University of Southern California Dissertations and Theses

Development of DNA methylation based biomarkers for the early detection of squamous cell lung cancer

(USC Thesis Other)

Development of DNA methylation based biomarkers for the early detection of squamous cell lung cancer

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DEVELOPMENT OF DNA METHYLATION BASED BIOMARKERS FOR THE
EARLY DETECTION OF SQUAMOUS CELL LUNG CANCER.

by

Paul P. Anglim

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOCHEMISTRY & MOLECULAR BIOLOGY)

December 2008

Copyright 2008 Paul Anglim
ii
Acknowledgments
So where does one start when writing an acknowledgment? I suppose at the beginning. I
want to say that I could write many more pages about all of the people mentioned here,
how much I love them and what they mean to me, but brevity is necessary.
First, my parents, for always trusting that my decisions were the right ones,
and supporting me in them, even if they didn’t always understand the rationale. Also, for
supporting me all the way through the process, for always giving me perspective, and
keeping me down to earth. You are both amazing role models who have always given
me the wings to fly, and yet the sense to stay grounded. Your support and inspiration has
helped me in not only all my choices but in my daily life and all the challenges I have
faced over the years. I fear the day when I cannot ask your counsel and appreciate every
day that I can. This simply could not have been accomplished without you.
Also the rest of my family, my brother Adrian, his fiancé Veronica, my
Uncle Joe and Aunt Breda and their daughters Tina and Michelle – despite my distance
for the past ten years (and I apologize) you have all loved, supported, and encouraged me
in my adventures, and more importantly, always given me a place to come home to. I
thank all of you, you have given me the strength to continue with it all.
When I decided to make the move, other than my family, I think it impacted
you, Monica, the most. I treasure you, you are my person, as we always said ‘where does
your mind end and mine begin’. Thank you for going through this with me, for sharing
both of our hard times, and for listening to mine. You have always listened, always been
there, and continue to do so. I intend that we continue to be there for each other for many
years to come. I love you, life without you seems improbable.
iii
To the Biotechs – my undergrad colleagues, especially Niamh, Ciara, Jules,
Audry, Geraldine and Caroline, thank you for being my best friends, and for braving this
ground before I did and advising me on how to do it. I am always grateful, and so lucky
to have you all as friends
Next the people who got me started when I moved to this place. The
O’Leary family, for giving me a place to stay and advice for many years, I will be ever
grateful. You all made the transition from Ireland to LA not only easy, but fun,
informative and amazing. I have made life long friends in you all, especially Kate, who
has guided, counseled, and made me laugh through so many of the experiences in the last
six years it would take another thesis to describe them all.
When I started at USC, my guidance very much came from Ludwig and
Meg. They are my great friends and were advisors not just for how to buy a bed at Ikea,
or find my way around, but for my whole life here. You both have helped me more than
you know and I am eternally grateful.
Then the move to Long Beach, the place I now call my second home. Jon,
I know you probably know this, but I will state it anyway. Your love, kindness, optimism,
and support got me through so many of the worst days in this degree. You not only
supported me, but educated, inspired and intellectually stimulated me. I often wonder if I
could have done it without you, and usually think not. I am sorry that you got to see the
worst of it, but perhaps we both got to see that for each other. Thank you for
understanding the process, for the counsel, advice and somedays just simple love and
support. I am glad to count you as one of my best friends, it is solace to know you are
always there. Also, thank you for the music, some of which I am listening to right now, it
iv
has always been an inspiration and a relief. Keep creating it, it is amazing and touches
many people.
And then there is the Long Beach crew, they are almost too numerous to
mention, but I shall try. I have had the privilege to be with such a phenomenal group of
people here: (in no particular order) RJ, Jilana, Brady, Jill, Sally, Suzi, Izak, Nicole,
Sarah – you all have been true friends. Thank you for the perspective, the advice and the
fun times. And for being the instigators! Thank you all for the love, support, and for
making me laugh after both the bad and the good times. One of the greatest privileges I
have had is to be able to spend time with all of you.
I must thank the families who have adopted me, The Talbergs, the Guess
family, the Miller family and last but not least Jean and Len – being part of your families
has been an honor, and thank you! You all make it so much easier to be so far from
home, and the time I have spent with you all means so much, I hope we can continue to
do so for many years.
At USC there are numerous people without whom this would not have been
possible. First of all, Ite, my advisor, thank you, for the opportunity, the support, the
kindness and the mentorship. To my colleagues, and also some of my best friends, again
in no particular order, (Shirley, Toshi, Sahar. Mel, Nicole, Kwang-Ho, Suhaida, Janice,
Jeff) Thank you all for sharing this experience with me, for the counsel, the support, the
technical advice, the personal interaction and just the general love. Without you going to
lab would not have been nearly as engaging, stimulating or fun. We have all shared this
experience, good and bad, and you all have always been there for me. I will love and
support you always, and consider myself lucky to have had the good fortune to work with
v
you all. I will always love you, and wish you all the best in your futures, and I look
forward to sharing your lives and careers with you.
I must also thank all of my collaborators, you have provided me with
excellent advise, and exposure to high level science. Thank you, those are lessons I will
take with me into my future.
For you all, I wish I could be more eloquent and expressive of my love and
gratitude in this writing. Thank you all, you mean so much to me, and each of you has
contributed so much to this degree, and my life in the last 6 years.

Peace and love,
Paul
vi

Table of Contents

Acknowledgments ii
List of Tables viii
List of Figures ix
Abstract x

Chapter 1: Introduction – DNA methylation-based biomarkers
for early detection of non-small cell lung cancer: an update 1
Abstract 1
Introduction 2
Early detection in lung cancer 3
DNA methylation 6
DNA methylation in non-small cell lung cancer (NSCLC) 7
DNA methylation in remote media 28
Selection of DNA methylation-based biomarkers for early detection of NSCLC 37
Conclusion 42
Author Contributions 44
Acknowledgements 45

Chapter 2. Identification of a panel of sensitive and specific DNA
methylation markers for squamous cell lung cancer 46
Abstract 46
Introduction 47
Materials and Methods 51
Tissue samples and DNA extraction 52
Methylation Analysis 52
Statistical Analysis 52
Results 54
Discussion 64
Conclusion 69
vii
Author Contributions 70
Acknowledgements 71

Chapter 3. Validation of a panel of sensitive and specific DNA
methylation markers for squamous cell lung cancer 72
Abstract 72
Introduction 73
Materials and Methods 76
Study Subjects 76
Tissue samples and DNA extraction 77
Methylation Analysis 78
Statistical Analysis 79
Results 80
Discussion 92
Conclusion 95
Acknowledgements 96

Chapter 4. A genome wide approach for identifying DNA
methylation based biomarkers of squamous cell lung cancer 97
Abstract 97
Introduction 97
Materials and Methods 101
Tissue samples and DNA extractions 101
Methylation analysis 102
Data Analysis 103
Results 105
Discussion 109
Conclusion 111
Acknowledgements 112

Chapter 5. Discussion 113
Bibliography 116
Appendix 139
Supplemental table 1 139
Supplemental table 2 142
viii
List of Tables
Table 1.1 Loci reported to be methylated in NSCLC in studies on single
loci/small panels of loci 9
Table 1.2 Alphebetical list of all loci focused on in studies using targeted
or genome wide approaches to examine DNA methylation at many
loci 20
Table 1.3 Alphabetical list of all genes whose DNA methylation status
has been examined in remote media 29
Table 2.1 Statistical analysis of differences in methylation levels between
tumor and adjacent non-tumor tissues 58
Table 2.2 AUC, Sensitivity & Specificity Analysis 64
Table 2.3 Putative biological role of the 22 statistically significantly
hypermethylated loci 66
Table 3.1 β-value of seven loci ranked as strongest potential biomarkers in
the GoldenGate assay 81
Table 3.2: Statistical analysis of differences in methylation levels
between tumor and adjacent non-tumor tissues 84
Table 3.3: Comparison of DNA methylation levels in Tumor Stage 85
Table 3.4: Importance Measure of biomarker panel 86

Table 3.5: Statistical analysis of differences in methylation levels between
tumor and adjacent non-tumor tissue 88
Table 3.6: Comparison of DNA methylation levels across ethnic groups 91
Table 3.7: Comparison of DNA methylation levels in Tumor Stage 92
Table 4.1: Common loci in the top 50 loci (by rank) in Frozen and FFPE fixed
Tissues 108

ix
List of Figures
Figure 2.1 DNA methylation analysis of 42 loci 45 squamous cell lung
cancer cases 55
Figure 2.2 Scatter plot analysis of the top 8 loci (as ranked by p-value) 61
Figure 2.3: Reciever Operator Characteristic curves for the top 8 loci
(as ranked by p-value) 63
Figure 3.1 DNA methylation for 4 loci in 42 tumor and adjacent
non-tumor lung cancer cases 83
Figure 3.2: Scatter plot representation of PMR values 84

Figure 3.3 DNA methylation analysis of 10 loci in 54 tumor and adjacent
non-tumor lung cancer cases 87

Figure 3.4: Scatter plot analysis of the 10-locus panel 89
Figure 3.5: Receiver Operating Characteristic (ROC) curves for the panel of
10 DNA methylation-based biomarkers 90

Figure 4.1: Correlation plots of the β-values between two independent
experiments 106
Figure 4.2: Venn Diagram of the number of loci that are ranked as potential
biomarkers from each data set 108

x
Abstract
Lung cancer is the number one cancer killer in the United States. This disease is divided
into two sub-types, small cell lung cancer, (10-15% of lung cancer cases), and non-small
cell lung cancer (NSCLC; 85-90% of cases). As NSCLC is the more common and less
aggressive sub-type, its early detection has very high potential for saving lives. No
routine screening method that facilitates early detection exists, and this is likely the
reason for the high mortality rate of this disease. Imaging and cytology based screening
strategies have been employed for early detection, and while some are sensitive, none has
been demonstrated to reduce lung cancer mortality. Developing specific molecular
markers that can compliment imaging techniques may facilitate a reduction in lung
cancer mortality. DNA methylation has emerged as a highly promising biomarker, and is
studied in multiple cancers.
NSCLC is further subdivided into four histological subtypes, each with its
own molecular profile. The goal of our work is to develop sensitive and specific DNA
methylation-based markers for early detection of squamous (SQ) lung cancer, which
accounts for 30% of all lung cancer cases. Using a combination of MethyLight and the
high-throughput DNA methylation analysis platform the GoldenGate assay, we identified
a 10-panel with high specificity (92.6%) for detection of SQ lung cancer. that used in
concert with sensitive imaging techniques may facilitate early detection of SQ lung
cancer. We further characterized the reproducibility of the GoldenGate assay, and
determined that we can obtain high quality data using DNA from FFPE tissue.
1
Chapter 1: Introduction - DNA methylation-based biomarkers
for early detection of non-small cell lung cancer: an update
This chapter was accepted in a peer-reviewed journal, Molecular Cancer

Paul P. Anglim
1
, Todd A. Alonzo
2
, and Ite A. Laird-Offringa
1 §

1
Departments of Surgery and of Biochemistry and Molecular Biology, Keck School of
Medicine, Norris Comprehensive Cancer Center, University of Southern California, Los
Angeles, CA 90089-9176, USA
2
Department of Preventive Medicine, University of Southern California, Arcadia, CA,
91006, USA

Abstract
Lung cancer is the number one cancer killer in the United States. This
disease is clinically divided into two sub-types, small cell lung cancer, (10-15% of lung
cancer cases), and non-small cell lung cancer (NSCLC; 85-90% of cases). Early
detection of NSCLC, which is the more common and less aggressive of the two sub-
types, has the highest potential for saving lives. As yet, no routine screening method that
enables early detection exists, and this is a key factor in the high mortality rate of this
disease. Imaging and cytology-based screening strategies have been employed for early
detection, and while some are sensitive, none have been demonstrated to reduce lung
cancer mortality. However, mortality might be reduced by developing specific molecular
markers that can complement imaging techniques. DNA methylation has emerged as a
2
highly promising biomarker and is being actively studied in multiple cancers. The
analysis of DNA methylation-based biomarkers is rapidly advancing, and a large number
of potential biomarkers have been identified. Here we present a detailed review of the
literature, focusing on DNA methylation-based markers developed using primary NSCLC
tissue. Viable markers for clinical diagnosis must be detectable in ‘remote media’ such
as blood, sputum, bronchoalveolar lavage, or even exhaled breath condensate. We discuss
progress on their detection in such media and the sensitivity and specificity of the
molecular marker panels identified to date. Lastly, we look to future advancements that
will be made possible with the interrogation of the epigenome.

Introduction
Worldwide lung cancer kills over one million people each year, and as the
leading cause of cancer death in men and second leading cause in women, it is a major
health problem [123]. This disease is largely smoking-associated. While in developed
countries smoking rates are decreasing, the use of tobacco products is increasing in
developing countries. In combination with a spike in the number of lung cancer cases in
never smokers, this ensures that lung cancer will remain a major health problem [123].
Clinically, lung cancer is divided into two subtypes, small cell lung cancer (SCLC) and
non-small cell lung cancer (NSCLC). SCLC is the more aggressive subtype, and accounts
for 10-15% of all cases. The remaining 85-90% of cases are classified as NSCLC, which
is further histologically subdivided into four categories; adenocarcinoma (AD), squamous
cell carcinoma (SQ), large cell carcinoma (LC) and ‘others’, for example cancers of
neuroendocrine origin.
3
In the United States lung cancer is the number one cancer killer in both men
and women, accounting for over 160,000 deaths each year [91]. Interestingly, it is not the
most commonly diagnosed cancer; breast and prostate cancer have a higher incidence. A
reason for this disparity is that early detection methods exist for breast and prostate
cancer, and these are widely used in the population. As a result, the five-year survival
rate is 89 and 99% (respectively) for these cancers, as opposed to a very low 15% for
lung cancer [91]. When early stage lung cancer is detected, the survival rate can increase
dramatically. For example, one report on detection of early stage cancers using low dose
spiral computed tomography (LDSCT) described a ten-year survival rate of 88% [83].
While there is concern that LDSCT leads to overdiagnosis (detection of indolent cancers
that would normally not lead to death), it is undisputed that effective early detection of
lesions that would otherwise progress to invasive cancer could reduce lung cancer
mortality. In an effort to achieve early detection many imaging and cytology-based
strategies have been employed, however none have yet been proven effective. Molecular
markers would provide an alternative approach and among them, DNA methylation
alterations show great promise. Here we present an update of the field of DNA
methylation markers for early lung cancer detection.

Early Detection of Lung Cancer
Original early detection methods for lung cancer were focused on screening
using chest X-ray and sputum cytology. Randomized controlled trials demonstrated no
reduction in mortality using these techniques [7, 62]. The question has been raised as to
whether these trials had enough statistical power to determine a mortality benefit [7, 41].
4
The Prostate, Lung, Colorectal and Ovarian cancer trial currently being conducted by the
National Cancer Institute is a larger trial and may conclusively reveal whether chest X-
ray screening can reduce mortality [7]. As discussed later, studies of molecular instead of
cytological changes in sputum samples appear promising [109].
Following the apparent failure of chest X-ray and sputum cytology as
effective screening techniques, attention was focused on a more sensitive imaging
method – Low Dose Spiral Computed Tomography (LDSCT). Several trials of LDSCT
as a screening tool in high-risk populations have been conducted [9, 27, 66, 67, 110, 152,
175]. It is clear that LDSCT is more sensitive than chest X-ray [66, 67], as it can detect
non-calcified nodules as small as 1 mm. Such high sensitivity comes with a price. The
number of non-calcified nodules detected is far greater than the number of actual cancers.
A Mayo Clinic study in 1999 reported that <2.0% of non-calcified nodules detected were
actually cancer [92]. This presents two potential problems for LDSCT as an early
detection method. Firstly, there is the potential for many false positive results, which
would result in low specificity if LDSCT were applied as a lung cancer screening tool.
The second problem is that in order to determine which nodules are actually cancer,
patients will require follow up procedures (further scans, possibly biopsies or resections).
These are costly, invasive, and can result in patient morbidity and mortality. Crestanello
et al. report that 9 out of 54 patients underwent surgery for benign nodules [35]. A review
of seven studies by Diederich and Wormanns reported that 4-55% of patients had
invasive procedures for benign lesions [41].
An increase in survival in LDSCT-screened lung cancer patients has been
reported; the IELCAP study reports an 88% 10-year survival [83]. Many argue that the
5
increased survival rate seen is due to an overdiagnosis bias. Using the Yankelevitz
criteria of overdiagnosis – a tumor volume doubling time (VDT) of > 400 days [193] –
27% of the detected cancers in a study by Lindell et al. would be considered
overdiagnosed [110]. In a review by Jett of a Japanese study, 33% of the cancers
detected have a VDT of >400 days [79], and hence would be considered overdiagnosed
[92]. Using a predictive model, Bach and colleagues recently examined the combined
results of LDSCT screening trials from three centers. They found an excess number of
cases diagnosed at each screening point compared to the predicted number, without a
decline in the number of advanced cancers being detected. This supports the notion of
overdiagnosis in LDSCT screening [6]. The true measure of efficacy of an early detection
method is a reduction in mortality. Whether LDSCT screening in high-risk populations
decreases lung cancer mortality remains unknown. The answer to this question will
hopefully be provided by of several ongoing randomized controlled trials (for example
the US-based National Lung Screening Trial, and in the Netherlands, the Dutch Lung
Cancer Screening Trial). The conclusions from such trials will determine the fate of
LDSCT as an early detection strategy.
Another imaging-based early detection approach is autofluoresence
bronchoscopy (AFB). This distinguishes between tumor and non-tumor tissue based on
the tumor-specific change in tissue autofluoresence. AFB has been shown to be effective
at detecting preneoplastic lesions and lung cancers [48]. The drawbacks of the method are
that it is invasive, it mainly detects centrally located cancers [120], and it is not highly
specific [81, 120].
6
Since imaging techniques have not yet proven effective as an early
detection method, a sensitive and specific screening strategy remains to be found. To fill
this void, research focus has shifted to molecular approaches. The goal is to identify
molecular markers (generally DNA, RNA or protein) that reflect characteristics of lethal
tumors, and that can be exploited for early detection of these lesions at the pre-invasive
stage. To function as molecular markers in a screening test, these molecules must be
detectible in remote media. If molecular markers that allow detection of cancer are
identified, they will require complementary highly sensitive imaging methods such as
LDSCT to locate the cancer. Identified molecular markers could be potentially targeted
by agents to help specifically enhance tumor imaging [186].

DNA Methylation
One highly promising molecular biomarker is DNA methylation. This
enzymatic addition of a methyl group at the 5-position of the cytosine in a CpG
(cytosine-guanine) dinucleotide is a normal process within cells. In cancer, despite a
global hypomethylation, one observes hypermethylation in regions of the genome
described as CpG islands [137, 160]. These islands are present in almost half of all genes
and are frequently promoter-associated [158]. The common occurrence of DNA
hypermethylation in all types of cancer makes it an ideal biomarker, one that has been
extensively investigated. An advantage of DNA methylation over protein-based markers
is that it is readily amplifiable and easily detectable using PCR-based approaches. In
addition, contrary to cancer-specific mutations, which could occur anywhere in a gene,
cancer-specific DNA hypermethylation occurs in defined regions, usually in or near the
7
promoter of genes. Thus, it is easy to devise targeted probes to measure this molecular
alteration. Conveniently, these probes can be readily combined into panels, which is
important because no single molecular alteration involved in cancer can be expected to be
present in every cancer case. Thus DNA methylation at a single gene would likely allow
detection of a subset of cancers. Assembly of a complementary panel of DNA
methylation probes would therefore increase sensitivity [5, 170]. Finally, it has been
demonstrated that methylated DNA can be isolated from ‘remote media’ making it well-
suited for non-invasive detection [13, 98].

Overview of DNA methylation analysis in NSCLC
In this review, we focus on DNA methylation-based biomarkers for early
detection of NSCLC. Because NSCLC is the less aggressive lung cancer subtype, and
accounts for 85-90% of all cases, its early detection holds the most promise for saving
lives. A plethora of studies describing DNA methylation in non-small-cell lung cancer
exist. These studies are summarized in three tables, which, due to their size, are attached
to this chapter as Tables 1-3. Each file lists the relevant loci in alphabetical order. Table
1 lists information from studies of less than 20 loci. Table 2 lists the results of DNA
methylation studies of 20 or more loci, or genome wide approaches. Lastly, Table 3
discusses loci studied in remote media from cancer patients. The contents of these tables
are discussed in more detail below.
Initial DNA methylation studies in NSCLC focused on single loci (or a small
number of well known loci) that were selected because of their potential functional role
in cancer. The goals of these studies were a) to see if methylation was involved in lung
8
cancer pathogenesis, or b) to determine if methylation of a given gene could be correlated
with clinical factors, and hence serve as a prognostic marker. This led to the
characterization of the DNA methylation status of many loci in NSCLC (listed in Table
1) [37, 119, 151, 166, 168, 176] [2, 15, 19, 21, 22, 30, 32, 42, 46, 53, 59, 73, 78, 84, 87,
88, 90, 99, 100, 106, 111-113, 115, 122, 125, 127, 128, 130, 141, 142, 147, 159, 161,
164, 165, 172-174, 183, 194-197] [20, 23, 28, 29, 31, 33, 55, 57, 60, 64, 74, 82, 93, 96,
103, 108, 149, 150, 156, 179, 180, 190, 191]. The information gathered in these studies
could be of clinical use for early detection, chemo prevention, diagnosis, treatment or
prognosis [51]. Further studies employed panels of 8-19 loci (including these previously
reported loci) for DNA methylation profiling [38, 97, 118, 139, 140, 148, 155, 163, 171,
182, 192, 199] (see Table file 1). This profiling was aimed at characterizing methylation
status of many loci in NSCLC, or in some cases, at identifying loci with the highest
methylation frequency in tumors versus non-tumor tissues, that could potentially be used
as DNA methylation-based biomarkers of the disease.
9

Table 1.1: Alphabetical list of all loci reported to be methylated in NSCLC in studies on
single loci/small panels of loci
HUG
O
a

Gene Name
b

Fraction
Methylated
c

Percent
Methylated
d

Material
e
Subtype
f
Method
g
Ref
h

ACAT
2
Acetyl-Coenzyme A
acetyltransferase 2
21/175 14 TU AD MSP 37
ADA
MTS8
ADAM metallopeptidase
with thrombospondin
type 1 motif, 8
29/50 58 TU MSP 75
16/24 67 TU AD
13/26 50 TU SQ
AGT Angiotensinogen 21/99 21 TU MSP 109
AKAP
12
A kinase (PRKA) anchor
protein (gravin) 12
67/175 39 TU AD MSP 37
APC
Adenomatosis polyposis
coli
8/25 32 TU MSP 105
7/7 100 TU AD ML 107
16/31 52 TU AD MSP 106
65/90 72 TU MSP 38
22/48 46 TU MSP 35
28/75 37 TU MSP 110
48/99 48 TU MSP 109
17/146 12 TU MSP 115
12/40 30 TU QMSP 113
17/31 55 TU QMSP 39
86/91 95 TU QMSP 40
24/105 25 TU MSP 114
95/99 96 TU QMSP 41
n/a n/a TU QAMA 111
19/28 68 TU
3D
Microarra
y
108
ATM
Ataxia telangiectasia
mutated
49/105 47 TU MSP 114
BCL2 B-cell CLL/lymphoma 2 28/120 23 TU MSP 42
19/48 40 TU AD
9/72 13 TU SQ
BMP3
Bone morphogenetic
protein 3B
45/91 46 TU MSP 31
BRCA
1
Breast cancer 1, early
onset
29/98 30 TU MSP 46
1/22 4 TU MSP 47

10
Table 1.1 continued
5/28 18 TU
3D
Microarr
ay
108
BRCA2 Breast cancer 2, early onset 41/98 42 TU MSP 46
CADM1 Cell adhesion molecule 1 11/14 78.5 TU BGS 94
21/48 44 TU BGS 95
45/103 44 TU BGS 96
29/68 43 AD
14/26 54 SQ
1/2 50 ADSQ
1/7 14 LCC
n/a n/a TU QAMA 111
CALCA
Calcitonin/calcitonin-related
polypeptide, alpha
6/7 86 TU AD ML 107
n/a n/a TU QAMA 111
18/28 64 TU
3D
Microarr
ay
108
CCND2 Cyclin D2 19/48 40 TU MSP 35
25/61 47 CL
CD9 CD9 molecule 3/19 13 TU MSP 48
CD44 CD44 molecule 5/28 18 TU
3D

Microarr
ay
108
CDH1 Cadherin-1 (E-cadherin) 130/224 58 TU MSP 49
2/7 29 TU AD ML 107
7/31 22 TU AD MSP 106
22/75 29 TU MSP 110
19/107 18 TU MSP 112
86/146 59 TU MSP 115
5/40 12 TU QMSP 113
27/31 87 TU QMSP 39
69/105 66 TU MSP 114
30/88 34.1 TU MSP 109
3/28 11 TU
3D
Microarr
ay
108
63/95 66 TU MSP 98
CDH13 Cadherin-13 (H-Cadherin) 130/305 43 TU MSP 50
18/42 43 TU MSP 33
15/30 50 CL MSP
9/20 45 TU MSP 51
5/7 71 CL MSP
40/150 27 TU MSP 105

11
Table 1.1 continued
70/146 48 TU MSP 115
11/40 28 TU QMSP 113
40/61 66 TU QMSP 52
21/63 34 TU MSP 45
26/88 29.5 TU MSP 109
15/28 54 TU
3D
Microarr
ay
108
CDKN2A
/p14
Cyclin-dependent kinase
inhibitor 2A
4/46 9 TU MSP 53
10/40 25 TU
Periphe
ral SQ
MSP
6/20 30 TU
Central
SQ
MSP 54
5/31 16 TU AD MSP 106
9/107 8 TU MSP 112
4/62 6 TU MSP 47
CDKN2A
/p16
Cyclin-dependent kinase
inhibitor 2A
22/54 41 TU MSP 53
64/122 52.5 TU MSP 56
11/18 61.1 TU SQ MSP 57
4/13 30.7 TU SQ MSP 58
8/20 40 TU
Central
SQ
MSP 54
15/40 48 TU
Periphe
ral SQ
MSP
49/224 21.9 TU MSP 49
44/150 29 TU MSP 105
6/7 86 TU AD ML 107
14/31 45 TU AD MSP 106
58/119 49 TU MSP 86
15/90 17 TU MSP 38
23/75 31 TU MSP 110
27/107 25 TU MSP 112
22/99 22 TU MSP 109
41/146 28 TU MSP 115
18/40 45 TU QMSP 113
12/29 41.4 TU MSP 59
7/31 23 TU QMSP 39
8/17 47 TU COBRA 116
5/9 56 CL
28/89 31 TU MSP 47
41/105 39 TU MSP 114
89/111 80.2 TU
Semi-
nested
MSP
55

12
Table 1.1 continued
48/61 79 TU QMSP 52
73/92 79.3 TU QMSP 60
33/63 53 TU MSP 45
n/a n/a TU QAMA 111
10/28 38 TU
3D
Microarr
ay
108
48/75 64 TU MSP 99
9/20 45 TU MSP 100
CDKN2B
/p15
Cyclin-dependent kinase
inhibitor 2B
4/20 20 TU
Central
SQ
MSP 54
4/40 10 TU
Periphe
ral SQ
MSP
2/28 7 TU
3D
Microarr
ay
108
CHFR
Checkpoint with forkhead and
ring finger domains
7/37 19 TU MSP 61
CHEK2 CHK2 checkpoint homolog 9/9 100 TU MSP 62
3/3 100 CL
CST6 Cystatin E/M 9/19 50 TU MSP 48
DAB2IP
DOC-2/DAB2 interactive
protein
63
m2a promoter region 19/47 40 CL MSP
26/70 37 TU MSP
m2b promoter region 16/47 34 CL MSP
25/70 36 TU MSP
DAPK1
Death associated protein kinase
1
40/122 32.8 TU MSP 64
2/31 6 TU AD MSP 106
15/90 17 TU MSP 38
10/23 43 CL BGS 34
12/32 37.5 TU
8/20 40 AD
4/12 33 SQ
21/75 28 TU MSP 110
20/107 19 TU MSP 112
37/146 25 TU MSP 115
17/40 43 TU QMSP 113
10/64 16 TU MSP 47
24/105 23 TU MSP 114
14/28 50 TU
3D
Microarr
ay
108
26/75 35 TU MSP 99
5/20 25 TU MSP 100
DBC1 Deleted in bladder cancer 1 n/a n/a TU QAMA 111

13
Table 1.1 continued
DKK3 Dickkopf homolog 3 32/238 13 TU MSP 105
DLC1 Deleted in liver cancer 1 11/18 61 TU COBRA 116
2/11 18 CL
DUOX1 Dual oxidase 1 11/39 28 TU AD MSP 65
DUOX2 Dual oxidase 2 15/39 38 TU AD MSP 65
EFEMP1
Epidermal growth factor-
containing fibulin like
extracellular matrix protein 1
12/32 37.5 TU MSP 66
EGFL7 EGF-like-domain, multiple 7 14/14 100 TU COBRA 116
5/11 56 CL
ENG Endoglin 11/16 69 TU COBRA 116
5/7 71 CL
Ep-CAM* Epithelial cell adhesion molecule 18/51 35 TU AD MSP 67
ESR1 Estrogen receptor 1 3/7 43 TU AD ML 107
n/a n/a TU QAMA 111
15/28 54 TU
3D
Microarr
ay
108
ESR2 Estrogen receptor 2 4/7 57 TU AD ML 107
FBN2 Fibrillin 2 14/16 88 CL MSP 68
62/126 49 TU MSP
FHIT Fragile Histidine Triad 34/99 34 TU MSP 70
28/91 31 TU MSP 69
68/254 27 TU MSP 70
117/224 52.2 TU MSP 49
40/107 37 TU MSP 112
24/63 39 TU MSP 45
GATA4 GATA binding protein 4 42/63 67 TU MSP 71
GATA5 GATA binding protein 5 26/63 41 TU MSP 71
GSTP1 Glutathione S-transferase pi 6/31 19 TU AD MSP 106
7/90 8 TU MSP 38
1/75 1 TU MSP 110
7/107 7 TU MSP 112
15/99 15 TU MSP 109
3/146 2 TU MSP 115
3/31 10 TU QMSP 39
2/21 9 TU MSP 47
2/7 29 TU AD ML 107
HOXA7 Homeobox A7 10/22 45 TU SQ COBRA 72
HOXA9 Homeobox A9 15/22 68 TU SQ COBRA 72
HRASLS HRAS-like suppressor 19/61 31.1 TU AD MSP 72
HS3ST2
Heparan sulfate D-glucosaminyl
3-O-sulfotransferase
28/40 70 TU QMSP 113

14
Table 1.1 continued
HT1RB
5-hydroxytryptamine receptor
1B
14/20 70 TU SQ
MS-
RDA
73
IL20RA Interleukin 20 receptor, alpha 45/175 26 TU AD MSP 37
LAMA3 Laminin, alpha 3 12/20 60 CL MSP 76
15/36 42 TU
11/19 58 AD
4/15 27 SQ
LAMB3 Laminin, beta 3 3/20 15 CL MSP 76
9/36 25 TU
6/19 32 AD
3/15 20 SQ
LAMC2 Laminin, gamma 2 5/20 25 CL MSP 76
8/36 22 TU
6/19 32 AD
2/15 13 SQ
54/146 37 TU MSP 115
MGMT
O
6
-methylguanine-DNA
methyltransferase
37/122 30.3 TU MSP 56
7/7 100 TU AD ML 107
13/31 42 TU AD MSP 106
22/53 42 TU AD MSP 103
25/70 36 SQ MSP
22/107 21 TU MSP 112
45/146 31 TU MSP 115
12/31 39 TU QMSP 39
18/83 21 TU MSP 47
15/105 10 TU MSP 114
34/90 38 TU OMSP 102
11/75 15 TU MSP 99
14/20 70 TU MSP 100
MINT1 Methylated in Tumor 1 33/146 23 TU MSP 115
MINT31* Methylated in Tumor 31 64/146 44 TU MSP 115
MINT32* Methylated in Tumor 32 33/146 23 TU MSP 115
MLH1
mutL homolog 1, colon cancer,
nonpolyposis type 2
5/75 7 TU MSP 110
18/99 18 TU MSP 109
2/146 1 TU MSP 115
62/105 59 TU MSP 114
43/77 55.8 TU MSP 77
8/28 29 TU
3D
Microarr
ay
108
MSH2
mutS homolog 2, colon cancer,
nonpolyposis type 1
18/99 18 TU MSP 109

15
Table 1.1 continued
43/77 55.8 TU MSP 77
MT3 Metallothionein 3 13/19 68 TU MSP 48
MTHFR
5,10-methylenetetrahydrofolate
reductase (NADPH)
7/7 100 TU AD ML 107
MYOD1 Myogenic differentiation 1 7/7 100 TU AD ML 107
14/90 16 TU MSP 38
n/a n/a TU QAMA 111
MYO18B Myosin XVIIIB 7/20 35 TU BGS 78
8/47 17 CL
NNAT Neuronatin 12/19 64 TU MSP 48
NRIP3
Nuclear receptor interacting
protein3
6/19 32 TU MSP 48
OLIG1
Oligodentrocyte transcription
factor 1
26/41 63 TU COBRA 79
OXTR Oxytocin receptor 1/19 6 TU MSP 48
PAX5α* Paired box 5 alpha 9/11 82 CL MSP 80
33/48 68.8 TU MSP
16/25 64 AD
17/23 74 SQ
PAX5β* Paired box 5 beta 9/11 82 CL MSP 80
27/48 56 TU MSP
13/25 52 AD
14/23 61 SQ
PER1 Period homolog 1 2/6 33 TU BGS 81
PGF Placenta growth factor 22/22 100 TU MSP 82
PGR* Progesterone receptor A 7/7 100 TU AD ML 107
PRKCDB
P
Protein kinase C delta binding
protein
11/14 79 TU BGS 93
44/107 41 TU MSP 112
PTGS2
Prostaglandin-endoperoxide
synthase 2
7/7 100 TU AD ML 107
11/20 55 TU MSP 100
PYCARD
PYD and CARD domain
containing
7/146 5 TU MSP 115
RAMP2
Receptor activity modifying
protein 2
14/32 43.7 TU MSP 66
RARB Retinoic acid receptor, beta 48/150 32 TU MSP 105
43/107 40 TU MSP 112
92/146 63 TU MSP 115
34/63 54 TU MSP 45
53/75 71 TU MSP 99
8/20 40 TU MSP 100
RARβ2* Retinoic acid receptor, beta 2 7/31 22 TU AD MSP 106
138/342 40 TU MSP 70
3/31 10 TU QMSP 39

16
Table 1.1 continued
19/29 65.5 TU MSP 59
n/a n/a TU QAMA 111
RASSF1
Ras association (RalGDS/AF-6)
domain family 1
7/7 100 TU AD ML 107
5/31 16 TU AD MSP 106
71/146 49 TU MSP 115
6/28 21 TU
3D
Microarr
ay
108
47/116 40.5 TU 84
17/35 48.6 AD
30/81 37 SQ
46/119 39 TU MSP 85
57/122 46.7 TU MSP 64
8/25 32 TU MSP 106
83/178 47 TU MSP 32
44/138 32 TU MSP 44
17/61 27.8 TU AD MSP 72
32/75 43 TU MSP 110
40/99 40 TU MSP 109
18/40 45 TU QMSP 113
14/31 45 TU QMSP 39
15/29 51.7 TU MSP 59
7/16 44 TU COBRA 116
10/14 71 CL
34/107 32 TU MSP 87
16/105 15 TU MSP 114
30/63 48 TU MSP 45
n/a n/a TU QAMA 111
RASSF2
Ras association (RalGDS/AF-6)
domain family 2
22/50 44 TU MSP 83
33/106 31 TU MSP 84
RASSF5
Ras association (RalGDS/AF-6)
domain family 5
17/61 27.8 TU AD MSP 72
RBP1
Retinol binding protein 1,
cellular
19/150 13 TU MSP 105
RECK
reversion-inducing-cysteine-rich
protein with Kazal motifs
35/55 53.6 TU MSP 88
RPRM
Reprimo, TP53 dependent G2
arrest mediator candidate
49/150 33 TU MSP 105
ROBO
Roundabout, axon guidance
receptor, homology 1
15/32 46.8 TU MSP 66
RUNX3
Runt-related transcription factor
3
6/25 24 TU MSP 89
3/11 27 AD
2/11 18 SQ
1/11 9 LC

17
Table 1.1 continued
15/75 20 TU MSP 110
SCGB3A
1
Secretoglobin, family 3A,
member 1
95/339 28 TU MSP 73
51/199 26 AD
42/132 32 SQ
2/8 25 LCC
SEMA3B
Sema domain, immunoglobulin
domain (Ig), short basic domain,
secreted, (semaphorin) 3B
65/138 47 TU MSP 44
SEMA3G
Sema domain, immunoglobulin
domain (Ig), short basic domain,
secreted, (semaphorin) 3G
13/14 93 TU COBRA 116
6/11 55 CL
SFRP1
Secreted Frizzled Related
Protein 1
20/31 64 TU AD MSP 106
15/29 52 CL MSP 92
44/80 55 TU MSP
111/146 76 TU MSP 115
81/238 34 TU MSP 91
SFRP2
Secreted Frizzled Related
Protein 2
123/146 84 TU MSP 115
123/238 52 TU MSP 91
SFRP4
Secreted Frizzled Related
Protein 4
43/146 29 TU MSP 115
SFRP5
Secreted Frizzled Related
Protein 5
100/146 69 TU MSP 115
78/238 33 TU MSP 91
SLIT2 Slit homolog 2 16/16 100 TU COBRA 116
3/3 100 CL
SLIT3 Slit homolog 3 0/17 0 TU COBRA 116
2/11 18 CL
SOCS1
Suppressor of cytokine signaling
1
13/40 33 TU QMSP 113
8/20 40 TU MSP 105
SOCS3
Suppressor of cytokine signaling
2
7/8 87.5 TU MSP 90
2/40 5 TU QMSP 113
SOX18
SRY (sex determining region
Y)-box 18
13/13 100 TU COBRA 116
8/11 73 CL
SPARC
Secreted protein, acidic,
cysteine-rich
81/150 54 TU MSP 106
SYNE1
Spectrin repeat containing,
nuclear envelope 1
88/175 50 TU AD MSP 37
TCF21 Transcription factor 21 19/22 86 TU RLGS 36
6/6 100 TU BGS
142/175 81 TU AD MSP 37
30/40 75 TU QMSP 113

18
Table 1.1 continued
TIMP3
TIMP metallopeptidase inhibitor
3
3/7 43 TU AD ML 107
28/107 26 TU MSP 112
2/15 13 TU COBRA 116
2/7 29 CL
4/21 19 TU MSP 47
12/28 43 TU
3D
Microarr
ay
108
TIMP4
TIMP metallopeptidase inhibitor
4
17/18 98 TU COBRA 116
9/14 63 CL
TMEFF2
Transmembrane protein with
EGF-like and two follistatin-like
domains 2
56/150 37 TU MSP 105
TNFRSF1
0C
Tumor necrosis factor receptor
superfamily, member 10c, decoy
without an intracellular domain
4/40 10 TU MSP 101
3/26 12 AD
1/14 7 SQ
9/40 23 TU QMSP 113
TNFRSF1
0D
Tumor necrosis factor receptor
superfamily, member 10d, decoy
with truncated death domain
11/40 28 TU QMSP 113
WIF1 WNT inhibitory factor 1 11/31 35 TU AD MSP 106
66/238 28 TU MSP 91
XRCC5
X-ray repair complementing
defective repair in Chinese
hamster cells 5
19/98 20 TU MSP 46
ZMYND1
0
Zinc finger MYND-type
containing 10
26/145 18.6 TU MSP 39
68/160 42 TU MSP 115
42/138 30 TU MSP 44
19/63 31 TU MSP 45

Table 1.1: Alphabetical list of all genes reported to be methylated in studies using less than 20 genes.
a
All
gene symbols are HUGO. In cases where the HUGO symbol has changed, the HUGO symbol is used and
the symbol at the time of publication is in parenthesis. * Denotes loci for which HUGO symbols cannot be
found
b
All gene names are from www.genecards.org.
c
Fraction methylated refers to the number of tumors
showing DNA methylation, and
d
percent methylated is derived from this.
e
Tissue type is either TU, tumor,
or CL, cell line.
f
Subtype refers to studies were a specific subtype of NSCLC was analyzed. AD is
adenocarcinoma, SQ is squamous cell carcinoma.
g
Method is the technique used to evaluate DNA
methylation. BGS -Bisulfite genomic sequencing, COBRA - Combined bisulfite restriction analysis, ML -
MethyLight, MSP - Methylation Sensitive PCR, MS-RDA - Methylation sensitive-representational
difference analysis, QAMA - Quantitative analysis of methylated alleles, QMSP - Quantitative MSP,
RLGS - Restriction landmark genome scanning.
h
Ref is the citation listing number in the bibliography and
equates to the citation number in the text.
19
Several loci identified in both types of studies (e.g. APC, CADM1, CDH1,
CDH13, CDKN2A/p14(ARF), CDKN2A/p16, DAPK, FHIT, GSTP1, MGMT, MLH1
and RASSF1A) are reported to be methylated multiple independent times in the literature
(reviewed in Table 1), and there is general consistency in the observed methylation
frequency for these loci. Any inconsistencies could have multiple explanations, for
example: the use of different techniques to study the methylation status, differences in the
population in each study, and a difference in the subtype composition of the NSCLC
collection studied.
To further characterize DNA methylation in NSCLC and facilitate the
discovery of new markers, more recent studies have employed approaches that analyze
large numbers of loci at one time. In these studies, the goal has been to identify DNA
methylation-based discriminators of tumor and normal tissues, and tumor subtypes. Some
of these approaches were targeted; the loci analyzed were selected based on their
relationship to cancer. Other approaches were not designed to interrogate DNA
methylation at specific loci, instead they examined the genome in greater depth and
identified potentially informative DNA methylation biomarkers based on comparative
profiling between tumor and non-tumor cells/tissues. The most promising loci to emerge
from these reports are reviewed in Table 2.
Table 1.2: Alphabetical list of all loci identified from studies using targeted or genome-
wide approaches to examine DNA methylation at many loci

HUGO
a
Gene Name
b
Method
c

Details
d
Fraction
Methylate
d
e

Ref
f

ADPRH ADP-ribosylarginine hydrolase
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
119
ALDH1A
3
Aldehyde dehydrogenase 1
family, member A3
MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
9/20 124
20
Table 1.2 continued
ASCL2
Achaete-scute complex homolog
2
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
BARHL2 BarH-like homeobox 2
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
17/20 127
BMP3 Bone morphogenetic protein 3B RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL T vs. AdjNTL
121
BNC1 Basonuclin 1 MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
18/20 124
BVES Blood vessel epicardial substance ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
CCNA1 Cyclin A1 MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
18/20 124
CD34 CD34 molecule RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
CDH1 Cadherin-1 (E-cadherin) ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
CDH13 Cadherin-13 (H-Cadherin)
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 adjNTL test set. 8 CpG sites
validated using BGS
122
ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
CD8B
CD8 antigen, beta polypeptide b1
chain
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
CDKN1C
Cyclin-dependent kinase inhibitor
1C
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
119
CDKN2A/
p16
Cyclin-dependent kinase inhibitor
2A
ML
Out of over 100 loci, 28 chosen
for evaluation, 7 show p<<0.0001
in 51 AD T vs. 38AdjNTL
27
CDX2 Caudal type homeobox 2 ML
Out of over 100 loci, 28 chosen
for evaluation, 7 show p<<0.0001
in 51 AD T vs. 38AdjNTL
27
CIDEB
Cell death-inducing DFFA-like
effector b
COBRA
8091 CpGs examined in lung
cancer cell lines, validation of
two genes in 8AD, 8SQ, 5 SCLC
123
CLEC3B
C-type lectin domain family 3,
member B
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
CTSZ Cathepsin Z MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
10/20 124
CYP1B1
Cytochrome P450, family 1,
subfamily B, polypeptide 1
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL T vs. AdjNTL
121
DAPK1 Death associated protein kinase ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
EVX2 Even-skipped homeobox 2
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
16/20 127
FABP3
Fatty acid binding protein 3,
intestinal
MSP
288 genes in lung cancer cell
lines using microarray. 5 genes in
22 T vs. AdjNTL
120
FGFR3
Fibroblast growth factor receptor
3
BGS
Studies a subset of 453
differentially expressed genes in
12 AD+3SQ vs. 5 normal adult
lung samples
125

21
Table 1.2 continued
GDNF Glial derived neurotrophic factor ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
GNAL
Guanine nucleotide binding
protein (G protein), alpha
activating activity polypeptide,
olfactory type
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
GPIBB
Glycoprotein Ib (platelet), beta
polypeptide
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
119
HOXA1 Homeobox A1 ML
Out of over 100 loci, 28 chosen
for evaluation, 7 show p<<0.0001
in 51 AD T vs. 38AdjNTL
27
HOXA5 Homeobox A5
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
HOXA7 Homeobox A7
MIRA/Mi
croarray
Cell line DNA studied using
MIRA and tiling arrays.
Confirmatory analysis in 22
T/AdjNTL cases from stage
1NSCLC
10/22 128
HOXA9 Homeobox A9
MIRA/Mi
croarray
Cell line DNA studied using
MIRA and tiling arrays.
Confirmatory analysis in 22
T/AdjNTL cases from stage
1NSCLC
15/22 128
HOXA11 Homeobox A11
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
IRF7 Interferon regulatory factor 7 MSP
288 genes in lung cancer cell
lines using microarray. 5 genes in
22 T vs. AdjNTL
120
IRX2 Iroquois homeobox 2
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
19/20 127
KCNH5
Potassium voltage-gated channel,
subfamily H (eag-related),
member 5
ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
LAPTM5
Lysosomal associated
multispanning membrane protein
5
BGS
Studies a subset of 453
differentially expressed genes in
12 AD+3SQ vs. 5 normal adult
lung samples
125
LHX2 LIM homeobox 2
MIRA/Mi
croarray/
MSP
Purification of methylated DNA
using methyl-binding domains
followed by microarray
hybridization. Confirmatory
analysis in primary NSCLC
tumors
7/12 126
LHX4 LIM homeobox 4
MIRA/Mi
croarray/
MSP
Purification of methylated DNA
using methyl-binding domains
followed by microarray
hybridization. Confirmatory
analysis in primary NSCLC
tumors
6/12 126

22
Table 1.2 continued
LOX Lysyl oxidase MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
19/20 124
MDK
Midkine (neurite growth-
promoting factor 2)
BGS
Studies a subset of 453
differentially expressed genes in
12 AD+3SQ vs. 5 normal adult
lung samples
125
MEIS1 Meis homeobox 1
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
17/20 127
MEOX2 Mesenchyme homeobox 2 BGS
Studies a subset of 453
differentially expressed genes in
12 AD+3SQ vs. 5 normal adult
lung samples
125
MGP matrix Gla protein
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
MGMT
O
6
-methylguanine-DNA
methyltransferase
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
119
ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
MLH3
mutL homolog 3, colon cancer,
nonpolyposis type 2
COBRA
8091 CpGs examined in lung
cancer cell lines, validation of
two genes in 8AD, 8SQ, 5 SCLC
123
MSX1 Msh homeobox 1 MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
11/20 124
MSX2 Msh homeobox 2
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
19/20 127
MTHFR
5,10-methylenetetrahydrofolate
reductase (NADPH)
ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
NPY Neuropeptide Y
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
NRCAM Neuronal cell adhesion molecule MSP
132 genes induced by 5-azadC,
31 methylated, top 8 analyzed in
20 T vs. 20 AdjNTL lung
18/20 124
NR2E1
Nuclear receptor subfamily 2,
group E, member 1
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
20/20 127
ONECUT
2
One cut homeobox 2
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
14/20 127
OPCML
Opioid binding protein/cell
adhesion molecule-like
ML
Out of over 100 loci, 28 chosen
for evaluation, 7 show p<<0.0001
in 51 AD T vs. 38AdjNTL
27
ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
OSR1 Odd-skipped related 1
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
20/20 127
OTX1 Orthodenticle homeobox 1
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
20/20 127
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121

23
Table 1.2 continued
PAX3 Paired box 3 MSP
288 genes in lung cancer cell
lines using microarray. 5 genes in
22 T vs. AdjNTL
120
PAX6 Paired box 6
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
17/20 127
PAX8 Paired box 8 ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
PDX1
Pancreatic and duodenal
homeobox 1
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
PTPRN2
Protein tyrosine phosphatase,
receptor type, N polypeptide 2
ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
PITX2 Paired-like homeodomain 2 ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
PYCARD
(ASC)
PY and CARD domain containing MSP
288 genes in lung cancer cell
lines using microarray. 5 genes in
22 T vs. AdjNTL
120
RARB Retinoic acid receptor, beta
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
119
ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
RASSF1
Ras association (RalGDS/AF-6)
domain family 1
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
ML
27 genes on 49 paired NSCLC T
and AdjNTL
117
RIPK3
Receptor-interacting serine-
threonine kinase 3
MSP
288 genes in lung cancer cell
lines using microarray. 5 genes in
22 T vs. AdjNTL
120
RUNX3 Runt-related transcription factor 3
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
SDK2 Sidekick homolog 2
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
SERPINB
5
Serpin peptidase inhibitor, clade
B (ovalbumin), member 5
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
SLC16A3
Solute carrier family 16, member
3
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
TAL1
T-cell acute lymphocytic
leukemia 1
RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
TBR1 T-box, brain, 1 RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
TCF21 Transcription factor 21 ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
TERT Telomerase reverse transcriptase
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122

24
Table 1.2 continued
TFAP2α Transcription factor AP-2 alpha
MIRA/Mi
croarray
4 SQ vs. AdjNTL on partial
genomic tiling arrays, detailed
analysis of gene subset on 20 SQ
T/AdjNTL
19/20 127
TLX1 T-cell leukemia homeobox 1 RGLS
1184 CpG islands in 16 NSCLC
T vs. AdjNTL
121
TMEFF2
Transmembrane protein with
EGF-like and two follistatin-like
domains 2
Microarra
y
245 CpG positions in 59
candidate genes in 26 SQ, 22 AD,
26 AdjNTL
59 119
TNFRSF2
5
Tumor necrosis factor receptor
superfamily, member 25
ML
Examined 42 candidate loci from
304 prescreened markers. 8 show
p<3E10-5 in SQ Tumor vs.
AdjNTL, 45 cases
28
TP73 Tumor protein p73
Illumina/
BGS
Examined 1536 CpG sites in 371
genes, identified 55-gene panel
that is 92% sensitive and 100%
specific on 12 adenocarcinoma &
12 AdjNTL test set. 8 CpG sites
validated using BGS
122
XAGE1A X antigen family, member 1
MALDI-
TOF
47 gene promoter regions in 96 T
with AdjNTL
118
ZNF577 Zinc finger protein 577
MIRA/Mi
croarray
AdjNTL on partial genomic tiling
arrays, detailed analysis of gene
subset on 20 SQ T/AdjNTL
18/20 127

Table 1.2: Alphabetical list of all genes focused on in studies using targeted or genome wide approaches to
examine methylation at many loci.
a
All gene symbols are HUGO. In cases where the HUGO symbol has
changed, the HUGO symbol is used and the symbol at the time of publication is in parenthesis.
b
All gene
names are from www.genecards.org.
c
Method is the technique used to evaluate DNA methylation. BGS -
Bisulfite genomic sequencing, COBRA - Combined bisulfite restriction analysis, MALDI-TOF - Matrix
assisted laser desorption ionization time of flight, MIRA - Methylated CpG-Island recovery assay, ML -
MethyLight, MSP - Methylation Sensitive PCR, MS-RDA - Methylation sensitive-representational
difference analysis, RLGS - Restriction landmark genome scanning.
d
Details describes the number of loci
examined, the source material used and numbers of tumors and the subtype of cancer. AD is
adenocarcinoma, SQ is squamous cell carcinoma. NSCLC is non-small cell lung cancer. T = tumor,
AdjNTL is adjacent non-tumor lung.
e
Fraction methylated refers to the number of tumors showing DNA
methylation.
f
Ref is the citation listing number in the bibliography and equates to the citation number in the
text.

One targeted approach is to use sodium bisulfite-treated DNA (in which
unmethylated Cs have been converted to Us) for semi-quantitative real time PCR
(MethyLight) to examine methylation levels of multiple loci. Three recent reports
described an examination of 27 loci in NSCLC [49], 28 loci in AD [170], and 42 loci in
SQ [5] using MethyLight. All three studies described a panel of loci with the ability to
sensitively and specifically detect cancer. Using a MALDI-TOF based approach 47 loci
were studied in tumor and non-tumor tissues from 96 patients. Six loci (CLEC3B
25
(previously TNA), MGP, RASSF1, SDK2, SERPINB5 and XAGE1A (previously
GAGED2)) with statistically significantly higher methylation in tumor samples compared
with non-tumor samples were identified [45]. A targeted microarray was used to study
the methylation status of 59 loci (245 CpGs) and a set of loci to discriminate SQ
(ADPRH (formerly ARH1), GP1BB, RARΒ and TMEFF2) and AD (CDKN1C, MGMT,
TMEFF2) from normal lung was identified [50]. In a similar system, the promoter
regions of 288 cancer-related genes were examined. Twenty-eight potential biomarker
loci were identified and 5 were further examined in lung cancer tissues, yielding two
(PAX3 and PYCARD/ASC) that showed frequent hypermethylation [56]. Restriction
landmark genomic scanning (RLGS) allows interrogation of up to 2000 promoter
sequences. In a study of 1184 CpG islands Dai et al. discovered 11 genes that are
differentially methylated in cancer, two of which are methylated in >50% of tumors
(GNAL and PDX1) [36]. A newer high throughput approach is the Illumina GoldenGate
platform, which examines 1505 CpG sites in 807 genes. Recently a panel of loci that
detects adenocarcinoma was discovered, of which 8 were further examined by bisulfite
genomic sequencing (ASCL2, CDH13, HOXA11, HOXA5, NPY, RUNX3, TERT and
TP73) [17]. A study using a large methylation microarray analyzed the promoter regions
of 8091 loci, identifying the frequently methylated CIDEB gene [80].
While these approaches can be used to determine the DNA methylation
status of large numbers of genes, a non locus-targeted approach that allows unbiased
interrogation of DNA methylation in the genome could examine far more loci. This could
yield additional biomarkers, as well as new information about general DNA methylation
patterns in lung cancer. Using an expression microarray one can identify genes induced in
26
cell lines treated with a DNA methylation inhibitor. Such genes are potential DNA
methylation targets. Using this approach, Shames et al. identified 132 tumor-specific
methylation candidates, 45 of which were further investigated, revealing seven potential
lung cancer markers (ALDH1A3, BNC1, CCNA1, CTSZ, LOX, MSX1 and NRCAM)
three of which showed frequent tumor-specific hypermethylation compared to non-
tumors [145]. Cortese et al. used a different approach, studying the DNA methylation of
genes that are differentially expressed in fetal vs. adult lung. Four loci (FGFR3,
LAPTM5, MDK, MEOX2) were identified as aberrantly methylated in lung cancer, one
with high frequency [34].
Using a methylated CpG island recovery assay coupled with microarray
analysis (MIRA-microarray), Rauch et al. enriched for CpG regions and then hybridized
this to a CpG microarray containing 12,192 CpG islands, >60% of which map to the 5’
end of known or putative genes. Multiple highly methylated loci were identified, of
which the top 50 were reported [134]. In follow-up studies they identified several loci as
markers for SQ lung cancer [136], including HOXA7 and HOXA9 [135]. It is of note that
while the non-targeted approaches have the potential to rapidly identify many more
biomarkers, the candidate biomarker loci must still be validated in primary tumors using
traditional approaches.
In general, there is not a large overlap between the top loci identified in the
targeted and non-targeted approaches. Several frequently methylated loci identified in
early studies, for example CDKN2A/p16, CDH13, MGMT and RASSF1 remain viable
markers when assessed in a larger context, underlining their importance in cancer
development/progression [5, 17, 45, 50, 170]. Methylation of genes that are occupied by
27
transcriptionally repressive polycomb group protein in embryonic stem cells, such as
members of the HOX and PAX families, was detected by targeted as well as genomic
approaches. This supports the notion that these genes may be prone to cancer-specific
methylation [187]. Further investigation of this group of genes is warranted.
Modest overlap between the top loci from the non-targeted studies is seen. This
might be expected as each of these approaches differ in their methods of experimentation,
data analysis and ranking of loci as biomarkers. It also indicates that further markers
remain to be identified and that development of the optimal panel will require additional
studies. Ongoing genome-wide analyses using a multitude of approaches will help solve
this issue, but it is important that these analyses be carried out on all histological subtypes
of lung cancer. As previously discussed, NSCLC is comprised of four histological sub-
groups. The two most common subtypes, adenocarcinoma and squamous cell lung
cancer, are quite distinct in both physical location and molecular profile [45, 50, 133,
162, 178, 198]. They show differential methylation profiles as reported by Field et al. and
Brena et al. [21], [50]. Indeed work in our lab supports the notion of different
methylation patterns in SQ and AD [5, 170]. The distinct nature of AD and SQ means an
optimal lung cancer methylation panel will probably require markers for both subtypes.
Markers for LC and other minor NSCLC groups, such as neuroendocrine cancers, remain
to be developed.

DNA Methylation in Remote Media
While using primary tissue to study methylation status is useful to discover
potential biomarkers, this material is not non-invasively accessible and is therefore not
28
useful for screening an at-risk population. The ideal system for early diagnosis is material
collected in a non-invasive/minimally invasive way that will contain methylated DNA.
For this, one looks to remote patient media – blood, naturally produced or induced
sputum, exhaled breath-condensate (EBC, non-invasive), and bronchoalveolar lavage
(BAL, semi-invasive). Multiple studies show that DNA methylation of certain loci can be
detected in blood, sputum and BAL (Table 3). A few show that genetic alterations can be
detected in EBC, as discussed below, although no published studies of DNA methylation
detection in this medium exist.

Table 1.3: Alphabetical list of all genes for which DNA methylation status has been
examined in remote media

HUGO
a
Gene Name
b

Fraction
Methylated
c

Samples
e

Material
f
Method
g
Ref
h

APC Adenomatosis polyposis coli 1/24 Cases BAL MSP 157
5/17 Cases BAL QMSP 39
3/10 Controls BAL QMSP
110/155 Cases BAL QMSP 156
28/67 Controls BAL QMSP
14/67 Cases BAL QMSP 160
1/102 Controls BAL QMSP
42/89 Cases
Plasma/
Serum
QMSP 41
0/50 Controls
Plasma/
Serum
QMSP
3/13 Cases Sputum QMSP 113
1/25 Controls

BHLHB
5
Basic helix-loop-helix domain
containing, class B, 5
12/98 Cases Sputum Nested MSP 147
11/92 Controls Nested MSP
CDH1 Cadherin-1 (E-cadherin) 13/27 Cases BAL QMSP 39
3/10 Controls BAL QMSP
CDH13 Cadherin-13 (H-Cadherin) 11/85 Cases BAL MSP 30
4/127 Controls BAL MSP
21/63 Cases Plasma MSP 45
6/36 Controls Plasma MSP
14/61 Cases Serum QMSP 52
3/53 Cases Serum MSP 143
29
Table 1.3 continued
27/98 Cases Sputum Nested MSP 147
23/92 Controls Nested MSP
19/72 Cases Sputum MSP 143
CDKN2
A/p16
Cyclin-dependent kinase
inhibitor 2A
12/19 Cases BAL MSP 155
4/24 Cases BAL MSP 157
1/7 Cases BAL QMSP 39
0/10 Controls BAL QMSP
14/85 Cases BAL MSP 30
8/127 Controls BAL MSP
17/50 Cases BAL QMSP 159
0/64 Controls BAL QMSP
4/20 Cases BAL MSP 100
14/68 Cases BAL MSP 99
9/67 Cases BAL QMSP 160
0/102 Controls BAL QMSP
103/136 Cases Plasma
Semi-nested
MSP
55
77/105 Cases Plasma Nested MSP 60
24/63 Cases Plasma MSP 45
3/36 Controls Plasma MSP
11/44
LC
survivor
s
Plasma Nested MSP 29
16/121 Smokers Nested MSP
7/74
Never
Smokers
Nested MSP
16/61 Cases Serum QMSP 52
14/100 Cases Serum MSP 143
3/9 Cases Serum MSP 142
12/35 Cases Serum QMSP 164
0/15 Controls Serum QMSP
15/72 Cases Serum MSP 140
19/53
LC
survivor
s
Sputum Nested MSP 29
30/118 Smokers Nested MSP
39/98 Cases Sputum Nested MSP 147
25/92 Controls Nested MSP
71/95 Cases Sputum
Semi-nested
MSP
55
3/13 Cases Sputum QMSP 113
2/25 Controls
1/29 Cases Sputum MSP 59
20/112 Controls
11/11 Cases Sputum Nested MSP 149
18/123 Controls Sputum
Nested
MSP

6/22 Cases Sputum MSP 148
29/72 Cases Sputum MSP 140
DAPK Death associated protein kinase

3/24

Cases

BAL

MSP

157
14/68 Cases BAL MSP 99
3/20 Cases BAL MSP 100
7/72 Cases Serum MSP 138
10/100 Cases Serum MSP 143

30
Table 1.3 continued
4/5 Cases Serum MSP 142
25/53
LC
Survivor
s
Sputum Nested MSP 29
21/118 Smokers Nested MSP
42/98 Cases Sputum Nested MSP 147
30/92 Controls Nested MSP
22/72 Cases Sputum MSP 140
FHIT Fragile Histidine Triad 7/24 Cases BAL MSP 157
19/85 Cases BAL MSP 30
36/127 Controls BAL MSP
20/63 Cases Plasma MSP 45
7/36 Controls Plasma MSP
GATA4 GATA binding protein 4 48/98 Cases Sputum Nested MSP 147
42/92 Controls Nested MSP
GATA5 GATA binding protein 5 10/45 Cases Serum MSP 140
34/98 Cases Sputum Nested MSP 147
26/92 Controls Nested MSP
31/72 Cases Sputum MSP 140
GSTP1 Glutathione S-transferase pi 1/3 Cases BAL QMSP 39
0/10 Controls BAL QMSP
1/2 Cases Serum MSP 142
HLHP* Unknown 42/98 Cases Sputum Nested MSP 147
36/92 Controls Nested MSP
HOXA9 Homeobox A9 14/22 Cases Sputum MSP 148
HS3ST2
Heparan sulfate D-glucosaminyl
3-O-sulfotransferase
5/13 Cases Sputum QMSP 113
3/25 Controls
IGFBP3
Insulin-like growth factor
binding protein 3
25/98 Cases Sputum Nested MSP 147
30/92 Controls Nested MSP
LAMC2 Laminin, gamma 2 72/98 Cases Sputum Nested MSP 147
70/92 Controls Nested MSP
MAGE
A1
Melanoma antigen family A, 1 11/22 Cases Sputum MSP 148
MAGE
B2
Melanoma antigen family B, 2 9/22 Cases Sputum MSP 148
MGMT
O
6
-methylguanine-DNA
methyltransferase

3/24

Cases

BAL

MSP

157
7/12 Cases BAL QMSP 39
0/10 Controls BAL QMSP
11/20 Cases BAL MSP 100
6/68 Cases BAL MSP 99
5/44
LC
survivor
s
Plasma Nested MSP 29
15/121 Smokers Nested MSP
2/74
Never
Smokers
Nested MSP
17/100 Cases Serum MSP 143
4/6 Cases Serum MSP 142
4/72 Cases Serum MSP 140
19/53
LC
survivor
s
Sputum Nested MSP 29

31

Table 1.3 continued
17/118 Smokers Nested MSP
23/72 Cases Sputum MSP 140
23/98 Cases Sputum Nested MSP 147
22/92 Controls Nested MSP
7/11 Cases Sputum Nested MSP 149
31/123 Controls Sputum
Nested
MSP

MLH1
mutL homolog 1, colon cancer,
nonpolyposis type 2
9/21 Cases Sputum MSP 77
PAX5α* Paired box 5 alpha 8/45 Cases Serum MSP 140
21/53
LC
Survivor
s
Sputum Nested MSP 29
14/118 Smokers Nested MSP
29/98 Cases Sputum Nested MSP 147
24/92 Controls Nested MSP
22/72 Cases Sputum MSP 140
PAX5β* Paired box 5 beta 3/53 Cases Serum MSP 140
13/53
LC
Survivor
s
Sputum Nested MSP 29
11/118 Smokers Nested MSP
41/98 Cases Sputum Nested MSP 147
32/92 Controls Nested MSP
22/72 Cases Sputum MSP 140
PTGS2
(COX2)
Prostaglandin-endoperoxide
synthase 2
5/20 Cases BAL MSP 100
RARB Retinoic acid receptor, beta 13/85 Cases BAL MSP 30
16/127 Controls BAL MSP
3/20 Cases BAL MSP 100
48/68 Cases BAL MSP 99
23/63 Cases Plasma MSP 45
6/36 Controls Plasma MSP
6/100 Cases Serum MSP 143
RARβ2* Retinoic acid receptor, beta 2 8/29 Cases Sputum MSP 59
58/112 Controls
0/3 Cases BAL QMSP 39
0/10 Controls BAL QMSP
27/50 Cases BAL QMSP 159
8/64 Controls BAL QMSP
27/67 Cases BAL QMSP 160
21/102 Controls BAL QMSP
RASSF1
(RASSF
1A)
Ras-association domain family
1A gene
31/111 Cases BAL QMSP 158
0/46 Controls BAL OMSP
4/14 Cases BAL QMSP 39
3/10 Controls BAL QMSP
15/85 Cases BAL MSP 30
5/127 Controls BAL MSP
5/20 Cases BAL MSP 100
20/67 Cases BAL QMSP 160
0/102 Controls BAL QMSP
16/44
LC
survivor
s
Plasma Nested MSP 29
32

Table 1.3 continued
2/121 Smokers Nested MSP
2/74
Never
Smokers
Nested MSP
24/63 Cases Plasma MSP 45
4/36 Controls Plasma MSP
10/12 Cases Serum Meth-DOP-PCR 141
11/100 Cases Serum MSP 143
7/72 Cases Serum MSP 140
27/80 Cases Serum MSP 144
0/50 Controls Serum MSP
13/53
LC
survivor
s
Sputum Nested MSP 29
8/118 Smokers Nested MSP
12/98 Cases Sputum Nested MSP 147
6/92 Controls Nested MSP
5/13 Cases Sputum QMSP 111
2/25 Controls
1/29 Cases Sputum MSP 59
1/112 Controls
19/72 Cases Sputum MSP 140
SEMA3
B
Sema domain, immunoglobulin
domain (Ig), short basic domain,
secreted, (semaphorin) 3B
45/50 Cases BAL QMSP 159
23/25 Controls BAL QMSP
SFRP1
Secreted Frizzled Related
Protein 1
68/98 Cases Sputum Nested MSP 147
71/92 Controls Nested MSP
SOCS1
Suppressor of cytokine signaling
1
6/20 Cases BAL MSP 100
TCF21 7/13 Cases Sputum QMSP 97
0/25 Controls Sputum QMSP
ZMYND
10
Zinc finger MYND-type
containing 10
19/63 Cases Plasma MSP 45
5/36 Controls Plasma MSP

Table 1.3: Alphabetical list of all genes whose DNA methylation status has been examined in remote
media.
a
All gene symbols are HUGO. In cases where the HUGO symbol has changed, the HUGO symbol is
used and the symbol at the time of publication is in parenthesis. * Denotes loci for which HUGO symbols
cannot be found
b
All gene names are from www.genecards.org.
c
Fraction methylated refers to the number
of tumors showing DNA methylation.
d
Samples is either Cases i.e. have tumor, or Controls, i.e. no tumor.
e
Media refers to which remote media was used in this study, BAL is bronchoalveolar lavage.
f
Method is the
technique used to evaluate DNA methylation. MSP - Methylation Sensitive PCR, QMSP - Quantitative
MSP.
g
Ref is the citation listing number in the bibliography and equates to the citation number in the text.

The ideal remote medium is blood – it can be applied to all patients, both
those at minimal and high risk, and is minimally invasive to obtain. It is reported that
cancer patients have a higher level of circulating DNA than non-cancer cases [146], and
33
that genetic [75, 95, 153], and epigenetic [104] alterations can be detected in said DNA.
It is postulated that this DNA is released due to necrotic cell death [188]. Over 25 loci
have been reported to be methylated in plasma/serum of NSCLC patients [12, 13, 40, 47,
53, 54, 84, 111, 173, 174, 181] (reviewed in Table 3). Several studies examined
methylation in primary tumor material and corresponding plasma/serum, and in these
cases methylation in blood was only seen in patients in which the primary tumor also
exhibited methylation [47, 53, 173]. Many of the most promising markers from Table 1
and 2 have not yet been investigated in blood.
There are, however, caveats to detection of DNA methylation in blood. It is
questioned as to whether there is enough methylated DNA in the blood to efficiently
detect tumors at an early enough stage for curative resection. While DNA quantity may
be low, ongoing research on more sensitive detection methods may overcome this issue.
Another potential problem is that blood as a remote medium is not organ-specific; loci
that are methylated in lung cancer may be methylated as well in other cancers, for
example TNFRSF10C and D [148] TCF21 [151], RUNX3 [108], APC [177], FBN2 [30].
Thus, methylation of these loci in blood could point to cancer in any one of several
organs. The best markers for lung cancer would therefore be ones that show methylation
only in lung cancer. Given the recent focus on more genome wide approaches to study
methylation in many cancer types, a comparison of DNA methylation profiles across
cancer sites should soon be possible. An alternative to this is to complement DNA
methylation marker screening with sensitive imaging techniques to identify the cancer
site. Another option is to examine remote media that are more lung-specific.
34
Sputum is produced by increased bronchial secretions, and is commonly
found in smokers, hence it can be used to screen high-risk populations. (In former or non-
smokers, it is much more difficult to obtain, though it can be induced). The advantages of
sputum as a screening tool include its non-invasive procurement, and the fact that it
contains cells from the lungs and lower respiratory tract. However, the material in sputum
is from the center of the lungs, and it may not be as useful for the detection of
adenocarcinoma, which generally occurs at the periphery. DNA methylation, mutations,
and microsatellite alterations have been detected in sputum, indicating it is a useful
source of tumor material [13, 109, 121]. Reports of DNA methylation in sputum are
summarized in Table 3 [12-15, 32, 129, 130, 148, 149, 183]. It has been demonstrated
that promoter methylation in sputum increases with cancer risk [13], increases as the time
to lung cancer decreases [14], and in the case of CDKN2A/p16 and/or MGMT, can be
found in sputum up to 3 years before diagnosis of squamous cell lung cancer [131]. A
study by Liu et al. using 50 matched tumor, plasma and sputum samples showed that
CDKN2A/p16 hypermethylation is detected in 84% of tumors, and 76% of sputum
samples from the same patients, demonstrating that this remote medium is potentially
effective in detecting lung cancer [111]. However, whether this detection is applicable to
all NSCLC subtypes remains to be determined.
Exhaled breath provides a source of materials that can reflect the disease state of
the lungs. Breath condensate, comprised mostly of water vapors, also contains lipids,
proteins, DNA and oxidation products – the levels of which may differ between healthy
and diseased subjects [85]. Several studies report the utility of EBC in detection of
asthma, chronic obstructive pulmonary disease (COPD) and cystic fibrosis [85]. EBC has
35
also been used for NSCLC detection. Carpagnano et al. reported detection of the
mitogenic factor endothelin-1 (ET1-1) in EBC of lung cancer patients. In a small study
they showed a statistically significant difference in ET-1 levels between healthy controls
and NSCLC patients, and between stage I-III and stage IV patients [25]. They have
shown similar results when looking at interleukin-6 [26]. While these studies are protein-
based, they do demonstrate the promise of EBC for early detection of lung cancer. Thus
far, there are no published reports of DNA methylation detection in EBC, although two
studies reported collecting sufficient DNA quantities to perform PCR-based assays for
microsatellite alterations and p53 mutations [24, 65]. Of concern is the fact that the p53
mutations detected in EBC differ from those found in the primary tumor from the same
patient [24, 65]. This raises concern regarding the origin of DNA obtained from EBC (it
may also come from cells in the esophagus, throat or mouth) and its utility as a remote
medium.
Bronchoalveolar lavage (BAL) is another potential screening material for
early detection of lung cancer. While obtaining lavage fluid is not as invasive as a biopsy,
it requires bronchoscopy. However, bronchoscopy is routinely performed in suspected
lung cancer cases and lavage fluid can be easily obtained during this procedure. An
advantage of BAL is that it allows localized harvesting of lung-specific material, so that
the fluid can be expected to contain lung cancer cells and/or DNA. Several investigations
of DNA methylation in BAL have been conducted [3, 28, 39, 69-71, 74, 98, 143, 165]
(Table 3). Results vary between studies. De Fraipont showed low levels of DNA
methylation in BAL from tumor-bearing patients, indicating that this would not be a good
medium for marker detection [39]. In contrast, Topalogu used a panel of loci and
36
detected 68% of their tumor cases by examining DNA methylation in the corresponding
BAL from the same patients [165]. Kim et al. also reported a good correlation between
methylation in tumors and BAL, ranging from 39-61% for the five loci they analyzed
[98]. DNA methylation has also been detected in control BAL from non-neoplastic
patients [70, 98, 143]. The detection of DNA methylation in cancer-free patients is cause
for concern if presence/absence of DNA methylation is being used as a diagnostic
measure of cancer. However, if a quantitative assay to determine DNA methylation levels
is applied, then one can determine a cut-off value, above which a sample would be
considered positive, as was done by Grote et al. [70] and Schmiemann et al. [143].
The analyses of DNA methylation markers in remote media are still in their early
stages, and although many show low sensitivity, the inclusion of more of the recently
identified promising markers (Table 2) in future studies would likely boost detection of
cancer cases. Published data so far supports the continued analysis of these fluids in
search of an early detection method that can, at the very least, complement imaging-
based screening of at risk subjects.

Selection of DNA methylation-based biomarkers for early detection of NSCLC
While a plethora of loci are reported to serve as potential DNA methylation-
based biomarkers for NSCLC, the important question is: Which should be chosen for
further evaluation, and eventually for screening of subjects? When performing a
screening test there are four potential outcomes. The first two of these, true-positive
results (TP, those who test positive and actually have cancer), and true-negative results
(TN, those who test negative and do not have cancer), are the desired outcome of a
37
screening test. However, false-negative results (FN, those who have cancer but do not
test positive), and false-positive results (FP, those who do not have cancer but test
positive), could do serious harm to the screening populations. False negative results have
the ramification of delaying diagnosis of the disease, hence endangering patients’ lives,
while false positive results significantly affect patient quality of life [102]. Sensitivity,
defined as TP/(TP+FN), and specificity, defined as TN/(TN+FP), measure the balance of
these results in the population. These measures can serve as the selection criteria to
determine which potential biomarkers are pursued further. An ideal DNA methylation-
based biomarker would be highly sensitive and specific in all populations studied,
regardless of age, gender, ethnicity, risk factors and tumor stage. However, given the
differences between NSCLC subtypes and smoking and non-smoking associated NSCLC,
markers that function accurately in a subset of the population could also be of use. The
likelihood of identifying a single marker with 100% sensitivity and specificity is
negligible.
The methylation frequency for many loci examined in early studies is quite
low in primary tumors (Table 1, for example, DAPK 16-47%, p16 23-81%, CDH13 28-
48%, and RASSF1A 15-54%). If the methylation frequency is low, sensitivity will suffer
as the locus yields too few cases. Even for the more frequently methylated loci listed in
Table 2, one DNA methylation marker cannot be expected to detect all cases of a
particular cancer. The way to address this problem is to study the DNA methylation
status of multiple loci (a panel) in a sample population. To ensure high sensitivity
individual loci in the panel should be highly penetrant, i.e. have a high frequency in the
population, and be complementary, i.e. detect different tumor cases.
38
While ensuring high sensitivity is important, given very sensitive imaging
approaches like LDSCT, the more critical issue in lung cancer screening is high
specificity. False-positive results precipitate not only patient anxiety, but also follow up
procedures that are invasive, costly, and have associated morbidity and mortality. The
incidence for lung cancer in the United States is 79.4/100,000 in men and 52.6/100,000 in
women [144]. This shows that less than 0.1% of the population will get lung cancer.
Hence, a population-based screening using any marker with a specificity of less that
99.9% will detect more false positive cases than true positive ones. Such a marker
therefore cannot function as a screening marker in the population at large. However, in
current smokers the risk of lung cancer is greatly increased (incidence of over 230 per
100,000 for both men and women [8]), and the specificity of a marker can be slightly
lower when screening is targeted to this high-risk group.
Sensitivity and specificity have been reported for several locus panels when
examining methylation in DNA isolated from primary tissue. The area under the curve
(AUC) of a receiver operating characteristic (ROC) curve is a measure of the ability of a
continuous marker to accurately classify tumor and non-tumor tissue. Such a curve is a
plot of sensitivity vs 1 minus specificity values associated with all dichotomous markers
that can be formed by varying the value threshold used to designate a marker “positive”.
An AUC of 1 corresponds to a marker with perfect accuracy, while an AUC of 0.5
corresponds to an uninformative marker. Shivapurkar et al. studied the DNA methylation
of 11 loci to distinguish between NSCLC and adjacent non-tumor lung tissue. Using a
logistic regression with a binary outcome indicator of tumor and non-tumor lung tissue,
and a marker panel as covariates, they demonstrated that a combination of HS3ST2
39
(3OST2), DAPK and TNFRSF10C (DcR1) gave an ROC curve with an AUC of 0.959
when comparing tumor and adjacent non-tumor lung tissue. This implies that this
combination of markers could sensitively and specifically detect lung cancer [148].
Ehrich et al. studied the methylation of 47 loci and developed a panel of 6 that could
distinguish cancer from adjacent normal tissue with >95% sensitivity and specificity [45].
Feng et al. developed a panel of 8 loci, of which the presence of methylation of one gene
was found in 80% of NSCLC tissues [49]. In an effort to develop markers for specific
NSCLC subtypes, Tsou et al. reported a panel of 4 loci with 94% sensitivity and 90%
specificity for AD [170], while Anglim et al. reported a panel of 4 loci that with 96.5%
sensitivity and 93.3% specificity for SQ lung cancer [5]. Both reports compare DNA
methylation in tumor and adjacent non-tumor tissue from the same patients. On a larger
scale, Bibikova et al. identified 55 loci that distinguished AD from adjacent non-tumor
lung with 100% sensitivity and 92% specificity [17]. These are all encouraging results,
implying that DNA methylation detection could serve as a viable early detection
biomarker, but these loci must be further validated in larger, racially/ethnicially and
gender balanced independent populations in order to ensure equal functionality for all
patients. Also, primary tissue would not be the source material tested in screening for
early detection, hence, promising loci must be interrogated for their potential to
sensitively and specifically detect cancer in remote media.
There are multiple reports of DNA methylation in blood, but not all assess
the sensitivity and specificity of the loci. In those that do, it appears that detection in
blood is commonly not sensitive [12, 54]. For example, sensitivity ranged from 7-27%
for CDH13, CDKN2A/p16, DAPK, GATA5, MGMT, PAX5α, PAX5β and RASSF1A in
40
serum, but is much higher in sputum for the same samples [12]. One way in which
investigators have tried to increase sensitivity is by defining a patient positive if a
minimum number of loci are methylated. For example, Fujiwara et al. also described a
low sensitivity of 49.5% when looking at methylation of at least one of 5 loci in serum
(CDKN2A/p16, DAPK, MGMT, RARΒ and RASSF1A) but specificity was 85% [54].
Recently a report examining the methylation of CDH13, CDKN2A/p16, FHIT, RARB,
RASSF1A and ZMYND10 (BLU) in which methylation of any 2 loci in plasma was
considered cancer positive showed 73% sensitivity and 82% specificity [84]. This
reinforces the notion that a panel of complementary loci is necessary. In an interesting
report, Bearzatto et al. showed that combining CDKN2A methylation with microsatellite
alterations in plasma increased sensitivity to 62%, and using CDKN2A methylation
combined with circulating DNA levels increased specificity to 80%, as opposed to
examining CDKN2A methylation alone [10]. While neither of these is ideal as a clinical
test, it is of note that the marker panels need not consist solely of DNA methylation-based
markers.
Many studies indicate that sputum could be a promising remote medium for
early detection. Shivapurkar et al. described a combination of 4 loci, APC,
CDKN2A/p16, HS3ST2 (3OST2), and RASSF1A that serve as a good panel for early
detection of NSCLC in sputum, with an AUC of 0.8 [148]. Similarly, Li et al. reported a
combination of FHIT and HYAL2 with 76% sensitivity and 85% specificity [109]. Wang
et al. described MLH1 methylation in sputum to have 60% sensitivity and 86%
specificity [183], and Belinsky showed that concomitant methylation of three or more of
a panel of 6 loci resulted in 64% sensitivity and specificity [14]. In contrast, Cirincione et
41
al. reported that 3 loci, CDKN2A/p16, RARβ2 and RASSF1A are of limited use in early
detection of lung cancer using sputum as a remote medium [32].
Detection of DNA methylation in bronchoalveolar lavage is also
documented. Grote et al. published two reports, using either APC or RASSF1A alone for
NSCLC detection. In both cases there is low sensitivity (30 and 34 % respectively) but
high specificity (98.5 and 100% respectively) [69, 71]. Using just CDKN2A/p16, Xie et
al. describe a higher sensitivity (64%) than any other reports on DNA methylation in
BAL when examining a single locus and a modest specificity (75%) [189]. Grote et al.
explored the use of marker combinations in two studies. In the first they used
CDKN2A/p16 and RARB2 in combination and showed 69% sensitivity and 87%
specificity in their population [70]. In another study they applied a marker panel (APC,
CDKN2A/p16, and RASSF1) to detect cancer in 247 patients, and reported 53%
sensitivity and, in cases without a previous history of cancer, >99% specificity [143]. It is
probable that the inclusion of more highly penetrant markers in such panels would
increase sensitivity. This again highlights the need for a panel of markers, and underlines
the need to combine molecular markers with imaging techniques.

Conclusion
Lung cancer is responsible for a million cancer deaths per year worldwide,
and its detrimental effects will continue to increase. Research focused on biomarker-
based early detection has the potential to reduce mortality rates. What will it take to
obtain functional DNA methylation markers for early lung cancer detection?
42
Sullivan-Pepe outlined the five phases of biomarker discovery[132]. The
first phase, clinical exploratory, consists of identification of promising markers. Much
work on identification of DNA-methylation based markers has already been done, as
described Tables 1 and 2, and a number of markers has been carried forward to phase
two, the clinical detection of established disease (Table 3). However, with the advent of
new techniques, a thorough evaluation of the epigenome of all types of cancer will soon
be possible. The pool of potential DNA methylation markers for lung cancer has by no
means been exhausted, and it is expected that additional high penetrance markers will be
identified. It will be important to examine DNA methylation in each of the major lung
cancer histological subtypes and ensure the functionality of identified markers in lung
cancers from both genders and all races/ethnic groups. In addition, given the fact that half
of all new lung cancer cases arise in ex-smokers or never smokers [1], and the observed
molecular differences between lung cancer from smokers and non-smokers [154], it
would be important to ensure representation of lung cancer from never smokers in these
marker screens. Standardization of epigenomic assay techniques and data analysis would
facilitate comparisons of DNA methylation profiles between cancer types, which may
allow the identification of true lung-cancer specific hypermethylation. Ideally, only
reproducibly hypermethylated high penetrance DNA methylation markers should be
carried forward to the analysis of systematically collected remote media (because remote
media are such a valuable resource). The most promising markers can then be tested in
retrospective longitudinal studies (phase three), in which materials collected prior to
disease onset are examined. Studies of DNA methylation in sputum and BAL collected
prior to diagnosis already look promising (e.g. [131] [143]), and results can improve
43
further with the inclusion of new high sensitivity/specificity marker panels. If results are
promising, prospective screening studies (phase four) should follow to determine the
extent and properties of detected disease and measure the false referral rate. Lastly, case
control studies should be done to measure any effect on lung cancer mortality.
If a strong DNA methylation marker panel were developed, the manner in which
it would be applied would depend on its sensitivity and specificity. It is unlikely that
DNA methylation markers, or any molecular markers for that matter, would be used on
their own. Instead, we envision that they will be applied in concert with high-resolution
imaging. In the near future, the prospect of genome-wide interrogation of DNA
methylation in lung cancer is extremely exciting. The resulting information may provide
not only new candidate markers for early detection, but also for monitoring response to
therapy and recurrence. In addition, methylation information could be linked to
pathobiology and clinical characteristics, potentially providing indicators for treatment
and prognosis. Much work remains to be done, but using epigenomics while building on
the experience and materials obtained from prior studies, we are well armed to make non-
invasive testing for early lung cancer detection a reality.

Authors’ contributions
PPA was involved in drafting the manuscript and generation of tables. TAA was involved
in reviewing and editing the manuscript. IALO mentored PPA, and revised manuscript
drafts. All authors reviewed and commented on the manuscript during its drafting and
approved the final version.
44
Acknowledgements
The authors thank members Laird-Offringa lab members for critical comments on the
manuscript. Grant support for IALO includes: National Institutes of Health/National
Cancer Institute R21 CA102247, R01 CA119029 and R01 CA120869, Whittier
Foundation Translational Research Grant, a STOP Cancer award, a Joan’s Legacy grant,
a Thomas Labrecque Foundation grant, and generous gifts from the Kazan, McClain,
Abrams, Fernandez, Lyons & Farrise Foundation, the Canary Foundation, Paul and
Michelle Zygielbaum, and Conya and Wallace Pembroke. None of the funding agencies
played any role in the collection, analysis, interpretation of the data, writing of the
manuscript, nor the decision to publish. The content is solely the responsibility of the
authors and does not represent the official views of the funding agencies.

45

Chapter 2. Identification of a panel of sensitive and specific
DNA methylation markers for squamous cell lung cancer

This chapter was accepted in a peer-reviewed journal, Molecular Cancer

Paul P Anglim
1*
, Janice S Galler
1*
, Michael N Koss
2
, Jeffrey A Hagen
3
, Sally Turla
2
,
Mihaela Campan
1
, Daniel J Weisenberger
1
, Peter W Laird
1,3
, Kimberly D Siegmund
4
and
Ite A Laird-Offringa
1,3 §

1
Departments of Surgery and of Biochemistry and Molecular Biology, Norris Cancer
Center, Keck School of Medicine, University of Southern California, Los Angeles, CA
90089-9176, USA;
2
Department of Pathology, Keck School of Medicine, University of
Southern California, Los Angeles, CA 90089-9092, USA;
3
Department of Surgery, Keck
School of Medicine, University of Southern California, Los Angeles, CA 90089-9202,
USA;
4
Department of Preventive Medicine, Keck School of Medicine, University of
Southern California, Los Angeles, CA 90089-9175, USA

*

These authors contributed equally to this work

Abstract
Background: Lung cancer is the leading cause of cancer death in men and women in the
United States and Western Europe. Over 160,000 Americans die of this disease every
46
year. The five-year survival rate is 15% - significantly lower than that of other major
cancers. Early detection is a key factor in increasing lung cancer patient survival. DNA
hypermethylation is recognized as an important mechanism for tumor suppressor gene
inactivation in cancer and could yield powerful biomarkers for early detection of lung
cancer. Here we focused on developing DNA methylation markers for squamous cell
carcinoma of the lung. Using the sensitive, high-throughput DNA methylation analysis
technique MethyLight, we examined the methylation profile of 42 loci in a collection of
45 squamous cell lung cancer samples and adjacent non-tumor lung tissues from the same
patients.
Results: We identified 22 loci showing significantly higher DNA methylation levels in
tumor tissue than adjacent non-tumor lung. Of these, eight showed highly significant
hypermethylation in tumor tissue (p < 0.0001): GDNF, MTHFR, OPCML, TNFRSF25,
TCF21, PAX8, PTPRN2 and PITX2. Used in combination on our specimen collection,
this eight-locus panel showed 95.6% sensitivity and 95.6% specificity.
Conclusion: We have identified 22 DNA methylation markers for squamous cell lung
cancer, several of which have not previously been reported to be methylated in any type
of human cancer. The top eight markers show great promise as a sensitive and specific
DNA methylation marker panel for squamous cell lung cancer.

Introduction
Cancer is responsible for one in four deaths in the US, making it the second most
common cause of death [91]. Lung cancer is the leading cancer killer in men and
women.
47
Over 160,000 Americans will die of this disease in 2007. In men, lung cancer accounts
for 31% of cancer deaths, killing more men than leukemia and prostate, colorectal, and
pancreatic cancer combined. In women, lung cancer accounts for 27% of all cancer
deaths, taking as many lives as breast and colorectal cancer combined [91]. The overall
five-year survival rate of lung cancer patients is 15%, significantly lower than that of
patients with prostate cancer (99.9%), breast cancer (88.5%) or colon cancer (64.1%)
[91]. This rate increases dramatically to greater than 50% when lung cancer is diagnosed
at an early stage. However, only 14-16% of cases are detected early [91].
In contrast to breast, colon, and prostate cancer, no routine screening method for
early detection of lung cancer exists. Methods based on imaging (chest X-ray, low dose
spiral computed tomography (LDSCT), autofluorescence bronchoscopy (AFB)), and
sputum cytology have been tested, however, none have proven ideal. Screening via chest
X-ray is not sufficiently sensitive [94], and trials demonstrated that its use in high risk
populations showed no decrease in mortality [62]. LDSCT screening can detect a number
of stage I lung cancers, with survival at 10 years reported as high as 88% [83]. However,
the possibility of lead-time bias and the high false positive rate [35] limit the utility of
this screening modality. These false positive tests frequently lead to invasive procedures
to remove lesions that later prove to be benign [41]. In addition, LDSCT appears to favor
detection of peripheral lesions, being less effective at detecting small pre-invasive/micro-
invasive lesions in the central airways [120]. Its effects on reducing lung cancer mortality
remain in question [6]. Autofluorescence bronchoscopy (AFB) also has a high false
positive rate [48, 81], and preferentially detects centrally located cancers. Screening by
sputum cytology can detect a number of aspymptomatic cases, but it has not been shown
48
to decrease lung cancer mortality [7]. Studies using molecular marker techniques on
sputum samples appear promising [109].
Given the poor five-year survival rates and limitations of current screening
techniques, it is clear that improved methods for early detection of lung cancer are
needed. One strategy is to develop sensitive and specific molecular markers that
distinguish cancer type and subtype, that are detectable in ‘remote’ patient media (e.g.
blood, sputum) by non-invasive/minimally invasive means, and that can be assayed using
a quantitative approach.
DNA methylation has emerged as a prime source of potential cancer-specific
biomarkers. In cancer, despite global DNA hypomethylation, many genes become
hypermethylated. Typically this occurs in CpG rich regions called CpG islands at/near
gene promoters. Methylation often results in the silencing of tumor suppressor or growth
regulatory genes [104]. Such cancer-specific hypermethylation results in differential
DNA methylation profiles between tumor and non-tumor tissues, which can be exploited
to distinguish the two, allowing DNA methylation to serve as a cancer-specific molecular
marker. Using bisulfite treatment, which embeds methylation information in the DNA
sequence, coupled with a sensitive and quantitative real-time PCR-based assay
(MethyLight), hypermethylated CpGs form stable, easily amplifiable, and readily
available biomarkers [43]. As no one locus can be expected to detect all cancers of a
particular type, reactions for multiple loci can be easily combined into panels of markers,
increasing the potential to detect lung cancer in a highly sensitive and specific manner.
Because our end goal is a non-invasive lung cancer detection method using DNA
methylation markers, it is worth noting that DNA hypermethylation has been detected in
49
remote patient media such as sputum, blood [11] and bronchoalveolar lavage (BAL) [39]
from lung cancer patients.
Lung cancer is divided clinically into two major subtypes – the rapidly
progressing small cell lung cancer (SCLC), and the more common non-small cell lung
cancer (NSCLC). As NSCLC accounts for > 85% of all lung cancer cases, and is less
aggressive than SCLC, there is a greater chance for early detection, resulting in increased
patient survival. NSCLC is divided into four major histological subtypes:
adenocarcinoma (AD), squamous cell carcinoma (SQ), large cell carcinoma and others
(carcinoids, neuroendocrine cancers, etc). A comparison of SQ and AD of the lung shows
differences in DNA hypermethylation profiles [45, 50, 167], in expression of therapeutic
targets [178], in the mutational and polymorphic spectra [162, 198] and in gene
expression profiles [133]. The region of the lung in which these tumors usually occur also
differs, with AD typically located at the periphery and SQ arising near the central
airways. Given the distinct nature of SQ and AD, it is to be expected that different
molecular markers would need to be developed to sensitively detect these two types of
lung cancer. We have recently identified a panel of DNA methylation markers for lung
adenocarcinoma [170]. Here we focus on the development of molecular markers for
squamous cell lung cancer.
SQ accounts for 25 – 35 % of all lung cancer cases in the United States [89]. Our
goal was to identify a panel of DNA markers that are frequently and highly methylated in
SQ lung tumors when compared to non-tumor lung. Such a panel may be used for non-
invasive/minimally invasive and potentially subtype-specific early detection of SQ lung
cancer. We envision that in the future, detection of DNA methylation markers in remote
50
media (blood, sputum, bronchoalveolar lavage) might complement less specific imaging-
based lung cancer screening tests, and if sensitivity and specificity are high enough,
might eventually be directly applied to the screening of high risk populations.

Materials and Methods
Tissue samples and DNA extraction
Samples were collected from the Los Angeles County Hospital archives, the Norris
Comprehensive Cancer Center archives and the National Disease Research Interchange
(NDRI). Study subjects included 21 males and 22 females ranging in age from 45 - 84 at
time of diagnosis (median age: 70 years old). Age and gender information was missing
for 2 patients. The study population was primarily Caucasian, with 35 Caucasians, 2
African Americans and race unknown for 8 patients. Information as to tumor stage was
available for 43 of the 45 patients. TNM status was either listed in the pathology report,
or discerned from the report using the International System for Staging Lung Cancer
[124]. This information was used to assign tumor stage. There were 6 stage IA, 25 stage
IB, 7 stage IIB and 5 stage IIIA patients. Sections were cut from separate, histologically
verified, tumor and adjacent non-tumor paraffin blocks. A 5µm slide was haematoxylin
& eosin (H&E) stained and coverslipped for histological confirmation of tumor
histological type, and presence or absence of tumor, by an expert lung pathologist
(MNK). Five adjacent 10µm slides were cut, H&E stained, and tumor or non-tumor
material was manually microdissected. DNA was extracted via proteinase K digestion
[105]. Briefly, cells were lysed in a solution containing 100mM Tris-HCl (pH 8.0), 10
51
mM EDTA (pH 8.0), 1 mg/mL proteinase K, and 0.05 mg/mL tRNA and incubated at
50
o
C overnight. The DNA was bisulfite converted as previously described [185]. All
studies were institutionally approved by the University of Southern California
Institutional Review Board (IRB# HS-016041, HS-06-00447), and the identities of
patients were not made available to laboratory investigators.

Methylation analysis
DNA methylation analysis was done by MethyLight as previously described
[185]. A pre-screen methylation analysis using cell lines and five sets of paired SQ/non-
tumor adjacent lung (distinct from the samples used in this study) were used to screen
over 300 DNA methylation loci, and led to the identification of 42 loci of interest, which
were evaluated in this study. The primer and probe sequences are described in the
supplemental data table. In addition to primer and probe sets designed specifically for the
locus of interest, two internal reference primer and probe sets directed against collagen
and ALU repeats were included in the analysis to normalize for input DNA [44, 184].
The percentage methylated reference (PMR) compares the level of methylation in the
sample to in vitro methylated control DNA. It is calculated by dividing the
GENE:reference ratio of a sample by the GENE:reference ratio of M.SssI-treated in vitro
methylated human DNA and multiplying by 100 [185]. PMRs were individually
calculated using the collagen and ALU controls and then averaged.

52
Statistical analysis
Using PMR as a continuous variable, methylation levels of tumor samples were
compared to adjacent non-tumor lung by means of the Wilcoxon signed rank test. The
large number of loci analyzed increases the potential for false discovery. To counteract
this risk, a multiple comparisons threshold was set and applied to those loci for which no
previous data demonstrated their methylation in SQ of the lung at the time of analysis
(Table 1, last column; [16]). To examine whether tumor-specific hypermethylation was
seen in early as well as later stages of SQ lung cancer, methylation levels in tumor and
adjacent non-tumor tissue were compared for “early” (stages IA and IB, n=31) and more
advanced cancers (stages II and III, n=12), as well as for each individual stage (IA, IB,
IIB and IIIA) using the Wilcoxon test. The same test was applied to the comparison of
methylation levels in tumor samples between the early and advanced cancers.
Associations with gender and age were tested using the Wilcoxon test to compare
methylation levels within the tumor sample collection only. As an indicator of the
potential utility of methylation of these loci as a marker for cancer, Receiver Operating
Characteristic (ROC) curves were calculated for each of our top markers, using the PMR
values for the tumor and adjacent non-tumor lung specimens. All statistical tests were
two-sided. Statistical tests were carried out using JMP (v 5.0.1a, SAS Institute Inc, NC).
To determine which combinations of markers would be most effective to correctly
identify tumor vs. non-tumor samples, we fit a random forest classifier to the data set,
using the R programming language (v 2.5; [86]) and 90 samples and 42 variables. Using
bootstrap samples of the data, we grew a forest of 30,000 trees. Splits were determined
using a random sample of five variables and trees were grown until there was only one
53
observation in each leaf. We determined error rates using the observations that were not
used to generate the trees. For each observation, its outcome was predicted by having the
majority vote from the trees that were generated without the original data point in their
bootstrap sample. These predicted values were compared against the true tissue type to
estimate prediction error.

Results
In an effort to develop sensitive and specific molecular markers for squamous cell
carcinoma (SQ) of the lung, the methylation status of 42 candidate loci was examined in
a collection of 45 tumors and histologically normal adjacent non-tumor lung samples
from the same patients. These 42 loci were identified in a pre-screen examination of the
methylation status of 304 MethyLight reactions on cell lines and a small number of
tumors distinct from the ones used in this study (data not shown). As our aim was to
identify novel high penetrance markers for lung SQ, many loci previously reported as
methylated in NSCLC/SQ were not included in our study due to their lower methylation
frequency. In five of the 42 loci (HRAS, MGMT, MTHFR, PAX8 and SLC38A4), the
region examined is not in a CpG island. In our pre-screen, multiple reactions in and
around the CpG islands of these loci were tested and the chosen reactions showed the
highest methylation in cancer. Paired histologically normal adjacent lung tissue samples,
derived from a separate non-cancer block of the lung cancer patients, were used as
control samples. Thus, our control tissue matched tumor tissue fully with respect to most
variables, including environmental exposures, age, gender, ethnicity and genetic
54
background. The use of paired control tissue from lung cancer patients, which may show
higher background methylation, ensures the identification of markers that are
hypermethylated in a cancer-specific manner. MethyLight provides a quantitative
measure for methylation at each locus; the percentage of methylated reference (PMR)
value reflects the level of DNA methylation at the locus examined compared to in vitro
methylated control DNA.
Figure 2.1 DNA methylation analysis of 42 loci 45 squamous cell lung cancer cases

Figure 2.1: DNA methylation analysis of 42 loci in tumor and adjacent non-tumor tissue from 45
squamous cell lung cancer cases Loci names are indicated on the top, sample identifiers are indicated on
the left hand side. The median PMR is calculated from the PMR value of all positive samples. Samples
with methylation ≥ the median PMR are in black, samples with a PMR < the median are in grey. Samples
in which there was no detectable methylation are marked white.
55
We observed a high methylation frequency (the fraction of samples showing any
methylation) for all 42 loci in both the tumors and the adjacent non-tumor tissues taken
from the same patient (Figure 1, Table 1).
The DNA methylation in histologically normal adjacent non-tumor lung is likely
due, on the one hand, to the sensitivity of MethyLight, and on the other, to age and/or
environmental exposure, and has been observed in other studies [38, 101, 139]. We
examined the statistical significance of differences in DNA methylation levels in tumor
versus adjacent non-tumor tissue using the PMR as a continuous variable. Out of the 42
loci studied, 13 were previously reported to be methylated in NSCLC. Hence, a marker
from these 13 was considered statistically significant if it attained the 0.05 level of
significance without correction for multiple testing. A marker from the remaining 29
targets was declared statistically significant if it exceeded the 5% false-discovery rate
threshold defined using the Benjamini and Hochberg [16] approach. Overall, twenty-five
of the 42 loci examined showed a statistically significant difference (highlighted in italics
in Table 1). Three markers – DIRAS3, MGMT, and HRAS - showed statistically
significant hypermethylation in non-tumor tissue. The importance of this suggested loss
of methylation in the tumors was not further explored here, as we are focused on
identifying positive methylation markers for SQ of the lung. The phenomenon could be
of interest for future studies. The remaining 22 loci were found to be statistically
significantly hypermethylated in the tumors (Table 1). This is the first report of
methylation in any cancer for five loci (CPVL, HOXC9, PAX8, PTPRN2, and SLC38A4),
flagging these loci as potential novel cancer markers. Eight loci (GDNF, MTHFR,
56
OPCML, TNFRSF25, TCF21, PAX8, PTPRN2, and PITX2) showed highly statistically
significant differences with p-values <0.0001.
Potential biomarkers should be effective in all patients regardless of cancer stage,
age, gender or ethnicity. We examined DNA methylation levels in tumors vs. adjacent
non-tumor tissue in relation to tumor stage. Because the number of cases was not very
large, we grouped stage IA and IB cases together (six IA and twenty-five IB), and stages
II and III (no IIA, seven IIB and five IIIA). Each of the eight highly significant loci
showed higher DNA methylation levels in tumors vs. adjacent non-tumor lung in both
early (stage I; n=31, p-value range = 1x10
-7
– 0.0041) and advanced (stage II/III; n=12, p-
value range = 6x10
-5
– 0.0194) lung cancer patients.
57
Table 2.1: Statistical analysis of differences in methylation levels between tumor and
adjacent non-tumor tissues.
a
Human Genome Organization nomenclature. Loci showing a
statistically significant difference in methylation between tumor and non-tumor tissue are highlighted in
italics. The top eight loci are noted in bold. Loci are ranked in order of ascending p-value.
b
Percentage of
samples with positive methylation value.
c
Median percent methylated reference calculated from positive
methylation values.
d
p-value calculated by Wilcoxon signed rank test.
e
To minimize the risk of false
discovery, a false-discovery rate threshold using the Benjamini and Hochberg approach was applied to all
loci not previously found to be methylated in squamous cell lung cancer (see methods).
f
This reaction is not
targeted to a CpG Island
g
This primer/probe set shares homology with the CpG island of the adjacent HNT
gene, which appears to have arisen via gene duplication. * Denotes loci previously reported to be
methylated in lung cancer tumor samples

58

Tumor Adjacent Non-Tumor
Gene Name
a

Methylation
Frequency
b

Median
PMR
c

Methylation
Frequency
b

Median
PMR
c

p-value
d

MC
Corr.
e

GDNF 95% 67.11 95% 3.29 5.0E-11 0.0017
MTHFR
f
100% 56.51 100% 25.95 2.0E-10 *
OPCML
g
95% 19.49 98% 5.80 1.0E-09 *
TNFRSF25 98% 50.52 93% 25.89 2.0E-07 *
TCF21 93% 60.64 88% 11.17 3.0E-07 *
PAX8
f
100% 83.90 100% 69.49 9.0E-06 0.0034
PTPRN2 80% 35.60 58% 4.02 1.0E-05 0.0052
PITX2 93% 19.37 95% 1.50 3.0E-05 0.0069
MT1G 95% 1.89 93% 0.59 0.0001 0.0086
PENK 93% 14.30 95% 6.82 0.0002 0.0103
GP1BB 98% 59.52 100% 42.68 0.0009 *
MGMT
f
100% 26.73 98% 33.96 0.0009 *
SLC38A4
f
56% 6.00 31% 0.00 0.0010 0.0121
SFRP2 98% 10.19 91% 2.67 0.0038 *
RARRES1 60% 9.27 47% 0.71 0.0048 0.0138
DIRAS3 100% 57.25 100% 63.92 0.0074 *
NEUROG1 37% 3.03 17% 0.08 0.0079 0.0155
WDR33 31% 0.34 9% 0.20 0.0089 0.0172
TFAP2A 46% 5.68 20% 10.00 0.0092 0.0190
SFRP1 91% 1.63 98% 0.56 0.0124 *
CYP1B1 58% 3.53 40% 0.41 0.0143 *
HOXC9 66% 10.62 60% 0.48 0.0157 0.0207
ABCB1 100% 27.07 100% 24.33 0.0193 0.0224
HRAS
f
100% 82.08 100% 90.01 0.0215 0.0241
GRIN2B 44% 28.03 31% 1.26 0.0235 0.0259
CACNA1G 93% 0.74 95% 0.44 0.0282 0.0276
HIC1 98% 33.81 100% 23.52 0.0315 0.0293
CPVL 66% 1.65 53% 0.33 0.0739 0.0310
SEZ6L 89% 5.08 95% 3.51 0.0745 0.0328
NEUROD1 44% 9.97 33% 0.70 0.0748 0.0345
CCND2 98% 1.74 96% 1.18 0.0827 0.0362
MINT1 98% 3.18 98% 2.49 0.0879 0.0379
DLEC 55% 19.39 64% 0.17 0.1169 *
RNR1 100% 43.28 100% 31.91 0.1930 0.0397
BLT1 98% 28.01 100% 25.63 0.2420 0.0414
ONECUT2 98% 17.23 100% 14.29 0.3369 *
PLAGL1 100% 45.72 100% 50.38 0.3747 0.0431
GATM 26% 11.95 69% 0.10 0.4462 0.0448
CDX1 98% 46.96 100% 44.40 0.7469 0.0466
TWIST1 66% 1.98 86% 1.07 0.7578 0.0483
TMEFF2 100% 11.54 100% 11.82 0.7591 *
RPA3 49% 0.24 49% 0.16 0.8863 0.0500
59
When analyzing each stage (IA, IB, IIB and IIIA) independently, the two most
significant markers (GDNF and MTHFR) showed significantly higher DNA methylation
levels in tumor vs. adjacent non-tumor in every stage, despite the modest number of
cases. Comparison of DNA methylation levels for the top eight markers in early vs.
advanced cancers showed no significant differences between the methylation levels in
these tumors, reinforcing the idea that these markers are not stage-specific. This is
important, since effective DNA methylation markers for SQ lung cancer must function on
every stage of cancer, but particularly on early stage tumors.
We also examined methylation in tumors in relation to age. HOXC9 showed
higher levels of DNA methylation in patients under the median age (70: p = 0.021) and
TCF21 showed increased DNA methylation in females (p = 0.047). However, if a
multiple comparisons correction were applied, these differences would not be significant.
DNA methylation of PAX8 appeared higher in males (p = 0.001; significant even with
application of a multiple comparison threshold), a factor that might require consideration
if it were to be developed for clinical use. As our population is primarily Caucasian, we
were not able to examine DNA methylation levels in relation to ethnicity. Studies are in
progress in a larger more ethnically diverse population, to examine the possible
relationship of DNA methylation to ethnicity.
To provide more insight into the distribution of DNA methylation levels in the
tumor and non-tumor samples, we plotted the distribution of PMR values for tumor and
non-tumor tissues for the eight most highly significant loci (Figure 2).

60
Figure 2.2 Scatter plot analysis of the top 8 loci (as ranked by p-value)

Figure 2.2: Scatter plot analysis of the top 8 loci (as ranked by p-value). T – tumor, AdjNTL – adjacent
non-tumor lung. This graphic representation of the PMR values for tumor and adjacent non-tumor lung
allows an indication of whether DNA methylation at this locus has potential as use as a biomarker in a
clinical setting. PMR values are represented by black dots. Grey bars represent the median PMR values.

These plots illustrate differences in the nature of these markers that are not
evident from the p-values. For example, GDNF appears to promise substantial specificity
and sensitivity due to frequently highly elevated DNA methylation of this locus in tumor
tissues. A similar pattern is seen in MTHFR, OPCML, and TNFRSF25. For TCF21,
PTPRN2, and PITX2, the DNA methylation levels of tumor tissues show a wider
distribution and more overlap with non-tumor samples. The PAX8 DNA methylation
values were tightly clustered, and while the difference is highly statistically significant (p
61
= 9 x10
-6
), the fold-difference is small, indicating that this marker may not be as useful in
the clinical setting.
The utility of clinical markers is often evaluated by generating a receiver
operating characteristic (ROC) curve, in which sensitivity versus 1-specificity at all
possible cutoff values is plotted. Ultimately, such ROC curves will be generated based on
methylation values detected in remote media. However, here we used ROC curves based
on the tumor and non-tumor PMR values to provide an early indication of the potential of
the top eight loci as cancer-specific markers. The area under the curve (AUC), an
indicator of marker performance, ranged from a modest 0.75 for PITX2 to a much better
0.9 for GDNF (Figure 3). The sensitivity and specificity values for each of the eight top
loci were individually calculated using the present tumor collection in a five-fold cross
validation (Table 2). The quantitative marker values were dichotomized at a level that
would minimize the classification error. Sensitivity ranged from 58-89% and specificity
from 69-100%.
While measurements for several individual markers look promising, it is
unrealistic to expect detection of all cases of a particular type of cancer using a single
biomarker. Thus, our goal is to develop a panel of DNA methylation markers that, used
in combination, can sensitively and specifically detect lung SQ. To assess the
performance of combinations of our markers in the identification of tumors, we fit a
random forest classifier to the data set, using 90 samples and 42 variables.

62
Figure 2.3: Receiver Operator Characteristic curves for the top 8 loci (as ranked by
p-value)

Figure 2.3: Receiver Operator Characteristic (ROC) curves for the top 8 loci as ranked by p-value. AUC –
area under the curve. Plots generated using JMP statistical software.

Using bootstrap samples of the data, we grew a forest of 30,000 trees. Splits
were determined using a random sample of five variables and trees were grown until
there was only one observation in each leaf. When the 42 loci were ranked using the
random forests classifier, the top eight loci were the same as when the data was ranked by
p-value or AUC value, and the order of the ranking is the same for the top four in all
three groups (data not shown). Using all 42 loci in combination, we observed 97.7%
sensitivity and 97.7% specificity. While this is encouraging, 42 loci are too many to test
63
in a clinical setting. Trimming the panel down to just the top eight loci resulted in 95.6%
sensitivity and 95.6% specificity. Further restricting our analysis to the four most highly
ranked loci maintained sensitivity at 95.6% while specificity dropped to 93.3%.
Table 2.2: AUC, Sensitivity & Specificity Analysis

5-fold cross validation
Locus AUC Sensitivity Specificity
GDNF 0.90 0.82 0.98
MTHFR 0.89 0.89 0.82
OPCML 0.87 0.76 0.89
TNFRSF25 0.82 0.76 0.91
TCF21 0.82 0.62 0.91
PAX8 0.77 0.64 0.69
PTPRN2 0.76 0.58 0.89
PITX2 0.75 0.60 1.00

Discussion
Thirteen of the 42 loci examined here were previously reported to be methylated
in lung cancer tumor samples. Consistent with the literature, eight loci (MTHFR,
OPCML, TNFRSF25, TCF21, SFRP2, SFRP1, CYP1B1, GPIΒB, DLEC and ONECUT2)
[20, 36, 50, 72, 78, 118, 126, 134, 151, 170, 177, 199] are hypermethylated in tumor
tissue in our study. Indeed, MTHFR, OPCML, TNFRSF25 and TCF21 show highly
statistically significant differences (p< 1 x10
-6
) between tumor and adjacent non-tumor
tissues in our study. The results for three loci are in contrast with the published literature.
MGMT, DIRAS3 (previously described as ARHI) and TMEFF2 (previously described as
HPP1) have been reported to be hypermethylated in lung cancer [20, 50, 60, 72, 74, 77,
78, 113, 114, 139, 157, 167, 199]. We found that MGMT and DIRAS3 were statistically
significantly more highly methylated in adjacent non-tumor than in SQ samples, while
for TMEFF2, we observed almost no difference in methylation levels between tumor and
non-tumor tissue (Table 1). The differences between our results and the published
64
literature may be due to a variety of reasons, including technical differences (such as the
use of the quantitative MethyLight versus qualitative methylation specific PCR, or the
less sensitive CpG island microarrays), the sampling of a different region of the gene,
differences in the lung cancer histologies studied (many studies contain a mix of NSCLC
samples), and ethnic/racial differences in the patient populations studied. In the case of
MGMT we sampled regions in and out of the CpG island in our pre-screen, and the
region outside of the CpG island looked more promising, and was therefore tested. Thus,
the primer/probe set we used differs from what has been published in the literature.
When examining the function of the 22 statistically significant potential markers
for SQ, four major functional categories emerged. Eight loci encode proteins involved in
signaling and growth regulation, seven loci encode transcription factors, four loci encode
proteins with metabolic function, and three loci belong to no particular group (Table 3).
Our strongest potential biomarkers, the eight most statistically significantly
hypermethylated loci, are scattered across the first three of these groups. Because our
focus is development of DNA methylation markers, our primary concern is consistent
methylation of a particular locus, not whether the associated gene is actually silenced by
methylation. Hence, genes in which the consistently hypermethylated locus is outside of
the CpG island can serve as markers (e.g. HRAS, MGMT, MTHFR, PAX8, SLC38A4),
even though the DNA methylation may not be of functional significance. While we have

65

Table 2.3: Putative biological role of the 22 statistically significantly hypermethylated
loci

Table 2.3:
a
Gene symbol is the Human Genome Organization nomenclature .
b
Gene Name and Function
as listed per [63]

Functional
Categories
Gene
Symbol
a

Gene Name
b
Gene Function
b

Signaling GDNF glial cell derived neurotrophic factor Growth factor
GP1BB glycoprotein I b, beta polypeptide
Platelet membrane
receptor
OPCML opioid binding protein/cell adhesion molecule-like
Cell adhesion
molecule
PENK proenkephalin
Opioid peptide
precursor
PTPRN2
protein tyrosine phosphatase, receptor type, N
polypeptide 2
Phoshatase
SFRP1 secreted frizzled-related protein 1
Wnt Signaling
modulator
SFRP2 secreted frizzled-related protein 2
Wnt signalling
modulator
TNFRSF25
tumor necrosis factor receptor superfamily,
member 25
Cell surface receptor
Transcription
Factor
HOXC9 homeobox C9 Transcription factor
NEUROD1 neurogenic differentiation 1 Transcription factor
NEUROG1 neurogenin 1 Transcription factor
PAX8 paired box gene - 8 Transcription factor
PITX2 paired-like homeodomain transcription factor 2 Transcription factor
TFAP2A transcription factor AP 2 alpha Transcription Factor
TCF21 transcription factor 21 Transcription factor
Metabolism CYP1B1
cytochrome p450 family 1, subfamily B,
polypeptide 1
Liver metabolism
MT1G metallothionein 1G Heavy metal binding
MTHFR
5,10 methylenetetrahydrofolate reductase
(NADPH)
Methyl group
metabolism
SLC38A4 solute carrier family 38, member 4
Amino acid
transporter
Other ABCB1
ATP-binding cassette, sub-family B (MDR/TAP),
member 1
Drug efflux pump
RARRES1 retinoic acid receptor responder 1 Unclear
WDR33 WD repeat domain 33 Unclear
66
not determined whether the genes for our eight top markers are silenced, there is
published evidence for the inactivation of some of these genes in lung cancer. For others,
their expression in cancer has not yet been investigated, and might be worth examining in
future, more mechanistic, studies. As six of the top eight loci show potentially
functionally relevant DNA hypermethylation in tumors, we will discuss what is known
about their role in cancer development.
OPCML, TNFRSF25 and TCF21 have been previously reported to be
hypermethylated in lung cancer [177], [126, 151] and based on their function,
methylation-induced silencing could favor tumor growth. Opioid binding protein/cell
adhesion molecule (OPCML) is an opioid receptor and is involved in cell-cell adhesion. It
binds opioid peptides (e.g. enkephalin) and causes apoptosis of lung cancer cell lines,
indicating it functions as a tumor suppressor gene. This inhibition was reversed by
nicotine [117], which may be of particular interest in lung cancer pathogenesis. It is of
note that PENK, which encodes the precursor peptide of the OPCML ligand enkephalin,
was also found to be significantly hypermethylated in tumor tissue in our studies. This
might suggest methylation-induced silencing of a tumor suppressor pathway. We recently
reported OPCML as highly methylated in lung adenocarcinoma, [170] indicating that it is
a potential AD/SQ lung cancer biomarker.
Tumor necrosis factor receptor superfamily member 25 (TNFRSF25) has been
shown to be methylated in bladder cancer, and very recently methylation in lung SQ was
reported [52, 126]. As this receptor mediates apoptosis, methylation-induced silencing
may facilitate evasion of cell death - a key step in cancer growth. The transcription factor
TCF21 has been reported to be more highly methylated in lung cancer tissue than non-
67
tumor adjacent lung, and overexpression in mouse xenografts results in a reduction in
tumor size and weight [151]. This implies a tumor suppressor function for TCF21,
therefore tumor-associated promoter DNA methylation, and possibly transcriptional
silencing, are not surprising.
For other genes, such as PITX2, PAX8 and PTPRN2, the biological consequences
of DNA methylation remain a question. Functionally, it is unclear how PITX2 silencing
would contribute to lung cancer growth. This member of the paired-like homeodomain
transcription factor family functions in left-right asymmetry in development [18], but has
no described function in adult lung. However, cancer-related methylation is reported in
other tissues in which the gene has no described function, for example, in acute myeloid
leukemia [169], breast cancer [116], and prostate cancer [76]. Interestingly, higher DNA
methylation levels of PITX2 are associated with greater recurrence of both breast and
prostate cancer [76, 116]. Whether such a link exists in lung cancer will require further
studies. Protein tyrosine phosphatase, receptor type, N polypeptide 2 (PTPRN2) is an
autoantigen involved in insulin dependent diabetes mellitus [107]. No previous reports of
methylation of PTPRN2 exist, making it a potentially novel cancer biomarker.
The most intriguing of the identified loci is the top marker GDNF, encoding glial
cell line-derived neurotrophic factor. GDNF has been reported to be overexpressed in
lung tumor tissue [61] and is silent in normal adult lung [53]. As a ligand for the RET
proto-oncogene, GDNF would be a likely candidate for promoting cancer progression,
and has been proposed to do so in pancreatic cancer [58]. DNA methylation of this locus
would seem contradictory. However, the high DNA methylation we report is at promoter
2 (located at the intron 1/exon 2 boundary of GDNF), a promoter that has been shown to
68
have low activity [68]. Indeed, in our preliminary studies, a primer designed against the
primary promoter of GDNF showed no hypermethylation (data not shown). It may be
possible that DNA methylation at the downstream promoter is somehow related to the
transcriptional activity from the upstream promoter. Given the fact that GDNF is, to our
knowledge, the strongest candidate DNA methylation marker for lung SQ identified to
date, this issue would be worth investigating further.
While the top eight markers identified in this study show highly significant
DNA hypermethylation in cancer, it will of course be important to validate these markers
in an independent collection of samples. Such studies are in progress using a specimen
collection balanced for gender and the major ethnic groups in the United States.

Conclusion
Our primary goal is to find sensitive and specific biomarkers for the early
detection of lung cancer. Differences in the biology and treatment of different lung cancer
histological subtypes warrant the development of markers for each cancer subtype. We
have recently reported a panel of DNA methylation markers for lung adenocarcinoma
[170]. Here we report the identification of promising DNA methylation markers for
squamous cell lung cancer. Statistical analysis of the difference in DNA methylation
levels between SQ tumor and adjacent non-tumor lung tissue identified 25 statistically
significant loci. Of these, three are potential negative DNA methylation markers (more
methylated in adjacent non-tumor tissues), while 22 are potential positive DNA
methylation markers. Of the 22 loci, we focused on those eight that were ranked most
significantly hypermethylated in the cancer versus paired non-cancer samples by p-values
69
and ROC curves. These eight loci are significantly hypermethylated in both early (stage
I) and more advanced cancers. Two of those eight loci (PAX8, PTPRN2) have never been
reported to be hypermethylated in human cancer specimens, and thus constitute
promising new candidate cancer markers. To our knowledge, the eight-locus panel
consisting of GDNF, MTHFR, OPCML, TNFRSF25, TCF21, PAX8, PTPRN2 and PITX2,
constitutes the highest sensitivity and specificity DNA methylation marker panel for lung
SQ reported to date. Following its validation on a separate set of tumor and non-tumor
lung samples, the next step will be to examine the DNA methylation of these loci in
remote media (such as blood, sputum, bronchoalveolar lavage) from lung cancer patients
and control non-cancer cases. In conjunction with our work on AD lung cancer and
ongoing studies of other NSCLC subtypes, we hope to develop a panel of markers for the
sensitive and specific detection of non-small cell lung cancer that would also identify the
histological subtype. The further development of DNA methylation markers promises to
be important not only for diagnostics, but also for prognostication, the ability to follow
response to therapy, and guidance in the choice of treatment.

Author’s contributions
PPA was involved in experimental execution and extensive data analysis, drafting the
manuscript, and generation of figures. JSG was involved in marker design, experimental
execution and initial analysis. MNK reviewed all histological slides prior to
microdissection. JAH provided samples and statistical discussions. ST helped locate and
section tissues from the Los Angeles County Hospital and provided the linked and de-
identified clinicopathological information. MC and DJW provided experimental advice
70
and designed several of the MethyLight reactions used in this study. PWL provided
experimental advice and discussion regarding data interpretation. KDS oversaw statistical
analysis and drafted statistical sections of the manuscript. IALO designed the study,
oversaw all aspects of the project, mentored PPA and JSG, and revised manuscript drafts.
All authors reviewed and commented on the manuscript during its drafting and approved
the final version.

Acknowledgements
The authors thank members of the Laird lab for help with MethyLight and probe/primer
design, and Laird-Offringa lab members for critical comments on the manuscript. We
thank Joe Hacia, Gyeong-Hoon Kang, Brian Pike, Jeffrey Tsou and Deborah Weener for
help with primer/probe design. This project was funded by grant support for IALO:
National Institutes of Health/National Cancer Institute R21 CA102247 and R01
CA119029, Whittier Foundation Translational Research Grant, a STOP Cancer award
and generous support by the Kazan, McClain, Abrams, Fernandez, Lyons & Farrise
Foundation and Paul and Michelle Zygielbaum. Two of the cancer samples used in this
study were provided by the Norris Comprehensive Cancer Center’s NIH-funded Slide
Retrieval and Tissue Discard Repository. None of the funding agencies played any role in
the collection, analysis, interpretation of the data, writing of the manuscript, nor the
decision to publish. The content is solely the responsibility of the authors and does not
necessarily represent the official views of the funding agencies.

71
Chapter 3. Validation of a panel of sensitive and specific DNA
methylation markers for squamous cell lung cancer

Abstract
Purpose: Lung cancer is the leading cause of cancer death in men and women in the
United States and Western Europe. Over 160,000 Americans die of this disease every
year. The overall five-year survival rate is 15%, significantly lower than that of other
cancers. Early detection is the key to increasing lung cancer patient survival. DNA
hypermethylation is recognized as an important mechanism for tumor suppressor gene
inactivation in cancer, and could yield powerful biomarkers for early detection of lung
cancer. We have previously identified promising DNA methylation-based biomarkers for
squamous cell (SQ) lung cancer. Here we focused on expanding our biomarker panel and
subsequently validating this panel in an independent population.
Experimental Design: We employed the high-throughput Illumina GoldenGate assay, to
examine DNA methylation in three matched SQ tumors and adjacent non-tumor lung.
Seven loci emerged as interesting, of which 3 had previously been examined using
MethyLight. The remaining 4 were examined in an independent population of 42 tumor
and adjacent non-tumor lung tissue samples from the same patients, using MethyLight.
In order to further develop our biomarker panel a random forests analysis was performed
using data from 42 previously examined loci, and the new 4 loci identified here on 38
cases that overlap between the two studies. A 10-locus panel was identified that can
72
sensitively and specifically detect SQ lung cancer. This panel was validated using
MethyLight in an independent population of 54 matched tumor and adjacent non-tumor
samples.
Results: Two of the 4 loci identified using the GoldenGate assay are combined with the
previously identified markers. This results in a 10-locus panel of highly promising DNA
methylation-based markers for SQ lung cancer, that shows high specificity for cancer
detection in the test and validation panels.
Conclusion: We have expanded our panel of biomarkers to include 10 loci, validated this
panel of biomarkers in an independent population, and observed that this panel can detect
multiple SQ cancers in differing ethnic groups and regardless of stage.

Introduction
Lung cancer is responsible for the highest number of cancer deaths in the United States,
and is predicted to kill more than 160,000 people in 2008 [91] . There are two major
subtypes, the aggressive small cell lung cancer (SCLC) which accounts for 10-15% of all
cancer cases, and the less aggressive non-small cell lung cancer (NSCLC), which is the
remaining 85-90% of cases. NSCLC is histologically further subdivided into four
categories, adenocarcinoma (AD), squamous cell carcinoma (SQ), large cell carcinoma,
and ‘others’ (carcinoids, neuroendocrine cancers) [123]. An essential strategy to reduce
cancer mortality its to develop screening methods for early detection. Such methods are
in place for other commonly diagnosed cancers, for example those of the breast, prostate
and colon, and are likely responsible for the high five-year survival and relatively low
mortality rates of these cancers. No effective early detection method for lung cancer
73
exists. Lung cancer currently has a five-year survival rate of 15%, but when detected at a
localized stage that rate can increase [83] . Hence, effective early detection might
increase five-year survival rates and decrease mortality.
Several strategies for early detection of lung cancer have been employed,
but with limited success. Large-scale trials of sputum cytology analysis and chest x-ray
imaging found no decrease in mortality [7, 62]. Low dose spiral computed tomography,
a highly sensitive screening method, has shown increased detection of lung cancer cases
[83], but it is still widely debated if employing this strategy will reduce lung cancer
mortality [6].
A newer approach to cancer detection is to use molecular markers that
exhibit a cancer-specific change. The goal of this research is to develop biomarkers that
can be detected sensitively and specifically and non-invasively in remote media, for
example in blood.
One potential molecular marker is DNA methylation. This is a normal
process in cells, but in cancer one sees a global decrease in methylation levels with a
concurrent increase in methylation at CpG rich regions (CpG islands; [104]). DNA
methylation is linked with transcriptional silencing, hence this modification in cancer
likely plays a mechanistic role in cancer development and progression [104]. From the
biomarker discovery standpoint, the effect of DNA methylation on gene expression is not
relevant. The important aspect is that methylation at a certain locus should be cancer
specific, and should have a high penetrance. DNA methylation is an ideal biomarker; the
material is stable, easily amplifiable, and its levels in cells can be assessed quantitatively
74
using such as techniques as MethyLight [44], or the Illumina GoldenGate and Infinium
assays.
As NSCLC is the less aggressive subtype, early detection of NSCLC has a
greater potential to have an effect on lung cancer mortality, hence we focus our efforts on
this area. Our specific goal is to develop DNA methylation-based early detection
markers for squamous cell lung cancer. Each histological subtype of lung cancer exhibits
characteristic molecular alterations, including DNA methylation [50, 162, 170, 178, 198].
Development of subtype-specific markers would facilitate earlier diagnosis of disease
subtype and may have important clinical implications. Using a single population, we have
previously identified a promising panel of loci for the sensitive and specific detection of
lung cancer [5] . Potential biomarkers should function in all patients, regardless of age,
gender, ethnicity and stage. In order to test this, any markers identified in an initial
screen must be further validated in a completely independent population. In an ideal
situation, such a population would be large and representative of the aforementioned
population variables.
We proposed to carry out such a validation, but because a new high-
throughput DNA methylation platform became available, we first tested whether we
could identify any new loci to add to our panel. We studied three tumors and paired
adjacent non-tumors from the same patients on the GoldenGate platform. We identified 7
loci of interest, of which 3 had previously been examined using MethyLight did not
appear to be promising biomarkers. The remaining four loci were studied in a population
of 42 tumor and adjacent non-tumor samples using MethyLight. Thirty-eight of the 42
cases used to screen these loci overlap with the 45 cases examined in our previously
75
published study. In order to determine the loci best suited to distinguish tumor from
adjacent non-tumor lung, a random forests analysis was performed on 46 loci (42 from
our original work, 4 identified using the GoldenGate assay), using the 38 common
samples. A 10-locus panel was identified, and further validated in an independent,
ethnically mixed population of 54 tumors and adjacent non-tumors from the same
patients.

Materials and Methods
Study subjects
GoldenGate assay test population. Study subjects included 2 males and 1 female, age
range 59-83. All subjects were Caucasian and all tumors were stage Ib.
Test population for MethyLight analysis of 4 loci: Study subjects included 18 males and
20 females ranging in age from 45 to 86 at time of diagnosis (median Age 70.5 years
old). Gender information is unknown for 4 patients. The study population was primarily
Caucasian (n=32) and included 2 African Americans. Ethnicity is unknown for 8 patients.
Tumor stage information was available for 36 patients. TNM status was either listed in
the pathology report, or discerned from the report using the International System for
Staging Lung Cancer [124]. This information was used to assign tumor stage. There were
6 stage IA, 20 stage IB, 5 stage IIB and 5 stage IIIA patients.
Validation population: Study subjects included 36 males and 18 females ranging in age
from 39 - 83 at time of diagnosis (median age: 66 years old). The study population was
primarily Caucasian (n=23) but included 10 African Americans, 8 Hispanic and 6 Asian
patients. Ethnicity is unknown for 7 patients. Tumor stage information was available for
76
29 patients. TNM status was either listed in the pathology report, or discerned from the
report using the International System for Staging Lung Cancer [124]. This information
was used to assign tumor stage. There were 5 stage IA, 10 stage IB, 1 stage IIA, 9 stage
IIB, 3 stage IIIA patients and 1 stage IV patient.

Tissue samples and DNA extraction
All studies were institutionally approved. Patient identity was not made known to the
laboratory investigators. Sections were cut from histologically verified tumor and
adjacent non-tumor frozen or FFPE blocks. A 5µm slide was haematoxylin & eosin
(H&E) stained and coverslipped for histological confirmation by authors M.N.K. and
S.C. Adjacent 10µm slides were cut, H&E stained, and tumor and non-tumor material
was microdissected from separate slides, with the exception of 2 cases in the validation
population, in which tumor and non-tumor material were microdissected from the same
slide. In these two adjacent regions of tumor and non-tumor material were avoided.
Samples were collected from the USC/Norris Comprehensive Cancer Center Tissue
Discard Repository, the Ontario Tumor Bank, the USC University Hospital archives and
the USC Norris Cancer Center archives. DNA was extracted from tumor and non-tumor
lung samples via proteinase K digestion [105]. Briefly, cells were lysed in a solution
containing 100mM Tris HCl (pH 8.0), 10 mM EDTA (pH 8.0), 1 mg/mL proteinase K,
and 0.05 mg/mL tRNA and incubated at 55
o
C overnight. The DNA was bisulfite
converted as previously described (2). In the case of the three samples applied to the
GoldenGate assay, the DNA was further purified using a phenol/choroform extraction to
ensure DNA was of high quality.
77

Methylation analysis
The Ilumina Golden Gate Cancer Panel 1 was used to examine DNA methylation in 3
patients. The assay was performed as recommended by Illumina. The level of DNA
methylation is reported as a β-value, with 1 being highly methylated and 0 being no
methylation. The β-value is a mean of 30 replicate measurements and is calculated by
dividing Cy5 fluorescence (representing methylation) by total fluorescence (Cy3 + Cy5;
Cy3 fluorescence represents unmethylated DNA). As the sample size was small, (n=3) a
statistical comparison was not possible. Therefore, two measures were applied to rank
the loci in order of ability to distinguish between tumor and adjacent non-tumor. Firstly,
the average β-value for tumor and adjacent non-tumor samples was calculated. Using
this, the fold difference in average β-values was calculated, with any difference > 3.5
considered as a reasonable change. An additional filter to remove loci with background
methylation was applied to the data: the difference in average β-value between tumor and
non-tumor samples must be > 0.3. Such stringent criteria were applied in order to ensure
only strong biomarker candidate loci were identified.
DNA methylation analysis was done by MethyLight as previously described (3). In
addition to primer and probe sets designed specifically for the gene of interest, an internal
reference primer and probe set directed against Alu repeats was included in the analysis
to normalize for input DNA (3). The percentage methylated reference (PMR) was
calculated as the GENE:reference ratio of a sample divided by the GENE:reference ratio
of Sss1-treated in vitro methylated human DNA and multiplying by 100 (2). The PMR
value thus compares the methylation in the sample to fully methylated control DNA.
78

Statistical analysis
Using PMR as a continuous variable, PMR values of tumor samples were compared to
non-tumor samples by means of the Wilcoxon test. In the test of four loci on 42 samples
only one locus MOS, had not been previously reported to be methylated in the literature,
thus a multiple comparisons correction was not warranted. As an indicator of the
potential utility as a marker for cancer, Receiver Operator Characteristic (ROC) curves
were calculated for each of our top markers, using the PMR values for the tumor and
control lung specimens and the statistical software JMP (v 5.0.1a, SAS Institute Inc, NC).
Clinical associations (age, gender), were tested by comparing methylation levels within
the tumor samples only, using the Wilcoxon test. Associations with ethnicity were tested
both using methylation levels in tumors only, and by comparing PMR values between
tumor and non-tumor tissues within each ethnic group, again using the Wilcoxon test.
Associations with stage were tested by comparing PMR values between tumor and non-
tumor tissues for early and advanced stage cancers. All statistical tests were two-sided,
and performed using the statistical software JMP.
To determine which combinations of markers would be most effective to
correctly identify tumor vs. non-tumor samples, we fit a random forest classifier to the
data set, using the R programming language (v 2.5; [86]) and 76 samples and 46
variables. Using bootstrap samples of the data, we grew a forest of 30,000 trees. Splits
were determined using a random sample of five variables and trees were grown until
there was only one observation in each leaf. We determined error rates using the
observations that were not used to generate the trees. For each observation, its outcome
79
was predicted by having the majority vote from the trees that were generated without the
original data point in their bootstrap sample. These predicted values were compared
against the true tissue type to estimate prediction error. To validate this panel in an
independent population we fit a random forest classifier to the data set, using the R
programming language (v 2.5; [86]) and 108 samples and 10 variables.

Results
In an effort to ensure we had identified as many strong candidate biomarker
loci as possible, we examined DNA methylation levels in three cases of tumor and
matched adjacent non-tumor lung from the same patients using the high-throughput
Illumina GoldenGate platform. This assay examines DNA methylation at 1505 loci
spanning 807 genes. As is the case for all the tumor samples used in this report, paired
histologically normal adjacent lung tissue samples were used as control samples. Thus,
our control tissue matched tumor tissue fully with respect to most variables, including
environmental exposures, age, gender, ethnicity and genetic background. The use of
paired control tissue from lung cancer patients, which may show higher background
methylation, ensures the identification of markers that are hypermethylated in a cancer-
specific manner.
Using stringent cutoffs for the fold difference in β-value between tumor and
adjacent non-tumor lung and the difference between average the β-value of tumor and
adjacent non-tumor samples, and a ranking of both criteria, seven potential biomarker
loci were identified (Table 3.1). Three (p16, HOXA11, and TERT) had previously been
80
examined using MethyLight and the data suggested that they would not be strong
markers for SQ lung cancer.

Table 3.1: β-value of seven loci ranked as strongest potential biomarkers in the
GoldenGate assay
Gene
Name
a

NT
1
b

NT
2
b

NT
3
b
T 1
c
T 2
c
T 3
c

Avr
NT
d

Avr
T
e
FD
f

FD
Rank
g

T-
N
h

T-N
Rank
i

Rank
Sum
j

MOS 0.04 0.06 0.13 0.55 0.62 0.36 0.08 0.51 6.69 2 0.43 3 5
HOXA9 0.21 0.20 0.14 0.66 0.91 0.65 0.18 0.74 4.09 5 0.56 1 6
p16 0.20 0.04 0.01 0.39 0.78 0.35 0.08 0.51 6.15 3 0.42 4 7
HOXA11 0.16 0.12 0.15 0.45 0.85 0.44 0.15 0.58 4.01 6 0.44 2 8
PAX6 0.02 0.03 0.00 0.00 0.66 0.29 0.02 0.32 19.07 1 0.30 7 8
TERT 0.07 0.10 0.12 0.58 0.07 0.56 0.10 0.40 4.18 4 0.31 6 10
TAL1 0.05 0.14 0.18 0.57 0.29 0.53 0.12 0.46 3.77 7 0.34 5 12

Table 3.1:
a
Gene Name as listed in the Illumina Cancer Panel 1. NT - Non-Tumor. T - Tumor.
b
β-values
for NT samples.
c
β-values for tumor samples.
d
Average β-value for all non-tumor samples.
e
Average β-
value for all tumor samples.
f
Fold difference between the average β-value for all tumor samples and
average β-value for all non-tumor samples.
g
Ranking of the FD, ranked as largest fold difference has the
lowest number.
h
Difference between the average β-value of all tumor samples and the average β-value of
all non-tumor samples.
i
Rank of the difference in the average β-values, largest difference has the lowest
numbers.
j
Sum of the ranks, lowest number is the highest ranked.

MethyLight primer/probe sets were designed for the remaining four (MOS,
HOXA9, PAX6, and TAL1) and their DNA methylation levels were examined in 42
tumor and adjacent non-tumor lung tissues from the same patients. Modest DNA
methylation of HOXA9, MOS and TAL1 is observed in adjacent non-tumor samples
(Figure 3.1, Table 3.2). This is likely due to environmental exposure as squamous cell
lung cancer is 90% smoking associated [123], but may also be due to the sensitivity of
the MethyLight assay, or aging associated. Using the PMR as a continuous variable, we
examined the statistical significance of differences in DNA methylation levels in tumor
versus adjacent non-tumor tissue. Three of the four loci (HOXA9, MOS and TAL1) were
81
highly statistically significantly hypermethylated (p-value 7 x 10
-9
, 7 x 10
-8
, and 3 x 10
-7

respectively) in tumor tissue when compared to adjacent non-tumor tissue (Table 3.2). To
avoid a spurious association, a multiple comparisons correction is applied to newly
discovered loci. However, in this study MOS is the only locus not previously reported to
be methylated in NSCLC, so such a correction is not warranted.
The ability of a clinical marker is often measured by the receiver operating
characteristic (ROC) curve in which sensitivity versus 1-specificity is plotted. These
curves are ultimately to be used in a clinical setting when these markers are being
detected in remote media, but here we use them as an early indication of the potential
clinical utility of our markers (Table 3.2, last column). HOXA9 and MOS show strong
AUC values (0.90 and 0.82 respectively).

82

Figure 3.1: DNA methylation for 4 loci in 42 tumor and adjacent non-tumor lung
cancer cases.

Figure 3.1: DNA methylation for 4 loci in 42 tumor and adjacent non-tumor lung cancer cases. Patient
samples colored dark gray have methylation levels higher than the overall (tumor and normal) median PMR
of all positive samples. Light Grey indicates lower methylation levels than overall median PMR and white
depicts no methylation
83
Table 3.2: Statistical analysis of differences in methylation levels between tumor and
adjacent non-tumor tissues

Tumor Adjacent Non-Tumor
Gene
Name
a

Methylation
Frequency
b

Median
PMR
c

Methylation
Frequency
b

Median
PMR
c
p-value
d
AUC
e

HOXA9 90.48 34.17 54.76 2.71 1.00E-10 0.90363
MOS 85.71 11.75 76.19 0.30 5.00E-07 0.81689
TAL1 59.52 20.41 28.57 2.10 0.0001 0.72052
PAX6 71.43 4.64 66.67 0.97 0.0912 0.60459

Table 3.2:
a
Human Genome Organization nomenclature. Loci showing a statistically significant difference
in methylation between tumor and non-tumor tissue are in bold. Loci are ranked in order of ascending p-
value.
b
Percentage of samples with positive methylation value.
c
Median percent methylated reference
calculated from positive methylation values.
d
p-value calculated by Wilcoxon signed rank test.
e
Area under
the curve (AUC) is an indicator of sensitivity and specificity.

A further indication of the ability of the loci to distinguish tumor from
adjacent non-tumor lung is the distribution of the PMR values (Figure 3.2). This
represents information that cannot be seen from the p-values. Examining these plots we
see a strong separation of PMR values for tumor and adjacent non-tumor lung for
HOXA9 and MOS, while there is more overlap between the PMR values for TAL1.

Figure 3.2: Scatter plot representation of PMR values

Figure 3.2: Scatter plot representation of PMR values of the 3 significant loci (as ranked by p-value) in 42
tumor and adjacent non-tumor lung samples. PMRs for tumor (T) are on the left and adjacent non-tumor
(adj - NT) on the right. PMR values are represented by black dots. Grey bars represent the median PMR
values.

Potential biomarkers should work in all patients regardless of age, gender,
ethnicity and should detect all stages of cancer. We examined DNA methylation levels in
tumors vs. adjacent non-tumor tissue in relation to tumor stage. Because the number of
84
cases was not very large, we grouped stage IA and IB cases together (six IA and twenty
IB), and stages II and III (no IIA, five IIB and five IIIA). The loci showed higher DNA
methylation levels in tumors vs. adjacent non-tumor lung in early (stage I; n=36, p-value
range = 6x10
-6
– 0.0007) lung cancer patients (Table 3.3). However, in advanced stage
patients only HOXA9 shows a statistically significant difference between tumor and
adjacent non-tumor tissue (p = 0.0017; Table 3.3). MOS approaches significance (p =
0.053; Table 3.3) We also examined methylation in tumors in relation to age and gender
and observed no statistically significant differences. As our population is primarily
Caucasian, we were not able to examine DNA methylation levels in relation to ethnicity.

Table 3.3: Comparison of DNA methylation levels in Tumor Stage

Early Stage (n=26)
b
Advanced Stage (n=10)
c

Gene Name
a

p-value
d
p-value
d

HOXA9 1.00E-06 0.0017
MOS 6.00E-06 0.053
TAL1 0.0007 0.0752

Table 3.3: Loci highlighted in grey show statistically significant differences in DNA methylation between
tumor and adjacent non-tumor tissue from the same patient. Genes ranked by statistical significance in all
samples.
a
Human Genome Organization nomenclature.
b
Early Stage consisted of 6 stage IA and 20 stage
IB tumors.
c
Advanced Stage consisted of 5 stage IIB, and 5 stage IIIA.
d
p-values calculated using the
Wilcoxon signed rank test.

In order to determine which of the 46 loci are the best discriminators of
tumor and adjacent non-tumor tissue, we conducted a random forests analysis. As there is
an overlap of 38 cases between our initial study (42 loci on 45 cases) and our current
work (4 loci on 42 cases), the random forests analysis was conducted using all 46 loci on
these 38 cases. Using a Gini index cutoff of >1.5, a10-locus panel was identified (Table
85
3.4) that strongly discriminates tumor from adjacent non-tumor tissue with 97.4%
sensitivity and 97.4% specificity.
Table 3.4: Importance Measure of biomarker panel
Gene Name
a
Importance Measure
b

GDNF 5.57
HOXA9 3.47
MOS 3.23
MTHFR 2.35
OPCML 2.33
PITX2 2.01
TNFRSF25 1.93
TCF21 1.81
PTPRN2 1.56
MT1G 1.53

Table 3.4: Loci are ranked in order of decreasing importance measure.
a
Human Genome Organization
nomenclature.
b
Importance measure based on random forests analysis

To examine the validity of this 10-locus panel for cancer detection, we
examined the DNA methylation status of these loci in an independent population of 54
tumor and matched adjacent non-tumor samples. We observed high methylation
frequency in tumor tissue and moderate to high methylation frequency in adjacent non-
tumor samples (Figure 3.3). As previously discussed, the methylation in the adjacent non-
tumor tissue could be attributed to the sensitivity of the MethyLight assay, environmental
exposure, or aging.
86

Figure 3.3: DNA methylation analysis of 10 loci in 54 tumor and adjacent non-
tumor lung cancer cases

Figure 3.3: DNA methylation of the 10-locus panel in 54 tumor and adjacent non-tumor lung cancer cases.
Patient samples colored dark gray have methylation levels higher than the overall (tumor and normal)
median PMR of all positive samples. Light Grey indicates lower methylation levels than overall median
PMR and white depicts no methylation.

We performed statistical analysis of the DNA methylation levels using
PMR as a continuous variable. All loci attained statistical significance (p <0.05; Table
3.5). Seven loci (GDNF, HOXA9, TCF21, MOS, OPCML, MT1G, and MTHFR) show
highly statistically significant differences in tumor versus adjacent non-tumor tissue (p-
87
value ≤ 0.00005; Table 3.5). As all the loci examined in this study were previously
demonstrated to be methylated, no multiple comparisons correction was applied.

Table 3.5: Statistical analysis of differences in methylation levels between tumor and
adjacent non-tumor tissue
Tumor Adjacent Non-Tumor
Gene Name
a

Methylation
Frequency
b

Median
PMR
c

Methylation
Frequency
b

Median
PMR
c
p-value
d

AUC
Value
e

GDNF 90.74 26.65 51.85 0.62 7.00E-10 0.84
HOXA9 90.74 34.83 50.00 0.90 7.00E-10 0.84
TCF21 98.15 37.52 98.15 8.67 3.00E-08 0.81
MOS 75.93 16.71 35.19 0.99 7.00E-08 0.79
OPCML
g
83.33 6.01 44.44 0.73 1.00E-07 0.79
MT1G 77.78 13.47 40.74 0.81 8.00E-06 0.74
MTHFR
f
98.15 47.81 88.89 25.51 5.00E-05 0.73
PTPRN2 81.48 5.91 62.96 0.63 0.0001 0.71
TNFRSF25 94.44 29.62 83.33 10.36 0.001 0.69
PITX2 77.78 9.02 64.81 2.04 0.0076 0.65

Table 3.5:
a
Human Genome Organization nomenclature. Loci showing a statistically significant difference
in methylation between tumor and non-tumor tissue are in bold. Loci are ranked in order of ascending p-
value.
b
Percentage of samples with positive methylation value.
c
Median percent methylated reference
calculated from positive methylation values.
d
p-value calculated by Wilcoxon signed rank test.
e
Area under
the curve (AUC) is an indicator of sensitivity and specificity.
f
This reaction is not targeted to a CpG Island
g
This primer/probe set shares homology with the CpG island of the adjacent HNT gene, which appears to
have arisen via gene duplication

Again, we examine the distribution of the methylation levels for each locus
to gain more insight into its potential use in a clinical setting. GDNF, HOXA9, MOS,
OPCML, TAL1 MT1G and PTPRN2 show a pattern of highly elevated DNA methylation
levels in tumor tissue, and generally lower methylation levels in non-tumor tissue. This
separation in the DNA methylation levels bodes well for their use in a clinical setting. In
contrast, MTHFR, TCF21, TNFRSF25 and PITX2 show less separation between tumor
and non-tumor tissues, with a substantial overlap in both.
88

Figure 3.4: Scatter plot analysis of the 10-locus panel

Figure 3.4: Scatter plot representation of PMR values the 10 loci in 54 tumor and adjacent non-tumor lung
samples. PMRs for tumor (T) are on the left and adjacent non-tumor (adj - NT) on the right. PMR values
are represented by black dots. Grey bars represent the median PMR values.

Any potential biomarker should work in all patients, regardless of gender, age, ethnicity
or clinical stage of cancer. We examined DNA methylation levels tumors in relation to
age and gender. OPCML (p = 0.0172) and TAL1 (p = 0.0496) showed higher DNA
methylation levels in tumors of patients above the median age of the population (66). In
relation to gender, PTPRN2 (p = 0.0438) showed higher methylation in tumor samples in
males. To examine DNA methylation levels in relation to ethnicity two approaches were
taken. Firstly, DNA methylation levels in tumors were compared between Caucasian
(n=23) versus non-Caucasian (n=24) patients.
89

Figure 3.5: Receiver Operating Characteristic (ROC) curves for the panel of 10
DNA methylation-based biomarkers

Figure 3.5: Receiver Operating Characteristic (ROC) curves for the panel of 10 DNA methylation-based
markers.

No significant difference in the DNA methylation levels was observed
(Table 3.6; first column). We then performed a comparison for each of the 10 loci of
DNA methylation levels in tumor and adjacent non-tumor tissue in each ethnic group
separately (Table 3.6). No one locus serves to detect cancer in all four ethnic groups.
GDNF, TCF21 distinguish tumor from non-tumor tissue in Caucasian, African American
and Hispanic groups. HOXA9 and MTHFR function as potential biomarkers in the
Caucasian and African American groups, while MT1G, OPCML and MOS function in
Caucasian and Hispanic groups. PITX2 detects cancer only in the Caucasian group. It is
clear that number of loci that serve as potential biomarkers in each ethnic group decreases
as the groups size does, hence the small sample size is likely a confounding factor.

90
Table 3.6: Comparison of DNA methylation levels across ethnic groups

Caucasian vs Non-
Caucasian
a

Caucasian
b
(n=24)
African-
American
b
(n=10)
Latino
b

(n=8)
Asian
b
(n=6)
Gene Name p-value
c
p-value
c
p-value
c
p-value
c
p-value
c

GDNF
0.1419 2.00E-06 0.0008 0.0405 0.5211
HOXA9 0.3543 4.00E-07 0.0007 0.263 0.37
TCF21
0.0721 0.0007 0.0412 0.0074 0.1282
MOS 0.2969 1.00E-04 0.0898 0.0092 0.1463
OPCML 0.6008 3.00E-05 0.1258 0.0246 0.468
MT1G 0.5769 2.00E-05 0.8361 0.031 0.455
MTHFR 0.1391 0.0479 0.0283 0.4309 0.0927
PTPRN2
0.6615 0.0051 0.2742 0.0653 0.5725
TNFRSF25 0.8066 0.1135 0.1304 0.0905 0.3785
PITX2
0.5059 0.0266 0.9377 0.7124 0.1424

Table 3.6: Loci highlighted in grey show statistically significant differences in DNA methylation between
tumor and adjacent non-tumor tissue from the same patient. Genes ranked by statistical significance in all
samples.
a
Comparison of PMR values tumor and non-adjacent tumor samples from the same patient.
b
Comparison of Tumor PMR values for Caucasian (n=23) versus Non-Caucasian samples (n=24).
c
p-values
calculated using the Wilcoxon signed rank test.

To examine wither the DNA methylation markers were significant for early
versus later stage cancer, cases were divided into two groups, early stage, consisting of
10 IB and 5 IA, versus more advanced stage cancers (1 stage IIA, 9 stage IIB, 3 stage
IIIA and 1 stage IV. Comparing DNA methylation levels between tumor and adjacent
non-tumor tissue for each group showed that not all loci can distinguish tumor and non-
tumor in early and late stage cancers. Six loci (GDNF, HOXA9, TCF21, MOS, OPCML,
and MT1G) can distinguish tumor from non-tumor lung in both early and late stage
cancers (Table 3.7). MTHFR and TNFRSF25 specifically detect early stage cancers,
while PTPRN2 and PITX2 specifically detect late stage cancers (Table 3.7). A
comparison of DNA methylation levels in tumors between the two groups showed no
significant difference.
91

Table 3.7: Comparison of DNA methylation levels in Tumor Stage
Early Stage (n=15)
b
Advanced Stage (n=14)
c

Gene Name
a
p-value
d
p-value
d

GDNF 0.0006 0.0043
HOXA9 8.00E-05 0.0017
TCF21 0.0014 0.0016
MOS 0.0001 0.0243
OPCML 0.0005 0.0075
MT1G 0.0131 0.0022
MTHFR 0.0048 0.4341
PTPRN2 0.084 0.0161
TNFRSF25 0.0037 0.1129
PITX2 0.0674 0.0161

Table 3.7: Loci highlighted in grey show statistically significant differences in DNA methylation between
tumor and adjacent non-tumor tissue from the same patient. Genes ranked by statistical significance in all
samples.
a
Human Genome Organization nomenclature.
b
Early Stage consisted of 5 stage IA and 10 stage
IB tumors.
c
Advanced Stage consisted of 1 stage IIA, 9 stage IIB, 3 stage IIIA and 1 stage IV.
d
p-values
calculated using the Wilcoxon signed rank test

To assess the performance of the 10-locus marker panel in the independent
population, the classifiers derived from the random forest analysis for the training set (38
tumors and 38 matched non-tumor samples) was applied to the test set (54 tumors and 54
matched non-tumors). The 10-locus panel achieved lower sensitivity in the test
population (68.5%), but retained high specificity (92.6%).

Discussion
As no one locus can be expected to identify all cases of a given cancer, it is
useful to have a panel of markers which can increase the sensitivity and specificity of
detection. Previously we had examined 42 loci for their ability to serve as DNA
methylation-based biomarkers of SQ lung cancer, and we identified a panel of 8 markers
with high sensitivity and specificity [5]. In order to expand our marker panel, and to
92
ensure that we were screening the strongest potential biomarkers, we undertook a study
of DNA methylation in SQ lung cancer using the high-throughput GoldenGate assay. In
analyzing this data, we applied very stringent criteria for locus selection, in order to
identify additional highly promising DNA methylation markers.
To further assess the four new loci identified in this analysis (HOXA9,
MOS, TAL1 and PAX6), their DNA methylation was examined in using MethyLight. To
initially characterize each locus, and compare its performance to the markers identified in
our previous studies, we analyzed the criteria that serve as indicators of the individual
behavior of each locus as a biomarker; methylation frequency in tumor and non-tumor
lung, p-value, AUC value, and scatter plots. Two of these loci (HOXA9 and MOS)
demonstrated highly significant differences in methylation between tumor and non-tumor
lung, high AUC values, and a strong separation of PMR values for tumor and adjacent
non-tumor tissues (Table 3.2, Figure 3.1, Figure 3.2). This, coupled with the fact that
these loci detect early as well as stage cancer (Table 3.3), suggests that HOXA9 and
MOS may be useful biomarkers for early detection of SQ lung cancer.
Combining the newly identified loci with our previously published 42 loci
allowed us to redefine our biomarker panel. A random forests analysis was run on all 46
loci in the 38 cases of tumor and adjacent non-tumor that were used in both studies. This
identified a ten-locus panel consisting of GDNF, HOXA9, TCF21, MOS, OPCML,
MT1G, MTHFR, PTPRN2, TNFRSF25 and PITX2. The revised marker panel shows
which very high sensitivity (97.4%) and specificity (97.4%) in the training set, similar to
the sensitivity and specificity reported for our original panel. As indicated by the
93
individual measures of clinical usefulness of loci as biomarkers in our test population,
HOXA9 and MOS are strong biomarkers and are included in our revised marker panel.
While this 10-locus panel appears strong in the population in which it was
developed, and panel must be further validated in an independent population to confirm
that the ability of the markers to distinguish tumor from adjacent non-tumor is not simply
a property of the population in which they were developed. We examined the DNA
methylation status of the loci in an ethnically mixed collection of 54 tumor and adjacent
non-tumor samples from the same patients for further study.
An examination of the individual measures of the clinical potential of this
panel in this new population is quite encouraging. All loci are statistically significant,
with seven loci having highly significant p-values (p ≤ 0.00005). AUC values range from
0.65 – 0.84. Examining the PMR distribution plots, we see that the majority of the loci
have a clear separation between PMR values for tumor and adjacent non-tumor samples,
implying that they would be clinically applicable. This panel has the characteristics
required of potential biomarkers. The top 6 (as ranked by p-value) of these 10 loci can
distinguish tumor from non-tumor tissue regardless of stage, while PTPRN2 and PITX2
appear to detect advanced stage cancers and MTHFR and TNFRSF25 detect early stage
cancers (Table 3.4). In relation to ethnicity, it is unclear whether each locus could detect
cancer in all ethnic groups. The small sample size of the individual ethnic groups may be
limiting, in particular in the case of Asians. In the population with the largest sample size
(Caucasian) all loci, except TNFRSF25, can detect cancer, while in the smallest
population (Asian) no loci appear to show significant DNA methylation differences
94
between tumor and non-tumor tissue. A larger sample size in each of the non-Caucasian
ethnic groups would be needed to confirm these results.
The performance of the panel on this independent population was
determined by a random forests analysis, using cutoff values determined in the training
population, hence it is a very stringent evaluation of this panel. While sensitivity
decreased, specificity remained very high. However, sensitivity to detect lung cancer is
less of an issue because highly sensitive imaging methods are available.
It is very encouraging that the specificity remains high. The primary
concern in screening for lung cancer is specificity, as false positive results of a lung
cancer test have serious consequences. They can result in a decreased quality of life for
patients, and in follow-up procedures that are expensive, invasive and have associated
morbidity [102]. Thus, our modest sensitivity, but 92.6% specific panel could be used in
concert with highly sensitive (but non-specific) imaging techniques, such as low dose
spiral computed tomography (LDSCT).

Conclusion
The successful identification of two (HOXA9 and MOS) additional DNA
methylation markers for SQ lung cancer using the GoldenGate assay indicates the
promise of this assay for identify potential biomarkers of disease. We have thus expanded
our previously reported DNA methylation-based marker panel to 10 loci, and have
validated this panel on an independent population. In doing so, we have developed a
moderately sensitive, but specific panel for early detection of lung cancer. Used in
concert with sensitive imaging techniques, such as LDSCT, this panel could facilitate
95
early detection of lung cancer. Having identified potential biomarkers in primary tissue,
the next step is to evaluate the ability of these markers to sensitively and specifically
detect SQ lung cancer when screened in remote media, such as blood, sputum or
broncheoalveolar lavage. To our knowledge, this is the largest and most promising panel
of markers for SQ lung cancer described to date.

Acknowledgements
I would like to thank Dr. Dan Weisenberger of the USC Epigenome Center for
performing the bisulfite conversion, organizing these samples on the Golden Gate assay,
and discussion of the results. I would also like to thank Dr. Kimberly Siegmund and
Amit Joshi for performing the random forests analysis, and for discussion of the data.

96
Chapter 4. A high-throughput approach for identifying DNA
methylation-based biomarkers of squamous cell lung cancer

Abstract
Lung cancer is the number one cancer killer in the United States. Compared to other
commonly diagnosed cancers, lung cancer has a very low five-year survival rate of 15%.
In an effort to reduce the mortality rate of this disease, early detection strategies have
been attempted. Imaging (chest X-ray and low dose spiral computed tomography
(LDSCT)) and sputum cytology analysis as screening approaches have thus far failed in
this endeavor. Research focus has shifted to developing molecular markers for lung
cancer that can be assayed in remote media (blood, sputum, bronchoalveolar lavage).
DNA methylation at CpG islands is a promising and powerful molecular marker in a
wide range of cancers. We have recently identified a panel of loci for sensitive and
specific detection of squamous cell lung cancer. However many more CpGs remain to be
evaluated. This can be done with new high-throughput platforms such as the Illumina
GoldenGate assay. It is unclear whether this platform provides reproducible data, and if
DNA derived from formalin fixed paraffin embedded tissues can be used. We have
demonstrated that the GoldenGate assay has a high reproducibility between independent
experiments, and show that FFPE tissues can generate high quality data on the
GoldenGate assay. Furthermore, we have identified 157 potential DNA methylation-
based biomarkers for SQ lung cancer.

97
Introduction
Lung cancer is the leading cancer killer, and is responsible for more that 160,000 deaths
annually in the Unites States [91]. Clinically, this cancer is divided into two major
subtypes, small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC). The
former is a very aggressive subtype and accounts for 10-15% of all lung cancer cases.
NSCLC accounts for 85-90% of all lung cancers and is further subdivided into four
histological groups: adenocarcinoma (AD), squamous cell carcinoma (SQ), large cell
carcinoma and ‘others’ (carcinoids, neuroendocrine cancers). The histological subtypes
of lung cancer exhibit different molecular profiles [162, 178, 198], and DNA methylation
profiles [4, 50, 170]. Biomarkers that can detect not only cancer, but also cancer subtype
could have important clinical applications, providing an earlier and more accurate
diagnosis of cancer. We are interested in developing early detection biomarkers for
squamous cell (SQ) lung cancer. SQ lung cancer accounts for 30% of all lung cancer
cases, and is 90% smoking associated [123]. As smoking is a risk factor, it becomes
easier to identify the high-risk population to which a screening strategy would be applied.
While a decline in smoking in the developed countries has caused a reduction in the
number of SQ cases, the increased use of tobacco products in developing countries
ensures that this cancer will continue as a public health problem [123].
A reason for the high mortality rate of lung cancer is the lack of a reliable
early detection method. For cancers such as breast and prostate, screening methods exist
that facilitate early detection and result in high five-year survival rates (88% and 99%
respectively). In contrast, lung cancer has a 15% five-year survival rate [91]. A
successful screening strategy facilitates the detection of a cancer at a stage when
98
resection and/or treatment can result in a decreased mortality rate and an increased five-
year survival rate. Screening strategies employing chest X-ray, spiral CT imaging, and
sputum cytology have been tested, however none has thus far proven effective in
decreasing mortality [7, 62]. However, the imaging methods are still under evaluation,
and the results of ongoing studies will help to determine whether their application to a
high-risk population results in a decrease in lung cancer mortality. (PLCO trial, [7]
LDSCT trials [6]).
While the fate of imaging based methods hangs in the balance, research has
turned to molecular markers as potential early detection methods. Molecular markers are
typically DNA or protein, and usually involve a modification thereof, or increase in
levels of, that is specific for cancer. One potentially powerful molecular marker is DNA
methylation. This modification of DNA consists of the enzymatic addition of a methyl
group to the 5’ position of cytosine in the context of a cytosine:guanine (CpG)
dinucleotide, and is a normal process in cells. In cancer, gene specific hypermethylation
is seen at CpG-rich regions known as CpG islands (>200 base pairs, CG content of >
0.55; [137, 160]). These islands are commonly associated with the promoter region of
genes, but are also found in introns and exons. Methylation at promoter CpG islands
frequently coincides with gene silencing [104], and mechanistically this is important for
cancer. However, from the biomarker discovery standpoint the location of the CpG island
and the effect of its methylation on gene expression are not relevant. Instead, the
important question is whether the methylation observed is cancer-specific and has high
penetrance, increasing its potential as a molecular marker. Several reports of DNA
99
methylation-based biomarkers of lung cancer describe panels of markers for cancer
detection [17, 50, 136, 148, 167, 170]
Multiple methods to assay DNA methylation are available. More traditional
methods, for example, MethyLight, methylation specific PCR, and bisulfite genomic
sequencing are relatively low throughput, but excellent for the detailed investigation of a
small number of loci. In biomarker discovery, the more loci one can examine, the higher
the potential for finding optimal biomarkers of the disease of interest. Recently, in the
epigenomic era, high-throughput methods that examine DNA methylation at large
numbers of loci have been developed. Methylation microarrays have been reported to
assay 8091 genes [80], and 12,192 CpGs [134]. In addition, Illumina Inc. has recently
developed bead-binding based high-throughput methods (www.illumina.com). In this
report we use the GoldenGate DNA methylation assay from Illumina Inc. This technique
utilizes a 96 well plate format. The bead-based assay allows the rapid and cost effective
study of DNA methylation at 1505 loci from 807 genes in each well. Two issues pertinent
to using this new assay are, its reproducibility, and the source material that can be used.
In a previous experiment using the GoldenGate assay (PAPER REF?), we
had examined DNA methylation in three cases of SQ lung cancer using tumor and
adjacent non-tumor samples from the same patients. In this work, these three samples
were again assayed, using new sections from the same tissue blocks. As different
sections, and hence DNA from a separate extraction were used, this cannot be considered
a technical replicate. However, by comparing results between independent experiments
using the same patient samples, this data affords us some opportunity to gain an
understanding of the reproducibility of this assay.
100
Patient tissues removed during resection are preserved using two methods,
freezing and formalin fixation followed by paraffin embedding (FFPE), the latter being
the most commonly used method. A significant difference between the two methods is
the quality of the preserved DNA. DNA from FFPE tissue shows extensive degradation
that increases with the age of the sample [138]. In contrast, the DNA from frozen tissue
samples is much better preserved. However, frozen tissue samples are in limited supply
compared to FFPE samples. Hence, if one can assay DNA from FFPE tissues then larger
numbers of tumor and adjacent non-tumor samples become available for study, allowing
mining of DNA methylation data in these collections.
The goals of this work were, to examine the reproducibility of the assay by
comparing 3 cases of tumor and adjacent non-tumor from the same patients that were
assayed independently using different sections from the tissue blocks, to test whether
DNA from FFPE preserved tumors can be used on this platform, and to apply a high-
throughput method for the identification of new biomarkers for squamous cell lung
cancer.

Materials and Methods
Study Subjects
The FFPE population was composed of 2 males and 3 females, and consisted of 4
Caucasians and ethnicity unknown for one patient. The median age of the population was
65. Stage information was available for all patients, and the population consisted of 1
stage Ia patient, 2 stage IIb patients and 2 stage IIb. The frozen population had 4 males
and 1 female and all are Caucasian. The median age of the population was 69. Stage
101
information was available for all patients, and the population consisted of 4 stage Ib
patients, and 1 stage IIb patients.

Tissue samples and DNA extraction
Samples were collected from the USC Norris Cancer Center archives and the National
Disease Research Interchange (NDRI). Frozen tissues were from the USC Norris Cancer
Center, while formalin fixed paraffin embedded (FFPE) samples were from the NDRI.
Sections were cut from separate, histologically verified, tumor and adjacent non-tumor
paraffin or frozen blocks. A 5µm slide was haematoxylin & eosin (H&E) stained and
coverslipped for histological confirmation of tumor type, and presence or absence of
tumor, by an expert lung pathologist (MNK). Four adjacent 10µm slides were cut, H&E
stained, and tumor or non-tumor material was manually microdissected. DNA was
extracted via proteinase K digestion [105]. Briefly, cells were lysed in a solution
containing 100mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), 1 mg/mL proteinase K,
and 0.05 mg/mL tRNA and incubated at 50
o
C overnight. DNA extraction for the initial
study of the three repeated samples is described in Chapter 3. The DNA was bisulfite
converted using the EZ DNA methylation kit (Zymo Research Orange, CA). All studies
were institutionally approved by the University of Southern California Institutional
Review Board (IRB# HS-016041, HS-06-00447), and the identities of patients were not
made available to laboratory investigators.

102
Methylation Analysis
Ten tumor and ten paired adjacent non-tumor samples from the same patients (five from
frozen samples, five from FFPE samples) were assayed using the Illumina Golden Gate
Assay. Samples were recovered from the bisulfite conversion process in 18 µl. 3 µl was
used for quality control analysis. 15 µl was sent to the USC Genomics Core and the
Golden Gate assay using Cancer Panel 1 was performed as recommended by Illumina
Inc. (www.illumina.com). The level of DNA methylation is reported as a β−value, with 1
representing most highly methylated and 0 indicating no detectable methylation. Each β−
value is a mean of 30 replicate measurements and is calculated by dividing Cy5
fluorescence (representing methylation) by total fluorescence (Cy3 + Cy5; Cy3
fluorescence represents unmethylated DNA).

Data Analysis
To examine reproducibility of the Illumina assay we compared results from two
independent experiments conducted on different sections from the same set of 3 tumor
and adjacent non-tumor tissues from the same patients which were preserved as frozen
tissue. The DNA was extracted two separate times for the experiments, and in the first
experiment it was phenol/chloroform extracted prior to analysis. In the interval between
the two experiments it became clear that this extra purification step was not necessary to
have good quality DNA for analysis on this platform. Two analyses were employed.
Firstly, a comparison of the number of loci ranked to be important in each experiment,
and secondly the overlap in these loci was determined. As only 3 cases were used, p-
value and AUC analysis are not informative, and hence were not used. The fold
103
difference in the β−value between the average β−value of the tumors and the average β−
value of the non-tumor samples was calculated, and samples were filtered such that the
average β−value was >0.1. A β−value of below 0.1 is considered background
methylation. All loci with a fold difference >2 and average tumor β−value >0.1 were
considered interesting. To rank the loci the data was sorted by relevance of each measure,
for fold difference and average tumor β−value it was in descending order. A rank of 1
was given to the best value, 2 to the next best, 3 to the next best etc. Identical values
were given the same rank, and the next value was that rank number plus n, where n
equals the number of identical values. For example, if there were 3 best values, then they
would all be ranked 1, and the next highest value would be ranked 4. The sum of the
ranks for the measures was calculated and the loci were ranked based on this, with the
lowest rank sum being the best ranked. Secondly, a plot of the β−values for each tumor
and adjacent non-tumor tissue was generated in which each sample was compared to
itself in the independent experiments. Graphs were generated using the R statistical
software (http://www.r-project.org/).
In order to evaluate which genes were most able to distinguish tumor from
adjacent non-tumor lung three measures of the data were employed. Using β−value as a
continuous variable, methylation levels of tumor samples were compared to adjacent non-
tumor lung by means of the Wilcoxon signed rank test, generating a p-value. The area
under the curve for tumor and adjacent non-tumor samples was also calculated, using the
R statistical software (http://www.r-project.org/). Finally, the fold difference in the β−
value between the average β−value of the tumors and the average β−value of the non-
tumor samples was calculated. To rank the genes in order of importance, a ranking was
104
assigned to each of the three measures of the data. The data was sorted by relevance of
each measure, for p-value it was sorted in ascending order, and for fold difference and
AUC it was in descending order. A rank of 1 was given to the best value (for example,
lowest p-value, highest fold difference/AUC), 2 to the next best, 3 to the next best etc.
Identical values were given the same rank, and the next value was that rank number plus
n, where n equals the number of identical values. The sum of the ranks for each of the
three measures was calculated and the loci were ranked based on this, with the lowest
rank sum being the best ranked. Based on this ranking the top 500 genes were selected.
An additional filter was placed on these genes, the average β−value of tumor samples had
to be greater than >0.1. A β−value of below 0.1 is considered background methylation.
The data was analyzed in three groups: all samples as one data set, frozen samples only,
and FFPE samples only.

Results
DNA isolated from independent sections of three of the frozen tissues
samples used in this study had been previously examined on the Illumina GoldenGate
assay (as discussed in chapter 3). In an effort to assess the reproducibility of the assay we
analyzed the data from these three cases separately and compared the results to those
obtained previously. In the prior analysis, 65 loci were identified as potential biomarkers
of squamous cell lung cancer as opposed to 104 loci in the second analysis. Fifty of the
65 loci identified in the first analysis are also found in the second analysis, indicating
good reproducibility of the data. All loci selected for further validation from the first
analysis (Chapter 3) were also found in the second analysis, and ranked highly. In a
105
further examination of the reproducibility of the assay the β−values for the tumor and
adjacent non-tumor samples from the independent experiments were plotted against each
other. In this analysis, the high R
2
values (range: 0.971 – 0.988) demonstrated a strong
correlation between β−values from independent experiments, and hence clearly
demonstrates the reproducibility of this assay (Figure 4.1).

Figure 4.1: Correlation plots of the β-values between two independent experiments

Figure 4.1: Plot of β−values for each tumor and non-tumor sample against itself in independent
experiments.

Three genes, OPCML, PITX2 and PENK overlap between the loci
identified as promising candidate markers of SQ lung cancer in the previously performed
MethyLight studies (Chapters 2 & 3) and the 1505 loci present on the Illumina platform.
106
Of these three, PITX2 and PENK are in the top ranked loci in both frozen and FFPE
tissues. OPCML is found only frozen tissues, where it has a very low rank.
We were curious as to how well DNA from FFPE tissues performed on the
Illumina GoldenGate platform, and hence we examined the data from these samples
separately. Using the 5 cases preserved by FFPE, 276 loci of interest were identified,
compared to 290 loci identified using frozen tissues (Figure 4.2). In an effort to assess the
quality of the data, we calculated how many non-informative values were found in each
sample. These non-informative values have no β−value and are represented as n/a in the
raw data. The percent failure rate was below 1% for FFPE samples; tumor percent failure
range 0.13-0.27%, non-tumor percent failure range 0.20-0.86%. The failure rate did not
appear to differ much between the tissue preservation methods (frozen tumor range 0.07-
0.33 %, frozen non-tumor range 0.20-0.53%). To further examine if FFPE tissues are
viable for use in this method we compared the commonality in loci identified as potential
biomarkers from each tissue type. It should be noted that two very small and potentially
very different sets of SQ tumors are being compared. An overlap of 158 loci was seen
between the two tissue types. Examining the commonality in the top 50 loci (by rank) in
frozen and FFPE samples shows that 12 loci overlap (Table 4.1), and the ranking position
differs between tissue preservation method. To ascertain whether there were any clear
differences in the tumor population that could account for this low overlap in common
loci, we examined age, ethnicity, gender and stage for the two sample populations. The
median age for the FFPE and frozen tissue populations were similar (65 and 69
respectively), and the nine samples for which ethnicity is known are Caucasian. Tumor
stage was available for all samples, and both populations have early stage tumors. The
107
FFPE population consists of 1 stage 1a, 2 stage IIa and 2 stage IIb patients, while the
frozen population had 4 stage 1b and 1 stage IIb patients. However, the gender make-up
of the population did differ. The FFPE population was composed of 2 males and 3
females, and frozen population had 4 males and 1 female.

Table 4.1: Common loci in the top 50 loci in Frozen and FFPE fixed tissues
Frozen
a
Ranking - Frozen
b
FFPE
c
Ranking - FFPE
b

p16_seq_47_S188_R 3 CDH13_E102_F 1
HOXA9_E252_R 6 SLC22A3_E122_R 2
MOS_E60_R 7 HTR1B_P222_F 5
HOXA11_P698_F 8 HOXA9_E252_R 7
HTR1B_P222_F 11 MOS_E60_R 9
HOXA5_P1324_F 15 TAL1_P594_F 10
HOXA9_P303_F 16 p16_seq_47_S188_R 13
TAL1_P594_F 17 HOXA11_P698_F 16
CDH13_E102_F 19 RARA_P1076_R 23
HOXB13_P17_R 21 HOXA5_P1324_F 25
RARA_P1076_R 43 HOXA9_P303_F 33
SLC22A3_E122_R 46 HOXB13_P17_R 36

Table 4.1:
a
Comparison of common loci between Frozen and FFPE tissues for the top 50 ranked loci in
each tissue type. Loci listed in order of ranking within the top 50 loci.
b
Ranking number reflects the
position of the locus in the top 50 ranked loci.
c
Comparison of common loci between FFPE and Frozen
tissues for the top 50 ranked loci in each tissue type
108

Figure 4.2: Venn diagram of the number of loci that are ranked as potential
biomarkers from each data set

Figure 4.2: Venn diagram of the discriminators of tumor and adjacent non-tumor tissue in each data set
(frozen tissues, FFPE tissues, and combined frozen and FFPE) and the overlap between each data set.

To identify new biomarkers of squamous cell lung cancer we examined DNA
methylation differences between tumor and adjacent non-tumor tissues. When all 10
cases were studied as one data set, 290 loci were identified as potential DNA methylation
markers for squamous cell lung cancer. When the loci identified in each of the three data
sets were compared 157 overlap (Figure 4.2), and these are most promising biomarkers.
These loci are not further examined in this study, but they will be further studied in
combination with GoldenGate methylation data from adenocarcinoma cases to determine
biomarkers that are common, as well as specific for, both subtypes (manuscript in
preparation).

109
Discussion
Comparing the data from two independent experiments conducted on
separate sections of three identical tumor and adjacent non-tumor samples clearly
revealed that that the reproducibility of this assay is high. The number of loci identified
as potential biomarkers in both experiments differs, but there is strong overlap in which
loci were identified. The correlation plots (Figure 4.1) comparing each tumor and non-
tumor sample to itself in the independent experiments clearly demonstrated a high
reproducibility between experiments. This analysis is independent of ranking or locus
selection and examines each of the 1505 β−values in each sample. It is therefore
unbiased and provides a clear view of how reproducible the data is. A further indication
of the validity of this assay is that the loci found to be potential biomarkers in our earlier
studies (Chapters 2 & 3) that are also examined on the GoldenGate assay, are identified
as promising biomarkers using the GoldenGate platform.
Many loci were identified in FFPE tissue, a comparable number to those
identified in frozen tissue, and the DNA was of high quality, indicating that one can use
FFPE tissues on this assay and obtain good quality data. In a comparison of the loci
identified as potential biomarkers in each tissue preservation method, frozen and FFPE,
there is an overlap of approximately 50% (Figure 4.2). The order of ranking of the top
ranked loci differs in each of the tissue preservation methods (Table 4.1). Both of these
observations may be cause for concern. However, the DNA quality appears to be high,
the median age of both populations is similar, and both populations are Caucasian,
therefore these factors are unlikely to account for a difference. Gender differences may
play a role, as the frozen tissue population was primarily male, however in previous
110
studies the differences in DNA methylation between genders has been minimal. The
difference in stage distribution between the populations may also have an effect. The
most plausible explanation is that the sample size of both populations was small (n=5),
and these tissues were from different patients and different sources. A further study with
a larger population would determine if this small overlap and differential ranking of top
loci is a consequence of small sample size and differing populations.
As the data was examined as three separate data sets the most stringent filter
for potential biomarkers could be to select those that are common in all data sets as this
selects potential biomarkers that function in independent populations. Using this
criterion, 157 loci appear to discriminate tumor from adjacent non-tumor tissue and are
potential biomarkers (Figure 4.2). To further characterize these potential biomarkers, one
must validate them in a larger, independent population. This will be performed using a
MethyLight primer/probe set, ideally one targeted to the same CpGs examined in the
Illumina assay. The potential biomarkers identified in this study can be used in
combination with data generated on adenocarcinoma samples to identify potential
biomarkers that are both common to, and specific for each lung cancer subtype.
The large number of loci ranked highly as discriminators between tumor
and adjacent non-tumor tissues implies that the GoldenGate assay is a strong candidate
for identifying potential SQ lung cancer biomarkers (Figure 4.2).

Conclusion
The combination of the identification of the same loci in independent
experiments, the strong overlap with loci identified using MethyLight, and the tight
111
correlation of β−values between independent experiments on the same samples indicates
that the reproducibility of this assay is high. Using FFPE tissues one can generate high
quality data on this platform and these tissues can be used for biomarker discovery. In
this study we have identified a collection of 157 loci that overlap between all data sets
and are potential biomarkers for SQ lung cancer. This data set demonstrates the promise
and reliability of the Illumina GoldenGate platform for identifying potential disease
biomarkers from both frozen and FFPE preserved tissue samples.

Acknowledgements
I would like to thank Dr. Dan Weisenberger of the USC Epigenome Center
for performing the bisulfite conversion, organizing these samples on the Golden Gate
assay, and discussion of the results. I would also like to thank Dr. Fei Pan of the USC
Epigenome Center for kindly generating the AUC and p-values used in this analysis. I
would like to thank Dr. Kim Siegmund for her analysis of the samples run two
independent times on different plates.
112
Chapter 5. Discussion

The goal of my thesis has been the identification of DNA methylation-based biomarkers
for development into early detection markers of squamous cell (SQ) lung cancer. In
Chapter 2 I describe our original work to identify a panel of eight markers that can
sensitively and specifically detect SQ lung cancer samples when compared to matched
adjacent non-tumor tissue. The comparison to matched adjacent non-tumor tissue means
that our control tissue fully matches our tumor tissue with respect to age, gender,
environmental exposure, genetic background and ethnicity. This stringent control ensures
that all putative biomarkers identified in our studies are cancer specific. While the
markers we identified were the strongest in the field, a new high-throughput technology
became available, the Illumina GoldenGate assay. Using this platform we examined
DNA methylation in a small set of 3 SQ tumors and matched adjacent non-tumor lung.
With the goal of identifying powerful biomarkers we applied stringent filtering steps
when selecting loci of interest. In Chapter 3 I describe the identification of 4 additional
candidate markers which were then evaluated in more detail. In combination with our
previously identified markers, we thus obtained a 10-marker panel that showed great
promise for sensitively and specifically detecting SQ lung cancer when applied to our
training set of tumor and matched adjacent non-tumor samples. We tested this panel on
an independent multiethnic set of samples. We found that sensitivity was reduced, but
specificity remained high. In biomarker discovery, the more loci one can examine, the
higher the potential to discover powerful potential biomarkers. The GoldenGate assay,
which examines DNA methylation at 1505 CpG sites in 807 genes is a potentially
113
powerful tool for biomarker identification. In Chapter 4 I describe how we further
explored the use of this technology. We had the opportunity to examine the
reproducibility of this assay and to examine if we can obtain high quality data when using
DNA isolated from formalin fixed paraffin embedded (FFPE) tissues. Our results reflect
high reproducibility, and indicate that FFPE tissues can be used to generate high quality
data. Formalin fixation followed by paraffin embedding is the more common method for
tissue sample fixation. If tissues preserved using this method can be used, then many
more samples are available for study. In an expansion of the work discussed in Chapter 3,
we assayed the same 3 cases a second time, and an additional 7 tumor and matched
adjacent non-tumor cases. An analysis of the ability to distinguish tumor from adjacent
non-tumor lung revealed 157 potential biomarkers. This is a larger number than our
previous experiment using 3 samples, and this is likely due to the larger sample size, and
the use of different filters. How to filter these large data sets is an area under constant
development.
We have developed a panel of specific markers that can detect early and
advanced stage tumors and function in multiple ethnic groups. While we are reporting
strong potential biomarkers, we are a long way from a clinical diagnostic test. The10-
locus panel from Chapter 3 appears to function well in Caucaian, African American and
Latino populations, but its ability to detect Asian cases remains in question. Further
studies using a larger population of Asian cases are required to answer this question. Our
final panel has moderate sensitivity and high specificity. Ideally, both of these measures
would be high. However, the most important measure when screening for lung cancer is
specificity. Specificity is the measure of false positive cases, and this is the least desired
114
result when screening any population. These false positives result in decreased quality of
patient life and often lead to invasive follow up procedures that have associated
morbidity and mortality. Sensitive imaging techniques (such as LDSCT) could be used in
concert with our markers to improve the specificity of detection, hence the moderate
sensitivity or our panel is of less concern. If this 10-locus panel is strong enough to be
developed into a clinical assay, it would be used in concert with a highly sensitive
imaging technique in the screening of a high-risk population. Another potential
application would be the identification of cancer using fine needle aspirate material from
a biopsy.
Our panel of markers has been developed and tested only in primary tissue
from lung cancer patients. In a non-invasive clinical diagnostic test primary tissue should
not be the source material. Hence, potential markers must be detectable in remote media,
such as blood, sputum or broncheoalveolar lavage. The next step for our 10-locus marker
panel is to test if we can detect their methylation in these body fluids. We currently have
a collection of tumor and matched plasma from the patients. Using this we can determine
if the loci are methylated in the tumors, and then examine the plasma for the presence of
methylation. Parallel studies will need to be done for sputum and broncheoalveolar
lavage. Such studies afford us the opportunity to see what percentage of patients with
methylation in tumors also have methylation in the remote media, and to gauge how
sensitive and specific of our marker panel may be.
Each of these media present individual problems for use in a patient
screening method. Blood would be the ideal remote medium, as it is easy to obtain and
non-invasive for the patient. Some of the loci in our marker panel have been reported to
115
be methylated in other cancers, hence, DNA methylation of a locus in blood could be due
to cancer in any organ. DNA methylation detection coupled with imaging would help to
resolve this issue. Sputum contains cells/DNA primarily from the central regions of the
lungs, and its production is associated with smoking. While this hinders its use for
screening of all NSCLC patients, sputum is a potentially effective remote media for SQ
lung cancer as this cancer is generally located centrally in the lungs, and is heavily
smoking associated. While broncheoalveolar lavage could be studied in all NSCLC
patients, it is not an ideal remote medium as a somewhat invasive procedure is required to
obtain it.
If our markers are detectable in remote media, and show strong specificity,
then this furthers their clinical potential. The next step would be to test their ability to
function in a screening strategy. This would be applied to a high-risk population, in the
case of SQ lung cancer that population would be heavy smokers, and in concert with a
sensitive imaging technique. The purpose of a screening strategy is to detect cancers at a
stage where a therapeutic intervention can result in a patient being in remission. The
effectiveness of such a strategy is measured by its effect on mortality. To determine if a
combination of marker testing and imaging would be effective, one would conduct a
randomized control trial. This would be a huge effort, involving patients from multiple
treatment centers, and take many years. If a decrease in morality were to be seen, then
this screening strategy could be routinely applied.
With a combination of our existing biomarker collection, ongoing efforts to
develop more markers, and parallel efforts in adenocarcinoma in the lung, we are taking
the baby steps to developing a clinical test for NSCLC. Such a test is a long way from
116
being available, but with the strong collection of markers being identified, and the
promising results of other studies in remote media, it is starting to appear as a speck on
the horizon.
117
Bibliography

1 American Cancer Society [http://www.cancer.org]

2 Agathanggelou A, Dallol A, Zochbauer-Muller S, Morrissey C, Honorio S,
Hesson L, Martinsson T, Fong KM, Kuo MJ, Yuen PW, Maher ER, Minna JD and Latif
F: Epigenetic inactivation of the candidate 3p21.3 suppressor gene BLU in human
cancers. Oncogene 2003, 22(10):1580-1588.

3 Ahrendt SA, Chow JT, Xu LH, Yang SC, Eisenberger CF, Esteller M,
Herman JG, Wu L, Decker PA, Jen J and Sidransky D: Molecular detection of tumor
cells in bronchoalveolar lavage fluid from patients with early stage lung cancer. J
Natl Cancer Inst 1999, 91(4):332-339.

4 Anglim P: Identification of a Panel of Sensitive and Specific markers for
detection of Squamous Cell Lung Cancer. Mol Cancer 2008(In Press).

5 Anglim PP, Galler JS, Koss MN, Hagen JA, Turla S, Campan M,
Weisenberger DJ, Laird PW, Siegmund KD and Laird-Offringa IA: Identification of a
panel of sensitive and specific DNA methylation markers for squamous cell lung
cancer. Mol Cancer 2008, 7(1):62.

6 Bach PB, Jett JR, Pastorino U, Tockman MS, Swensen SJ and Begg CB:
Computed tomography screening and lung cancer outcomes. Jama 2007, 297(9):953-
961.

7 Bach PB, Kelley MJ, Tate RC and McCrory DC: Screening for lung
cancer: a review of the current literature. Chest 2003, 123(1 Suppl):72S-82S.

8 Bain C, Feskanich D, Speizer FE, Thun M, Hertzmark E, Rosner BA and
Colditz GA: Lung cancer rates in men and women with comparable histories of
smoking. J Natl Cancer Inst 2004, 96(11):826-834.

9 Bastarrika G, Garcia-Velloso MJ, Lozano MD, Montes U, Torre W, Spiteri
N, Campo A, Seijo L, Alcaide AB, Pueyo J, Cano D, Vivas I, Cosin O, Dominguez P,
Serra P, Richter JA, Montuenga L and Zulueta JJ: Early lung cancer detection using
spiral computed tomography and positron emission tomography. Am J Respir Crit
Care Med 2005, 171(12):1378-1383.

10 Bearzatto A, Conte D, Frattini M, Zaffaroni N, Andriani F, Balestra D,
Tavecchio L, Daidone MG and Sozzi G: p16(INK4A) Hypermethylation detected by
fluorescent methylation-specific PCR in plasmas from non-small cell lung cancer.
Clin Cancer Res 2002, 8(12):3782-3787.

118
11 Belinsky SA: Gene-promoter hypermethylation as a biomarker in lung
cancer. Nat Rev Cancer 2004, 4(9):707-717.

12 Belinsky SA, Grimes MJ, Casas E, Stidley CA, Franklin WA, Bocklage TJ,
Johnson DH and Schiller JH: Predicting gene promoter methylation in non-small-cell
lung cancer by evaluating sputum and serum. Br J Cancer 2007, 96(8):1278-1283.

13 Belinsky SA, Klinge DM, Dekker JD, Smith MW, Bocklage TJ, Gilliland
FD, Crowell RE, Karp DD, Stidley CA and Picchi MA: Gene promoter methylation in
plasma and sputum increases with lung cancer risk. Clin Cancer Res 2005,
11(18):6505-6511.

14 Belinsky SA, Liechty KC, Gentry FD, Wolf HJ, Rogers J, Vu K, Haney J,
Kennedy TC, Hirsch FR, Miller Y, Franklin WA, Herman JG, Baylin SB, Bunn PA and
Byers T: Promoter hypermethylation of multiple genes in sputum precedes lung
cancer incidence in a high-risk cohort. Cancer Res 2006, 66(6):3338-3344.

15 Belinsky SA, Nikula KJ, Palmisano WA, Michels R, Saccomanno G,
Gabrielson E, Baylin SB and Herman JG: Aberrant methylation of p16(INK4a) is an
early event in lung cancer and a potential biomarker for early diagnosis. Proc Natl
Acad Sci U S A 1998, 95(20):11891-11896.

16 Benjamini Y and Hochberg Y: Controlling the false discovery rate: a
practical and powerful approach to multiple testing. J R Statist Soc B 2005, 57:289-
300.

17 Bibikova M, Lin Z, Zhou L, Chudin E, Garcia EW, Wu B, Doucet D,
Thomas NJ, Wang Y, Vollmer E, Goldmann T, Seifart C, Jiang W, Barker DL, Chee MS,
Floros J and Fan JB: High-throughput DNA methylation profiling using universal
bead arrays. Genome Res 2006, 16(3):383-393.

18 Blum M, Steinbeisser H, Campione M and Schweickert A: Vertebrate left-
right asymmetry: old studies and new insights. Cell Mol Biol (Noisy-le-grand) 1999,
45(5):505-516.

19 Brabender J, Usadel H, Danenberg KD, Metzger R, Schneider PM, Lord
RV, Wickramasinghe K, Lum CE, Park J, Salonga D, Singer J, Sidransky D, Holscher
AH, Meltzer SJ and Danenberg PV: Adenomatous polyposis coli gene promoter
hypermethylation in non-small cell lung cancer is associated with survival. Oncogene
2001, 20(27):3528-3532.

119
20 Brabender J, Usadel H, Metzger R, Schneider PM, Park J, Salonga D, Tsao-
Wei DD, Groshen S, Lord RV, Takebe N, Schneider S, Holscher AH, Danenberg KD and
Danenberg PV: Quantitative O(6)-methylguanine DNA methyltransferase
methylation analysis in curatively resected non-small cell lung cancer: associations
with clinical outcome. Clin Cancer Res 2003, 9(1):223-227.

21 Brena RM, Morrison C, Liyanarachchi S, Jarjoura D, Davuluri RV,
Otterson GA, Reisman D, Glaros S, Rush LJ and Plass C: Aberrant DNA methylation
of OLIG1, a novel prognostic factor in non-small cell lung cancer. PLoS Med 2007,
4(3):e108.

22 Breuer RH, Snijders PJ, Sutedja GT, Sewalt RG, Otte AP, Postmus PE,
Meijer CJ, Raaphorst FM and Smit EF: Expression of the p16(INK4a) gene product,
methylation of the p16(INK4a) promoter region and expression of the polycomb-
group gene BMI-1 in squamous cell lung carcinoma and premalignant
endobronchial lesions. Lung Cancer 2005, 48(3):299-306.

23 Burbee DG, Forgacs E, Zochbauer-Muller S, Shivakumar L, Fong K, Gao
B, Randle D, Kondo M, Virmani A, Bader S, Sekido Y, Latif F, Milchgrub S, Toyooka S,
Gazdar AF, Lerman MI, Zabarovsky E, White M and Minna JD: Epigenetic inactivation
of RASSF1A in lung and breast cancers and malignant phenotype suppression. J
Natl Cancer Inst 2001, 93(9):691-699.

24 Carpagnano GE, Foschino-Barbaro MP, Mule G, Resta O, Tommasi S,
Mangia A, Carpagnano F, Stea G, Susca A, Di Gioia G, De Lena M and Paradiso A: 3p
microsatellite alterations in exhaled breath condensate from patients with non-small
cell lung cancer. Am J Respir Crit Care Med 2005, 172(6):738-744.

25 Carpagnano GE, Foschino-Barbaro MP, Resta O, Gramiccioni E and
Carpagnano F: Endothelin-1 is increased in the breath condensate of patients with
non-small-cell lung cancer. Oncology 2004, 66(3):180-184.

26 Carpagnano GE, Resta O, Foschino-Barbaro MP, Gramiccioni E and
Carpagnano F: Interleukin-6 is increased in breath condensate of patients with non-
small cell lung cancer. Int J Biol Markers 2002, 17(2):141-145.

27 Carter D, Vazquez M, Flieder DB, Brambilla E, Gazdar A, Noguchi M,
Travis WD, Kramer A, Yip R, Yankelevitz DF and Henschke CI: Comparison of
pathologic findings of baseline and annual repeat cancers diagnosed on CT
screening. Lung Cancer 2007, 56(2):193-199.

28 Chan EC, Lam SY, Tsang KW, Lam B, Ho JC, Fu KH, Lam WK and
Kwong YL: Aberrant promoter methylation in Chinese patients with non-small cell
lung cancer: patterns in primary tumors and potential diagnostic application in
bronchoalevolar lavage. Clin Cancer Res 2002, 8(12):3741-3746.
120

29 Chang HC, Cho CY and Hung WC: Downregulation of RECK by
promoter methylation correlates with lymph node metastasis in non-small cell lung
cancer. Cancer Sci 2007, 98(2):169-173.

30 Chen H, Suzuki M, Nakamura Y, Ohira M, Ando S, Iida T, Nakajima T,
Nakagawara A and Kimura H: Aberrant methylation of FBN2 in human non-small
cell lung cancer. Lung Cancer 2005, 50(1):43-49.

31 Choi N, Son DS, Song I, Lee HS, Lim YS, Song MS, Lim DS, Lee J, Kim
H and Kim J: RASSF1A is not appropriate as an early detection marker or a
prognostic marker for non-small cell lung cancer. Int J Cancer 2005, 115(4):575-581.

32 Cirincione R, Lintas C, Conte D, Mariani L, Roz L, Vignola AM, Pastorino
U and Sozzi G: Methylation profile in tumor and sputum samples of lung cancer
patients detected by spiral computed tomography: a nested case-control study. Int J
Cancer 2006, 118(5):1248-1253.

33 Cooper WN, Dickinson RE, Dallol A, Grigorieva EV, Pavlova TV, Hesson
LB, Bieche I, Broggini M, Maher ER, Zabarovsky ER, Clark GJ and Latif F: Epigenetic
regulation of the ras effector/tumour suppressor RASSF2 in breast and lung cancer.
Oncogene 2008, 27(12):1805-1811.

34 Cortese R, Hartmann O, Berlin K and Eckhardt F: Correlative gene
expression and DNA methylation profiling in lung development nominate new
biomarkers in lung cancer. Int J Biochem Cell Biol 2008, 40(8):1494-1508.

35 Crestanello JA, Allen MS, Jett JR, Cassivi SD, Nichols FC, 3rd, Swensen
SJ, Deschamps C and Pairolero PC: Thoracic surgical operations in patients enrolled
in a computed tomographic screening trial. J Thorac Cardiovasc Surg 2004,
128(2):254-259.

36 Dai Z, Lakshmanan RR, Zhu WG, Smiraglia DJ, Rush LJ, Fruhwald MC,
Brena RM, Li B, Wright FA, Ross P, Otterson GA and Plass C: Global methylation
profiling of lung cancer identifies novel methylated genes. Neoplasia 2001, 3(4):314-
323.

37 Dai Z, Popkie AP, Zhu WG, Timmers CD, Raval A, Tannehill-Gregg S,
Morrison CD, Auer H, Kratzke RA, Niehans G, Amatschek S, Sommergruber W, Leone
GW, Rosol T, Otterson GA and Plass C: Bone morphogenetic protein 3B silencing in
non-small-cell lung cancer. Oncogene 2004, 23(20):3521-3529.

\

121
38 Dammann R, Strunnikova M, Schagdarsurengin U, Rastetter M, Papritz M,
Hattenhorst UE, Hofmann HS, Silber RE, Burdach S and Hansen G: CpG island
methylation and expression of tumour-associated genes in lung carcinoma. Eur J
Cancer 2005, 41(8):1223-1236.

39 de Fraipont F, Moro-Sibilot D, Michelland S, Brambilla E, Brambilla C and
Favrot MC: Promoter methylation of genes in bronchial lavages: a marker for early
diagnosis of primary and relapsing non-small cell lung cancer? Lung Cancer 2005,
50(2):199-209.

40 Di Vinci A, Gelvi I, Banelli B, Casciano I, Allemanni G and Romani M:
Meth-DOP-PCR: an assay for the methylation profiling of trace amounts of DNA
extracted from bodily fluids. Lab Invest 2006, 86(3):297-303.

41 Diederich S and Wormanns D: Impact of low-dose CT on lung cancer
screening. Lung Cancer 2004, 45 Suppl 2:S13-19.

42 Dunn JR, Panutsopulos D, Shaw MW, Heighway J, Dormer R, Salmo EN,
Watson SG, Field JK and Liloglou T: METH-2 silencing and promoter
hypermethylation in NSCLC. Br J Cancer 2004, 91(6):1149-1154.

43 Eads CA, Danenberg KD, Kawakami K, Saltz LB, Blake C, Shibata D,
Danenberg PV and Laird PW: MethyLight: a high-throughput assay to measure DNA
methylation. Nucleic Acids Res 2000, 28(8):E32.

44 Eads CA, Danenberg KD, Kawakami K, Saltz LB, Danenberg PV and Laird
PW: CpG island hypermethylation in human colorectal tumors is not associated
with DNA methyltransferase overexpression. Cancer Res 1999, 59(10):2302-2306.

45 Ehrich M, Field JK, Liloglou T, Xinarianos G, Oeth P, Nelson MR, Cantor
CR and van den Boom D: Cytosine methylation profiles as a molecular marker in
non-small cell lung cancer. Cancer Res 2006, 66(22):10911-10918.

46 Esteller M, Corn PG, Baylin SB and Herman JG: A gene
hypermethylation profile of human cancer. Cancer Res 2001, 61(8):3225-3229.

47 Esteller M, Sanchez-Cespedes M, Rosell R, Sidransky D, Baylin SB and
Herman JG: Detection of aberrant promoter hypermethylation of tumor suppressor
genes in serum DNA from non-small cell lung cancer patients. Cancer Res 1999,
59(1):67-70.

48 Feller-Kopman D, Lunn W and Ernst A: Autofluorescence bronchoscopy
and endobronchial ultrasound: a practical review. Ann Thorac Surg 2005,
80(6):2395-2401.

122
49 Feng Q, Hawes SE, Stern JE, Wiens L, Lu H, Dong ZM, Jordan CD, Kiviat
NB and Vesselle H: DNA methylation in tumor and matched normal tissues from
non-small cell lung cancer patients. Cancer Epidemiol Biomarkers Prev 2008,
17(3):645-654.

50 Field JK, Liloglou T, Warrak S, Burger M, Becker E, Berlin K, Nimmrich I
and Maier S: Methylation discriminators in NSCLC identified by a microarray
based approach. Int J Oncol 2005, 27(1):105-111.

51 Fong KM, Sekido Y, Gazdar AF and Minna JD: Lung cancer. 9:
Molecular biology of lung cancer: clinical implications. Thorax 2003, 58(10):892-900.

52 Friedrich MG, Weisenberger DJ, Cheng JC, Chandrasoma S, Siegmund
KD, Gonzalgo ML, Toma MI, Huland H, Yoo C, Tsai YC, Nichols PW, Bochner BH,
Jones PA and Liang G: Detection of methylated apoptosis-associated genes in urine
sediments of bladder cancer patients. Clin Cancer Res 2004, 10(22):7457-7465.

53 Fromont-Hankard G, Philippe-Chomette P, Delezoide AL, Nessmann C,
Aigrain Y and Peuchmaur M: Glial cell-derived neurotrophic factor expression in
normal human lung and congenital cystic adenomatoid malformation. Arch Pathol
Lab Med 2002, 126(4):432-436.

54 Fujiwara K, Fujimoto N, Tabata M, Nishii K, Matsuo K, Hotta K, Kozuki
T, Aoe M, Kiura K, Ueoka H and Tanimoto M: Identification of epigenetic aberrant
promoter methylation in serum DNA is useful for early detection of lung cancer.
Clin Cancer Res 2005, 11(3):1219-1225.

55 Fukami T, Fukuhara H, Kuramochi M, Maruyama T, Isogai K, Sakamoto
M, Takamoto S and Murakami Y: Promoter methylation of the TSLC1 gene in
advanced lung tumors and various cancer cell lines. Int J Cancer 2003, 107(1):53-59.

56 Fukasawa M, Kimura M, Morita S, Matsubara K, Yamanaka S, Endo C,
Sakurada A, Sato M, Kondo T, Horii A, Sasaki H and Hatada I: Microarray analysis of
promoter methylation in lung cancers. J Hum Genet 2006, 51(4):368-374.

57 Fukui T, Kondo M, Ito G, Maeda O, Sato N, Yoshioka H, Yokoi K, Ueda
Y, Shimokata K and Sekido Y: Transcriptional silencing of secreted frizzled related
protein 1 (SFRP 1) by promoter hypermethylation in non-small-cell lung cancer.
Oncogene 2005, 24(41):6323-6327.

58 Funahashi H, Okada Y, Sawai H, Takahashi H, Matsuo Y, Takeyama H and
Manabe T: The role of glial cell line-derived neurotrophic factor (GDNF) and
integrins for invasion and metastasis in human pancreatic cancer cells. J Surg Oncol
2005, 91(1):77-83.

123
59 Furonaka O, Takeshima Y, Awaya H, Ishida H, Kohno N and Inai K:
Aberrant methylation of p14(ARF), p15(INK4b) and p16(INK4a) genes and location
of the primary site in pulmonary squamous cell carcinoma. Pathol Int 2004,
54(8):549-555.

60 Furonaka O, Takeshima Y, Awaya H, Kushitani K, Kohno N and Inai K:
Aberrant methylation and loss of expression of O-methylguanine-DNA
methyltransferase in pulmonary squamous cell carcinoma and adenocarcinoma.
Pathol Int 2005, 55(6):303-309.

61 Garnis C, Davies JJ, Buys TP, Tsao MS, MacAulay C, Lam S and Lam
WL: Chromosome 5p aberrations are early events in lung cancer: implication of
glial cell line-derived neurotrophic factor in disease progression. Oncogene 2005,
24(30):4806-4812.

62 Gavelli G and Giampalma E: Sensitivity and specificity of chest X-ray
screening for lung cancer: review article. Cancer 2000, 89(11 Suppl):2453-2456.

63 GeneCards: [http://www.genecards.org].

64 Gery S, Komatsu N, Kawamata N, Miller CW, Desmond J, Virk RK,
Marchevsky A, McKenna R, Taguchi H and Koeffler HP: Epigenetic silencing of the
candidate tumor suppressor gene Per1 in non-small cell lung cancer. Clin Cancer
Res 2007, 13(5):1399-1404.

65 Gessner C, Kuhn H, Toepfer K, Hammerschmidt S, Schauer J and Wirtz H:
Detection of p53 gene mutations in exhaled breath condensate of non-small cell lung
cancer patients. Lung Cancer 2004, 43(2):215-222.

66 Gohagan J, Marcus P, Fagerstrom R, Pinsky P, Kramer B and Prorok P:
Baseline findings of a randomized feasibility trial of lung cancer screening with
spiral CT scan vs chest radiograph: the Lung Screening Study of the National
Cancer Institute. Chest 2004, 126(1):114-121.

67 Gohagan JK, Marcus PM, Fagerstrom RM, Pinsky PF, Kramer BS, Prorok
PC, Ascher S, Bailey W, Brewer B, Church T, Engelhard D, Ford M, Fouad M,
Freedman M, Gelmann E, Gierada D, Hocking W, Inampudi S, Irons B, Johnson CC,
Jones A, Kucera G, Kvale P, Lappe K, Manor W, Moore A, Nath H, Neff S, Oken M,
Plunkett M, Price H, Reding D, Riley T, Schwartz M, Spizarny D, Yoffie R and Zylak C:
Final results of the Lung Screening Study, a randomized feasibility study of spiral
CT versus chest X-ray screening for lung cancer. Lung Cancer 2005, 47(1):9-15.

124
68 Grimm L, Holinski-Feder E, Teodoridis J, Scheffer B, Schindelhauer D,
Meitinger T and Ueffing M: Analysis of the human GDNF gene reveals an inducible
promoter, three exons, a triplet repeat within the 3'-UTR and alternative splice
products. Hum Mol Genet 1998, 7(12):1873-1886.

69 Grote HJ, Schmiemann V, Geddert H, Bocking A, Kappes R, Gabbert HE
and Sarbia M: Methylation of RAS association domain family protein 1A as a
biomarker of lung cancer. Cancer 2006, 108(2):129-134.

70 Grote HJ, Schmiemann V, Geddert H, Rohr UP, Kappes R, Gabbert HE and
Bocking A: Aberrant promoter methylation of p16(INK4a), RARB2 and SEMA3B
in bronchial aspirates from patients with suspected lung cancer. Int J Cancer 2005,
116(5):720-725.

71 Grote HJ, Schmiemann V, Kiel S, Bocking A, Kappes R, Gabbert HE and
Sarbia M: Aberrant methylation of the adenomatous polyposis coli promoter 1A in
bronchial aspirates from patients with suspected lung cancer. Int J Cancer 2004,
110(5):751-755.

72 Gu J, Berman D, Lu C, Wistuba, II, Roth JA, Frazier M, Spitz MR and Wu
X: Aberrant promoter methylation profile and association with survival in patients
with non-small cell lung cancer. Clin Cancer Res 2006, 12(24):7329-7338.

73 Guo M, Akiyama Y, House MG, Hooker CM, Heath E, Gabrielson E, Yang
SC, Han Y, Baylin SB, Herman JG and Brock MV: Hypermethylation of the GATA
genes in lung cancer. Clin Cancer Res 2004, 10(23):7917-7924.

74 Guo M, House MG, Hooker C, Han Y, Heath E, Gabrielson E, Yang SC,
Baylin SB, Herman JG and Brock MV: Promoter hypermethylation of resected
bronchial margins: a field defect of changes? Clin Cancer Res 2004, 10(15):5131-
5136.

75 Hagiwara N, Mechanic LE, Trivers GE, Cawley HL, Taga M, Bowman ED,
Kumamoto K, He P, Bernard M, Doja S, Miyashita M, Tajiri T, Sasajima K, Nomura T,
Makino H, Takahashi K, Hussain SP and Harris CC: Quantitative detection of p53
mutations in plasma DNA from tobacco smokers. Cancer Res 2006, 66(16):8309-
8317.

76 Hampton T: New Markers may help predict prostate cancer relapse
risk. Journal of the American Medical Association 2006, 295(19):2234-2238.

77 Hanabata T, Tsukuda K, Toyooka S, Yano M, Aoe M, Nagahiro I, Sano Y,
Date H and Shimizu N: DNA methylation of multiple genes and clinicopathological
relationship of non-small cell lung cancers. Oncol Rep 2004, 12(1):177-180.

125
78 Harden SV, Tokumaru Y, Westra WH, Goodman S, Ahrendt SA, Yang SC
and Sidransky D: Gene promoter hypermethylation in tumors and lymph nodes of
stage I lung cancer patients. Clin Cancer Res 2003, 9(4):1370-1375.

79 Hasegawa M, Sone S, Takashima S, Li F, Yang ZG, Maruyama Y and
Watanabe T: Growth rate of small lung cancers detected on mass CT screening. Br J
Radiol 2000, 73(876):1252-1259.

80 Hatada I, Fukasawa M, Kimura M, Morita S, Yamada K, Yoshikawa T,
Yamanaka S, Endo C, Sakurada A, Sato M, Kondo T, Horii A, Ushijima T and Sasaki H:
Genome-wide profiling of promoter methylation in human. Oncogene 2006,
25(21):3059-3064.

81 Haussinger K, Becker H, Stanzel F, Kreuzer A, Schmidt B, Strausz J,
Cavaliere S, Herth F, Kohlhaufl M, Muller KM, Huber RM, Pichlmeier U and Bolliger
Ch T: Autofluorescence bronchoscopy with white light bronchoscopy compared with
white light bronchoscopy alone for the detection of precancerous lesions: a
European randomised controlled multicentre trial. Thorax 2005, 60(6):496-503.

82 He B, You L, Uematsu K, Zang K, Xu Z, Lee AY, Costello JF, McCormick
F and Jablons DM: SOCS-3 is frequently silenced by hypermethylation and
suppresses cell growth in human lung cancer. Proc Natl Acad Sci U S A 2003,
100(24):14133-14138.

83 Henschke CI, Yankelevitz DF, Libby DM, Pasmantier MW, Smith JP and
Miettinen OS: Survival of patients with stage I lung cancer detected on CT screening.
N Engl J Med 2006, 355(17):1763-1771.

84 Hsu HS, Chen TP, Hung CH, Wen CK, Lin RK, Lee HC and Wang YC:
Characterization of a multiple epigenetic marker panel for lung cancer detection
and risk assessment in plasma. Cancer 2007, 110(9):2019-2026.

85 Hunt J: Exhaled breath condensate: an evolving tool for noninvasive
evaluation of lung disease. J Allergy Clin Immunol 2002, 110(1):28-34.

86 Ihaka R and Gentleman R: R: a language for data analysis and graphics.
J. Comput. Graph. Statist. 1996, 5:299-314.

87 Irimia M, Fraga MF, Sanchez-Cespedes M and Esteller M: CpG island
promoter hypermethylation of the Ras-effector gene NORE1A occurs in the context
of a wild-type K-ras in lung cancer. Oncogene 2004, 23(53):8695-8699.

126
88 Ito M, Ito G, Kondo M, Uchiyama M, Fukui T, Mori S, Yoshioka H, Ueda
Y, Shimokata K and Sekido Y: Frequent inactivation of RASSF1A, BLU, and
SEMA3B on 3p21.3 by promoter hypermethylation and allele loss in non-small cell
lung cancer. Cancer Lett 2005, 225(1):131-139.

89 Janssen-Heijnen ML and Coebergh JW: Trends in incidence and
prognosis of the histological subtypes of lung cancer in North America, Australia,
New Zealand and Europe. Lung Cancer 2001, 31(2-3):123-137.

90 Jarmalaite S, Kannio A, Anttila S, Lazutka JR and Husgafvel-Pursiainen K:
Aberrant p16 promoter methylation in smokers and former smokers with nonsmall
cell lung cancer. Int J Cancer 2003, 106(6):913-918.

91 Jemal A, Siegel R, Ward E, Murray T, Xu J and Thun MJ: Cancer
statistics, 2007. CA Cancer J Clin 2007, 57(1):43-66.

92 Jett JR: Limitations of screening for lung cancer with low-dose spiral
computed tomography. Clin Cancer Res 2005, 11(13 Pt 2):4988s-4992s.

93 Kaira K, Sunaga N, Tomizawa Y, Yanagitani N, Ishizuka T, Saito R,
Nakajima T and Mori M: Epigenetic inactivation of the RAS-effector gene RASSF2 in
lung cancers. Int J Oncol 2007, 31(1):169-173.

94 Kaneko M, Eguchi K, Ohmatsu H, Kakinuma R, Naruke T, Suemasu K and
Moriyama N: Peripheral lung cancer: screening and detection with low-dose spiral
CT versus radiography. Radiology 1996, 201(3):798-802.

95 Khan S, Coulson JM and Woll PJ: Genetic abnormalities in plasma DNA
of patients with lung cancer and other respiratory diseases. Int J Cancer 2004,
110(6):891-895.

96 Kikuchi S, Yamada D, Fukami T, Maruyama T, Ito A, Asamura H,
Matsuno Y, Onizuka M and Murakami Y: Hypermethylation of the TSLC1/IGSF4
promoter is associated with tobacco smoking and a poor prognosis in primary
nonsmall cell lung carcinoma. Cancer 2006, 106(8):1751-1758.

97 Kim DS, Cha SI, Lee JH, Lee YM, Choi JE, Kim MJ, Lim JS, Lee EB, Kim
CH, Park TI, Jung TH and Park JY: Aberrant DNA methylation profiles of non-small
cell lung cancers in a Korean population. Lung Cancer 2007, 58(1):1-6.

98 Kim H, Kwon YM, Kim JS, Lee H, Park JH, Shim YM, Han J, Park J and
Kim DH: Tumor-specific methylation in bronchial lavage for the early detection of
non-small-cell lung cancer. J Clin Oncol 2004, 22(12):2363-2370.

127
99 Kim JS, Han J, Shim YM, Park J and Kim DH: Aberrant methylation of
H-cadherin (CDH13) promoter is associated with tumor progression in primary
nonsmall cell lung carcinoma. Cancer 2005, 104(9):1825-1833.

100 Kim JS, Kim H, Shim YM, Han J, Park J and Kim DH: Aberrant
methylation of the FHIT gene in chronic smokers with early stage squamous cell
carcinoma of the lung. Carcinogenesis 2004, 25(11):2165-2171.

101 Kim YT, Lee SH, Sung SW and Kim JH: Can aberrant promoter
hypermethylation of CpG islands predict the clinical outcome of non-small cell lung
cancer after curative resection? Ann Thorac Surg 2005, 79(4):1180-1188; discussion
1180-1188.

102 Kramer B, Gohagan J and Prorok P: Cancer Screening: Theory and
Practice. In. Edited by B. Kramer: Marcel Dekker; 1999: 630.

103 Kuramochi M, Fukuhara H, Nobukuni T, Kanbe T, Maruyama T, Ghosh
HP, Pletcher M, Isomura M, Onizuka M, Kitamura T, Sekiya T, Reeves RH and
Murakami Y: TSLC1 is a tumor-suppressor gene in human non-small-cell lung
cancer. Nat Genet 2001, 27(4):427-430.

104 Laird PW and Jaenisch R: The role of DNA methylation in cancer genetic
and epigenetics. Annu Rev Genet 1996, 30:441-464.

105 Laird PW, Zijderveld A, Linders K, Rudnicki MA, Jaenisch R and Berns A:
Simplified mammalian DNA isolation procedure. Nucleic Acids Res 1991,
19(15):4293.

106 Lee MN, Tseng RC, Hsu HS, Chen JY, Tzao C, Ho WL and Wang YC:
Epigenetic inactivation of the chromosomal stability control genes BRCA1, BRCA2,
and XRCC5 in non-small cell lung cancer. Clin Cancer Res 2007, 13(3):832-838.

107 Li Q, Borovitskaya AE, DeSilva MG, Wasserfall C, Maclaren NK, Notkins
AL and Lan MS: Autoantigens in insulin-dependent diabetes mellitus: molecular
cloning and characterization of human IA-2 beta. Proc Assoc Am Physicians 1997,
109(4):429-439.

108 Li QL, Kim HR, Kim WJ, Choi JK, Lee YH, Kim HM, Li LS, Kim H,
Chang J, Ito Y, Youl Lee K and Bae SC: Transcriptional silencing of the RUNX3 gene
by CpG hypermethylation is associated with lung cancer. Biochem Biophys Res
Commun 2004, 314(1):223-228.

109 Li R, Todd NW, Qiu Q, Fan T, Zhao RY, Rodgers WH, Fang HB, Katz RL,
Stass SA and Jiang F: Genetic deletions in sputum as diagnostic markers for early
detection of stage I non-small cell lung cancer. Clin Cancer Res 2007, 13(2):482-487.
128

110 Lindell RM, Hartman TE, Swensen SJ, Jett JR, Midthun DE, Tazelaar HD
and Mandrekar JN: Five-year lung cancer screening experience: CT appearance,
growth rate, location, and histologic features of 61 lung cancers. Radiology 2007,
242(2):555-562.

111 Liu Y, An Q, Li L, Zhang D, Huang J, Feng X, Cheng S and Gao Y:
Hypermethylation of p16INK4a in Chinese lung cancer patients: biological and
clinical implications. Carcinogenesis 2003, 24(12):1897-1901.

112 Liu Y, Gao W, Siegfried JM, Weissfeld JL, Luketich JD and Keohavong P:
Promoter methylation of RASSF1A and DAPK and mutations of K-ras, p53, and
EGFR in lung tumors from smokers and never-smokers. BMC Cancer 2007, 7:74.

113 Liu Y, Lan Q, Siegfried JM, Luketich JD and Keohavong P: Aberrant
promoter methylation of p16 and MGMT genes in lung tumors from smoking and
never-smoking lung cancer patients. Neoplasia 2006, 8(1):46-51.

114 Luo RZ, Fang X, Marquez R, Liu SY, Mills GB, Liao WS, Yu Y and Bast
RC: ARHI is a Ras-related small G-protein with a novel N-terminal extension that
inhibits growth of ovarian and breast cancers. Oncogene 2003, 22(19):2897-2909.

115 Luxen S, Belinsky SA and Knaus UG: Silencing of DUOX NADPH
oxidases by promoter hypermethylation in lung cancer. Cancer Res 2008, 68(4):1037-
1045.

116 Maier S, Nimmrich I, Koenig T, Eppenberger-Castori S, Bohlmann I,
Paradiso A, Spyratos F, Thomssen C, Mueller V, Nahrig J, Schittulli F, Kates R, Lesche
R, Schwope I, Kluth A, Marx A, Martens JW, Foekens JA, Schmitt M and Harbeck N:
DNA-methylation of the homeodomain transcription factor PITX2 reliably predicts
risk of distant disease recurrence in tamoxifen-treated, node-negative breast cancer
patients--Technical and clinical validation in a multi-centre setting in collaboration
with the European Organisation for Research and Treatment of Cancer (EORTC)
PathoBiology group. Eur J Cancer 2007, 43(11):1679-1686.

117 Maneckjee R and Minna JD: Opioids induce while nicotine suppresses
apoptosis in human lung cancer cells. Cell Growth Differ 1994, 5(10):1033-1040.

118 Marsit CJ, Houseman EA, Christensen BC, Eddy K, Bueno R, Sugarbaker
DJ, Nelson HH, Karagas MR and Kelsey KT: Examination of a CpG island methylator
phenotype and implications of methylation profiles in solid tumors. Cancer Res 2006,
66(21):10621-10629.

129
119 Marsit CJ, Kim DH, Liu M, Hinds PW, Wiencke JK, Nelson HH and
Kelsey KT: Hypermethylation of RASSF1A and BLU tumor suppressor genes in
non-small cell lung cancer: implications for tobacco smoking during adolescence. Int
J Cancer 2005, 114(2):219-223.

120 McWilliams A, MacAulay C, Gazdar AF and Lam S: Innovative
molecular and imaging approaches for the detection of lung cancer and its
precursor lesions. Oncogene 2002, 21(45):6949-6959.

121 Miozzo M, Sozzi G, Musso K, Pilotti S, Incarbone M, Pastorino U and
Pierotti MA: Microsatellite alterations in bronchial and sputum specimens of lung
cancer patients. Cancer Res 1996, 56(10):2285-2288.

122 Mizuno K, Osada H, Konishi H, Tatematsu Y, Yatabe Y, Mitsudomi T,
Fujii Y and Takahashi T: Aberrant hypermethylation of the CHFR prophase
checkpoint gene in human lung cancers. Oncogene 2002, 21(15):2328-2333.

123 Molina JR, Yang P, Cassivi SD, Schild SE and Adjei AA: Non-small cell
lung cancer: epidemiology, risk factors, treatment, and survivorship. Mayo Clin Proc
2008, 83(5):584-594.

124 Mountain CF: The international system for staging lung cancer. Semin
Surg Oncol 2000, 18(2):106-115.

125 Nagatake M, Osada H, Kondo M, Uchida K, Nishio M, Shimokata K,
Takahashi T and Takahashi T: Aberrant hypermethylation at the bcl-2 locus at 18q21
in human lung cancers. Cancer Res 1996, 56(8):1886-1891.

126 Nakas CT and Alonzo TA: ROC graphs for assessing the ability of a
diagnostic marker to detect three disease classes with an umbrella ordering.
Biometrics 2007, 63(2):603-609.

127 Nakata S, Sugio K, Uramoto H, Oyama T, Hanagiri T, Morita M and
Yasumoto K: The methylation status and protein expression of CDH1, p16(INK4A),
and fragile histidine triad in nonsmall cell lung carcinoma: epigenetic silencing,
clinical features, and prognostic significance. Cancer 2006, 106(10):2190-2199.

128 Nishioka M, Kohno T, Tani M, Yanaihara N, Tomizawa Y, Otsuka A,
Sasaki S, Kobayashi K, Niki T, Maeshima A, Sekido Y, Minna JD, Sone S and Yokota J:
MYO18B, a candidate tumor suppressor gene at chromosome 22q12.1, deleted,
mutated, and methylated in human lung cancer. Proc Natl Acad Sci U S A 2002,
99(19):12269-12274.

130
129 Olaussen KA, Soria JC, Park YW, Kim HJ, Kim SH, Ro JY, Andre F and
Jang SJ: Assessing abnormal gene promoter methylation in paraffin-embedded
sputum from patients with NSCLC. Eur J Cancer 2005, 41(14):2112-2119.

130 Palmisano WA, Crume KP, Grimes MJ, Winters SA, Toyota M, Esteller M,
Joste N, Baylin SB and Belinsky SA: Aberrant promoter methylation of the
transcription factor genes PAX5 alpha and beta in human cancers. Cancer Res 2003,
63(15):4620-4625.

131 Palmisano WA, Divine KK, Saccomanno G, Gilliland FD, Baylin SB,
Herman JG and Belinsky SA: Predicting lung cancer by detecting aberrant promoter
methylation in sputum. Cancer Res 2000, 60(21):5954-5958.

132 Pepe MS, Etzioni R, Feng Z, Potter JD, Thompson ML, Thornquist M,
Winget M and Yasui Y: Phases of biomarker development for early detection of
cancer. J Natl Cancer Inst 2001, 93(14):1054-1061.

133 Raponi M, Zhang Y, Yu J, Chen G, Lee G, Taylor JM, Macdonald J,
Thomas D, Moskaluk C, Wang Y and Beer DG: Gene expression signatures for
predicting prognosis of squamous cell and adenocarcinomas of the lung. Cancer Res
2006, 66(15):7466-7472.

134 Rauch T, Li H, Wu X and Pfeifer GP: MIRA-Assisted Microarray
Analysis, a New Technology for the Determination of DNA Methylation Patterns,
Identifies Frequent Methylation of Homeodomain-Containing Genes in Lung
Cancer Cells. Cancer Res 2006, 66(16):7939-7947.

135 Rauch T, Wang Z, Zhang X, Zhong X, Wu X, Lau SK, Kernstine KH,
Riggs AD and Pfeifer GP: Homeobox gene methylation in lung cancer studied by
genome-wide analysis with a microarray-based methylated CpG island recovery
assay. Proc Natl Acad Sci U S A 2007, 104(13):5527-5532.

136 Rauch TA, Zhong X, Wu X, Wang M, Kernstine KH, Wang Z, Riggs AD
and Pfeifer GP: High-resolution mapping of DNA hypermethylation and
hypomethylation in lung cancer. Proc Natl Acad Sci U S A 2008, 105(1):252-257.

137 Rideout WM, 3rd, Eversole-Cire P, Spruck CH, 3rd, Hustad CM, Coetzee
GA, Gonzales FA and Jones PA: Progressive increases in the methylation status and
heterochromatinization of the myoD CpG island during oncogenic transformation.
Mol Cell Biol 1994, 14(9):6143-6152.

138 Romero RL, Juston AC, Ballantyne J and Henry BE: The applicability of
formalin-fixed and formalin fixed paraffin embedded tissues in forensic DNA
analysis. J Forensic Sci 1997, 42(4):708-714.

131
139 Safar AM, Spencer H, 3rd, Su X, Coffey M, Cooney CA, Ratnasinghe LD,
Hutchins LF and Fan CY: Methylation profiling of archived non-small cell lung
cancer: a promising prognostic system. Clin Cancer Res 2005, 11(12):4400-4405.

140 Sano A, Kage H, Sugimoto K, Kitagawa H, Aki N, Goto A, Fukayama M,
Nakajima J, Takamoto S, Nagase T, Yatomi Y, Ohishi N and Takai D: A second-
generation profiling system for quantitative methylation analysis of multiple gene
promoters: application to lung cancer. Oncogene 2007, 26(45):6518-6525.

141 Sathyanarayana UG, Toyooka S, Padar A, Takahashi T, Brambilla E, Minna
JD and Gazdar AF: Epigenetic inactivation of laminin-5-encoding genes in lung
cancers. Clin Cancer Res 2003, 9(7):2665-2672.

142 Sato M, Mori Y, Sakurada A, Fujimura S and Horii A: The H-cadherin
(CDH13) gene is inactivated in human lung cancer. Hum Genet 1998, 103(1):96-101.

143 Schmiemann V, Bocking A, Kazimirek M, Onofre AS, Gabbert HE, Kappes
R, Gerharz CD and Grote HJ: Methylation assay for the diagnosis of lung cancer on
bronchial aspirates: a cohort study. Clin Cancer Res 2005, 11(21):7728-7734.

144 Cancer of the Lung and Bronchus
[http://seer.cancer.gov/statfacts/html/lungb.html]

145 Shames DS, Girard L, Gao B, Sato M, Lewis CM, Shivapurkar N, Jiang A,
Perou CM, Kim YH, Pollack JR, Fong KM, Lam CL, Wong M, Shyr Y, Nanda R,
Olopade OI, Gerald W, Euhus DM, Shay JW, Gazdar AF and Minna JD: A genome-wide
screen for promoter methylation in lung cancer identifies novel methylation
markers for multiple malignancies. PLoS Med 2006, 3(12):e486.

146 Shapiro B, Chakrabarty M, Cohn EM and Leon SA: Determination of
circulating DNA levels in patients with benign or malignant gastrointestinal disease.
Cancer 1983, 51(11):2116-2120.

147 Shigematsu H, Suzuki M, Takahashi T, Miyajima K, Toyooka S,
Shivapurkar N, Tomlinson GE, Mastrangelo D, Pass HI, Brambilla E, Sathyanarayana
UG, Czerniak B, Fujisawa T, Shimizu N and Gazdar AF: Aberrant methylation of
HIN-1 (high in normal-1) is a frequent event in many human malignancies. Int J
Cancer 2005, 113(4):600-604.

148 Shivapurkar N, Stastny V, Suzuki M, Wistuba, II, Li L, Zheng Y, Feng Z,
Hol B, Prinsen C, Thunnissen FB and Gazdar AF: Application of a methylation gene
panel by quantitative PCR for lung cancers. Cancer Lett 2007, 247(1):56-71.

132
149 Shivapurkar N, Stastny V, Xie Y, Prinsen C, Frenkel E, Czerniak B,
Thunnissen FB, Minna JD and Gazdar AF: Differential methylation of a short CpG-
rich sequence within exon 1 of TCF21 gene: a promising cancer biomarker assay.
Cancer Epidemiol Biomarkers Prev 2008, 17(4):995-1000.

150 Shivapurkar N, Toyooka S, Toyooka KO, Reddy J, Miyajima K, Suzuki M,
Shigematsu H, Takahashi T, Parikh G, Pass HI, Chaudhary PM and Gazdar AF:
Aberrant methylation of trail decoy receptor genes is frequent in multiple tumor
types. Int J Cancer 2004, 109(5):786-792.

151 Smith LT, Lin M, Brena RM, Lang JC, Schuller DE, Otterson GA,
Morrison CD, Smiraglia DJ and Plass C: Epigenetic regulation of the tumor
suppressor gene TCF21 on 6q23-q24 in lung and head and neck cancer. Proc Natl
Acad Sci U S A 2006, 103(4):982-987.

152 Sone S, Nakayama T, Honda T, Tsushima K, Li F, Haniuda M, Takahashi
Y, Hanaoka T, Takayama F, Koizumi T, Kubo K, Yamanda T, Kondo R, Fushimi H and
Suzuki T: CT findings of early-stage small cell lung cancer in a low-dose CT
screening programme. Lung Cancer 2007, 56(2):207-215.

153 Sozzi G, Musso K, Ratcliffe C, Goldstraw P, Pierotti MA and Pastorino U:
Detection of microsatellite alterations in plasma DNA of non-small cell lung cancer
patients: a prospect for early diagnosis. Clin Cancer Res 1999, 5(10):2689-2692.

154 Sun S, Schiller JH and Gazdar AF: Lung cancer in never smokers--a
different disease. Nat Rev Cancer 2007, 7(10):778-790.

155 Suzuki M, Shigematsu H, Iizasa T, Hiroshima K, Nakatani Y, Minna JD,
Gazdar AF and Fujisawa T: Exclusive mutation in epidermal growth factor receptor
gene, HER-2, and KRAS, and synchronous methylation of nonsmall cell lung
cancer. Cancer 2006, 106(10):2200-2207.

156 Suzuki M, Shigematsu H, Nakajima T, Kubo R, Motohashi S, Sekine Y,
Shibuya K, Iizasa T, Hiroshima K, Nakatani Y, Gazdar AF and Fujisawa T:
Synchronous alterations of Wnt and epidermal growth factor receptor signaling
pathways through aberrant methylation and mutation in non small cell lung cancer.
Clin Cancer Res 2007, 13(20):6087-6092.

157 Suzuki M, Shigematsu H, Shames DS, Sunaga N, Takahashi T, Shivapurkar
N, Iizasa T, Frenkel EP, Minna JD, Fujisawa T and Gazdar AF: DNA methylation-
associated inactivation of TGFbeta-related genes DRM/Gremlin, RUNX3, and
HPP1 in human cancers. Br J Cancer 2005, 93(9):1029-1037.

158 Suzuki MM and Bird A: DNA methylation landscapes: provocative
insights from epigenomics. Nat Rev Genet 2008, 9(6):465-476.
133

159 Tai KY, Shiah SG, Shieh YS, Kao YR, Chi CY, Huang E, Lee HS, Chang
LC, Yang PC and Wu CW: DNA methylation and histone modification regulate
silencing of epithelial cell adhesion molecule for tumor invasion and progression.
Oncogene 2007, 26(27):3989-3997.

160 Takai D and Jones PA: The CpG island searcher: a new WWW
resource. In Silico Biol 2003, 3(3):235-240.

161 Takai D, Yagi Y, Wakazono K, Ohishi N, Morita Y, Sugimura T and
Ushijima T: Silencing of HTR1B and reduced expression of EDN1 in human lung
cancers, revealed by methylation-sensitive representational difference analysis.
Oncogene 2001, 20(51):7505-7513.

162 Tam IY, Chung LP, Suen WS, Wang E, Wong MC, Ho KK, Lam WK, Chiu
SW, Girard L, Minna JD, Gazdar AF and Wong MP: Distinct epidermal growth factor
receptor and KRAS mutation patterns in non-small cell lung cancer patients with
different tobacco exposure and clinicopathologic features. Clin Cancer Res 2006,
12(5):1647-1653.

163 Tang M, Torres-Lanzas J, Lopez-Rios F, Esteller M and Sanchez-Cespedes
M: Wnt signaling promoter hypermethylation distinguishes lung primary
adenocarcinomas from colorectal metastasis to the lung. Int J Cancer 2006,
119(11):2603-2606.

164 Tessema M, Willink R, Do K, Yu YY, Yu W, Machida EO, Brock M, Van
Neste L, Stidley CA, Baylin SB and Belinsky SA: Promoter methylation of genes in
and around the candidate lung cancer susceptibility locus 6q23-25. Cancer Res 2008,
68(6):1707-1714.

165 Topaloglu O, Hoque MO, Tokumaru Y, Lee J, Ratovitski E, Sidransky D
and Moon CS: Detection of promoter hypermethylation of multiple genes in the
tumor and bronchoalveolar lavage of patients with lung cancer. Clin Cancer Res
2004, 10(7):2284-2288.

166 Toyooka KO, Toyooka S, Virmani AK, Sathyanarayana UG, Euhus DM,
Gilcrease M, Minna JD and Gazdar AF: Loss of expression and aberrant methylation
of the CDH13 (H-cadherin) gene in breast and lung carcinomas. Cancer Res 2001,
61(11):4556-4560.

167 Toyooka S, Toyooka KO, Maruyama R, Virmani AK, Girard L, Miyajima
K, Harada K, Ariyoshi Y, Takahashi T, Sugio K, Brambilla E, Gilcrease M, Minna JD
and Gazdar AF: DNA methylation profiles of lung tumors. Mol Cancer Ther 2001,
1(1):61-67.

134
168 Toyooka S, Toyooka KO, Miyajima K, Reddy JL, Toyota M,
Sathyanarayana UG, Padar A, Tockman MS, Lam S, Shivapurkar N and Gazdar AF:
Epigenetic down-regulation of death-associated protein kinase in lung cancers. Clin
Cancer Res 2003, 9(8):3034-3041.

169 Toyota M, Kopecky KJ, Toyota MO, Jair KW, Willman CL and Issa JP:
Methylation profiling in acute myeloid leukemia. Blood 2001, 97(9):2823-2829.

170 Tsou JA, Galler JS, Siegmund KD, Laird PW, Turla S, Cozen W, Hagen
JA, Koss MN and Laird-Offringa IA: Identification of a panel of sensitive and specific
DNA methylation markers for lung adenocarcinoma. Mol Cancer 2007, 6:70.

171 Tsou JA, Shen LY, Siegmund KD, Long TI, Laird PW, Seneviratne CK,
Koss MN, Pass HI, Hagen JA and Laird-Offringa IA: Distinct DNA methylation
profiles in malignant mesothelioma, lung adenocarcinoma, and non-tumor lung.
Lung Cancer 2005, 47(2):193-204.

172 Tzao C, Tsai HY, Chen JT, Chen CY and Wang YC: 5'CpG island
hypermethylation and aberrant transcript splicing both contribute to the
inactivation of the FHIT gene in resected non-small cell lung cancer. Eur J Cancer
2004, 40(14):2175-2183.

173 Ulivi P, Zoli W, Calistri D, Fabbri F, Tesei A, Rosetti M, Mengozzi M and
Amadori D: p16INK4A and CDH13 hypermethylation in tumor and serum of non-
small cell lung cancer patients. J Cell Physiol 2006, 206(3):611-615.

174 Usadel H, Brabender J, Danenberg KD, Jeronimo C, Harden S, Engles J,
Danenberg PV, Yang S and Sidransky D: Quantitative adenomatous polyposis coli
promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients
with lung cancer. Cancer Res 2002, 62(2):371-375.

175 van Klaveren RJ, Habbema JDF, Pedersen JH, de Koning HJ, Oudkerk M
and Hoogsteden HC: Lung cancer screening by low-dose spiral computed
tomography. Eur Respir J 2001, 18(5):857-866.

176 Virmani AK, Rathi A, Sathyanarayana UG, Padar A, Huang CX,
Cunnigham HT, Farinas AJ, Milchgrub S, Euhus DM, Gilcrease M, Herman J, Minna JD
and Gazdar AF: Aberrant methylation of the adenomatous polyposis coli (APC) gene
promoter 1A in breast and lung carcinomas. Clin Cancer Res 2001, 7(7):1998-2004.

177 Virmani AK, Tsou JA, Siegmund KD, Shen LY, Long TI, Laird PW,
Gazdar AF and Laird-Offringa IA: Hierarchical clustering of lung cancer cell lines
using DNA methylation markers. Cancer Epidemiol Biomarkers Prev 2002, 11(3):291-
297.

135
178 Vischioni B, Oudejans JJ, Vos W, Rodriguez JA and Giaccone G:
Frequent overexpression of aurora B kinase, a novel drug target, in non-small cell
lung carcinoma patients. Mol Cancer Ther 2006, 5(11):2905-2913.

179 Wang G, Hu X, Lu C, Su C, Luo S and Luo ZW: Promoter-
hypermethylation associated defective expression of E-cadherin in primary non-
small cell lung cancer. Lung Cancer 2008.

180 Wang J, Walsh G, Liu DD, Lee JJ and Mao L: Expression of Delta
DNMT3B variants and its association with promoter methylation of p16 and
RASSF1A in primary non-small cell lung cancer. Cancer Res 2006, 66(17):8361-
8366.

181 Wang Y, Yu Z, Wang T, Zhang J, Hong L and Chen L: Identification of
epigenetic aberrant promoter methylation of RASSF1A in serum DNA and its
clinicopathological significance in lung cancer. Lung Cancer 2007, 56(2):289-294.

182 Wang Y, Zhang D, Zheng W, Luo J, Bai Y and Lu Z: Multiple gene
methylation of nonsmall cell lung cancers evaluated with 3-dimensional microarray.
Cancer 2008, 112(6):1325-1336.

183 Wang YC, Lu YP, Tseng RC, Lin RK, Chang JW, Chen JT, Shih CM and
Chen CY: Inactivation of hMLH1 and hMSH2 by promoter methylation in primary
non-small cell lung tumors and matched sputum samples. J Clin Invest 2003,
111(6):887-895.

184 Weisenberger DJ, Campan M, Long TI, Kim M, Woods C, Fiala E, Ehrlich
M and Laird PW: Analysis of repetitive element DNA methylation by MethyLight.
Nucleic Acids Res 2005, 33(21):6823-6836.

185 Weisenberger DJ, Siegmund KD, Campan M, Young J, Long TI, Faasse
MA, Kang GH, Widschwendter M, Weener D, Buchanan D, Koh H, Simms L, Barker M,
Leggett B, Levine J, Kim M, French AJ, Thibodeau SN, Jass J, Haile R and Laird PW:
CpG island methylator phenotype underlies sporadic microsatellite instability and is
tightly associated with BRAF mutation in colorectal cancer. Nat Genet 2006,
38(7):787-793.

186 Weissleder R and Pittet MJ: Imaging in the era of molecular oncology.
Nature 2008, 452(7187):580-589.

187 Widschwendter M, Fiegl H, Egle D, Mueller-Holzner E, Spizzo G, Marth
C, Weisenberger DJ, Campan M, Young J, Jacobs I and Laird PW: Epigenetic stem cell
signature in cancer. Nat Genet 2007, 39(2):157-158.

136
188 Widschwendter M and Menon U: Circulating methylated DNA: a new
generation of tumor markers. Clin Cancer Res 2006, 12(24):7205-7208.

189 Xie GS, Hou AR, Li LY, Gao YN and Cheng SJ: Aberrant p16 promoter
hypermethylation in bronchial mucosae as a biomarker for the early detection of
lung cancer. Chin Med J (Engl) 2006, 119(17):1469-1472.

190 Xu L and Jain RK: Down-regulation of placenta growth factor by
promoter hypermethylation in human lung and colon carcinoma. Mol Cancer Res
2007, 5(9):873-880.

191 Xu XL, Wu LC, Du F, Davis A, Peyton M, Tomizawa Y, Maitra A,
Tomlinson G, Gazdar AF, Weissman BE, Bowcock AM, Baer R and Minna JD:
Inactivation of human SRBC, located within the 11p15.5-p15.4 tumor suppressor
region, in breast and lung cancers. Cancer Res 2001, 61(21):7943-7949.

192 Yanagawa N, Tamura G, Oizumi H, Takahashi N, Shimazaki Y and
Motoyama T: Promoter hypermethylation of tumor suppressor and tumor-related
genes in non-small cell lung cancers. Cancer Sci 2003, 94(7):589-592.

193 Yankelevitz DF, Kostis WJ, Henschke CI, Heelan RT, Libby DM,
Pasmantier MW and Smith JP: Overdiagnosis in chest radiographic screening for lung
carcinoma: frequency. Cancer 2003, 97(5):1271-1275.

194 Yano M, Toyooka S, Tsukuda K, Dote H, Ouchida M, Hanabata T, Aoe M,
Date H, Gazdar AF and Shimizu N: Aberrant promoter methylation of human DAB2
interactive protein (hDAB2IP) gene in lung cancers. Int J Cancer 2005, 113(1):59-66.

195 Yue W, Dacic S, Sun Q, Landreneau R, Guo M, Zhou W, Siegfried JM, Yu
J and Zhang L: Frequent inactivation of RAMP2, EFEMP1 and Dutt1 in lung cancer
by promoter hypermethylation. Clin Cancer Res 2007, 13(15 Pt 1):4336-4344.

196 Zhang P, Wang J, Gao W, Yuan BZ, Rogers J and Reed E: CHK2 kinase
expression is down-regulated due to promoter methylation in non-small cell lung
cancer. Mol Cancer 2004, 3:14.

197 Zhong S, Fields CR, Su N, Pan YX and Robertson KD: Pharmacologic
inhibition of epigenetic modifications, coupled with gene expression profiling,
reveals novel targets of aberrant DNA methylation and histone deacetylation in lung
cancer. Oncogene 2007, 26(18):2621-2634.

137
198 Zhou W, Heist RS, Liu G, Neuberg DS, Asomaning K, Su L, Wain JC,
Lynch TJ, Giovannucci E and Christiani DC: Polymorphisms of vitamin D receptor
and survival in early-stage non-small cell lung cancer patients. Cancer Epidemiol
Biomarkers Prev 2006, 15(11):2239-2245.

199 Zochbauer-Muller S, Fong KM, Virmani AK, Geradts J, Gazdar AF and
Minna JD: Aberrant promoter methylation of multiple genes in non-small cell lung
cancers. Cancer Res 2001, 61(1):249-255.

138
Appendix 1

Supplemental Table 1

Gene
Name
Reacti
on ID
Forward primer
sequence
Reverse primer
sequence
Probe sequence (5' with 6FAM
and 3' with BHQ1)
ABCB
1
HB-
051
TCGGGTCGGGAGTAG
TTATTTG
CGACTATACTCAAC
CCACGCC
6FAM-
ACGCTATTCCTACCCAACCAA
TCAACCTCA-BHQ-1
CACN
A1G
HB-
162
GTCGTCGGCGTTATTT
TAGAAAGTT
CACCGACGCCCAAC
ACA
6FAM-
ACGCTCCGCTCCCGAATACCC
A-BHQ-1
CCND
2
HB-
040
GGAGGGTCGGCGAGG
AT
TCCTTTCCCCGAAA
ACATAAAA
6FAM-
CACGCTCGATCCTTCGCCCG-
BHQ-1
CDX1
HB-
195
TGAGCGGTTGTTCGTC
GTC
AAATCCCCCGCGCA
TACTA
6FAM-
CCTAAAACCGCCGCTACCGA
CCG-BHQ-1
CPVL HB
427
ATTTACGTAGGGTAG
GCGGTATTTAC
AACGCTACAAAAAC
AACCGACTAA
6FAM-
ATCCTTACGCGCCGCGACTCA
A-BHQ-1
CYP1
B1
HB
238
GGGTTAGCGGTTGTT
GAGGTAAC
CCGCGTTTCCGATC
AATAA
6FAM-
AAAACGCGAACCGAAACTAC
CGCCT-BHQ-1
DIRAS
3
HB-
043
GCGTAAGCGGAATTT
ATGTTTGT
CCGCGATTTTATATT
CCGACTT
6FAM-
CGCACAAAAACGAAATACGA
AAACGCAAA-BHQ-1
DLEC HB
225
TCGTTGCGTATTTAAG
ATATTTCGTATT
CGTAACGCTCATTC
TCGCTACC
6FAM-
TAATCAAACTTACGCTCACTT
CGTCGCCG-BHQ-1
GATM HB
401
TTTTTGTAGTCGCGTT
TCGTTTC
GCCGACCCCACCAC
CTAT
6FAM-
CGACAACCAATAAAACCGCG
AAAAACGA-BHQ-1
GDNF HB
222
TCGTTTGTTCGCGTAG
GTGTC
CGATATAAAACAAC
ACCAAACAAACAAC
6FAM-
TCCCATAACTTCATCTTAAAA
TCCCGTCCG-BHQ-1
GP1B
B
HB-
398
TGGGAGCGGAAGTTT
GAGC
AACGCGCGCTACAA
CGAC
6FAM-
AATAACACACTATCGCCGAA
AACCGCAAA-BHQ-1
GRIN2
B
HB
250
GTCGGATTTACGCGT
CGAGT
CTACCGCCGCGCTA
AAATAC
6FAM-
ACGCACGAAACTTCACCTAC
AACGTATCG-BHQ-1
HIC1
HB-
168
GTTAGGCGGTTAGGG
CGTC
CCGAACGCCTCCAT
CGTAT
6FAM-
CAACATCGTCTACCCAACAC
ACTCTCCTACG-BHQ-1
HOXC
9
HB
440
GGGAGTTGCGCGATC
G
CTTCTCCTCTTTATA
CTTACTACCGACA
6FAM-
CGCGTCCGCCTCGAACGAAA
AC-BHQ-1
RASSF
1
HB-
144
GAGCGATGACGGAAT
ATAAGTTGG
CGTCCACAAAATAA
TTCTAAATCAACTA
6FAM-
CACTCTTACCCACACCGCCGA
CG-BHQ-1
LTB4R
HB-
070
GCGTTGGTTTTATCGG
AAGG
AAACCGTAATTCCC
GCTCG
6FAM-
GACTCCGCCCAACTTCGCCAA
AA-BHQ-1
MGM
T
HB-
159
CTAACGTATAACGAA
AATCGTAACAACC
AGTATGAAGGGTAG
GAAGAATTCGG
6FAM-
CCTTACCTCTAAATACCAACC
CCAAACCCG-BHQ-1
MINT1
HB-
161
GGGTTGAGGTTTTTTG
TTAGCG
CCCCTCTAAACTTC
ACAACCTCG
6FAM-
CTACTTCGCCTAACCTAACGC
139
ACAACAAACG-BHQ-1

MT1G
HB-
204
CGTTTAAGGGATTTTG
TATTTGGTTTAT
CCGCTAAATCCGCA
CCG
6FAM-
CGCGATCCCGACCTAAACTAT
ACGCA-BHQ-1
MTHF
R
HB-
058
TGGTAGTGAGAGTTT
TAAAGATAGTTCGA
CGCCTCATCTTCTCC
CGA
6FAM-
TCTCATACCGCTCAAAATCCA
AACCCG-BHQ-1
NEUR
OD1
HB
259
GTTTTTTGCGTGGGCG
AAT
CCGCGCTTAACATC
ACTAACTAAA
6FAM-
CGCGCGACCACGACACGAAA
-BHQ-1
NEUR
OG1
HB
261
CGTGTAGCGTTCGGG
TATTTGTA
CGATAATTACGAAC
ACACTCCGAAT
6FAM-
CGATAACGACCTCCCGCGAA
CATAAA-BHQ-1
ONEC
UT2
HB
243
CGTTACGTATATCGC
GCGG
CAAAAACCTCCTXX
XATAAACGACGAAT
6FAM-
AACACGCAATTACGCGCTTTT
ATACGCA-BHQ-1
OPCM
L
HB-
209
CGTTTCGAGGCGGTA
TCG
CGAACCGCCGAAAT
TATCAT
6FAM-
AACAACTCCATCCCTAACCGC
CACTTTCT-BHQ-1
PAX-8
HB-
210
ATCGATCGGTTTTATT
TCGTTGAG
ACCAATCCGCGACC
TACG
6FAM-
ACCTCGCCAAACCCATCTCCC
AAAA-BHQ-1
PENK
HB-
163
GGTTAATTATAAAGT
GGTTTTAGTAGTCGG
CAACGTCTCTACGA
AATCACGAAC
6FAM-
AACGCCTACCTCGCCGTCCCG
-BHQ-1
PITX2 HB
234
GGAGTGACGTGACGT
TAGTAGAGATTT
AACCGCGCAACCGA
ACT
6FAM-
CGCCCGCGCGCCACTATACA-
BHQ-1
PLAG
L1
HB-
199
ATCGACGGGTTGAAT
GATAAATG
CTCGACGCAACCAT
CCTCTT
6FAM-
ACTACCGCGAACGACAAAAC
CCACG-BHQ-1
PTPR
N2
HB
392
MGTTUUAAUAGUTTM
GGGTUUAGTUAUAT
AACTRCKCTTTCTCA
RCKCCTC
6FAM-
TAAAACGACCGCGTACTCGC
CAAAAAA-BHQ-1
RARR
ES1
HB
322
GGCGAGTCGGATCGG
AA
CGCAAACTCCTACA
ACAAACGA
6FAM-
CGCGCGACGCTTCACTTCTTC
AA-BHQ-1
RNR1
HB-
071
CGTTTTGGAGATACG
GGTCG
AAACAACGCCGAAC
CGAA
6FAM-
ACCGCCCGTACCACACGCAA
A-BHQ-1
RPA 3 HB
104
AGCGCGATTGCGATT
TAGG
TTTCTCGACACCAA
TCAACGAA
6FAM-
TCCAACTTCGCCAATTAAATA
CGCGAAA-BHQ-1
SEZ6L
HB-
184
GCGTTAGTAGGGAGA
GAAAACGTTC
ATACCAACCGCCTC
CTCTAACC
6FAM-
CCGTCGACCCTACAAAATTTA
ACGCCA-BHQ-1
SFRP1
HB-
201
GAATTCGTTCGCGAG
GGA
AAACGAACCGCACT
CGTTACC
6FAM-
CCGTCACCGACGCGAAAACC
AAT-BHQ-1
SFRP2 HB
279
TTTATAATTTTGATTT
TTTTACGGTATTGG
GAAACCGCCTCGAC
GAACT
6FAM-
CTCGAATCTCCAACCACCGTT
CAACAA-BHQ-1
SLC38
A4
HB
430
GATTTGAGGACGCGG
GC
TTCCCCCGCGAAAA
CTAACT
6FAM-
CGCGACCGCCCGAAATCCTA
CT-BHQ-1
TCF21 HB
359
GAGAGTTTTAATTGC
GAGAATGGG
TCTTCTTAATAAAC
GCCTTCCTCC
6FAM-
CCAAACCGCCGCGACCCTTCT
-BHQ-1
TFAP2
HB CGTTAATTTTTAAAGT CCGACAACCAACAC 6FAM-
140
A 314 ATTTTTATGGATCG TTTACGC CGAAACCGAAAAAAACATAT
CCGTTCACG-BHQ-1
TMEF
F2
HB-
274
CGACGAGGAGGTGTA
AGGATG
CAACGCCTAACGAA
CGAACC
6FAM-
TATAACTTCCGCGACCGCCTC
CTCCT-BHQ-1
TNFR
SF25
HB-
080
GCGGAATTACGACGG
GTAGA
ACTCCATAACCCTC
CGACGA
6FAM-
CGCCCAAAAACTTCCCGACTC
CGTA-BHQ-1
TWIST
HB-
047
GTAGCGCGGCGAACG
T
AAACGCAACGAATC
ATAACCAAC
6FAM-
CCAACGCACCCAATCGCTAA
ACGA-BHQ-1
WDR3
3
HB
435
GTTATTACGTATTGGC
GGGACG
ACGCAAATCGAACC
TCACAAA
6FAM-
CCGCGATCCAAACGCGCG-
BHQ-1

141

Supplemental Table 2

Gene
Name Reaction ID
Forward primer
sequence Reverse primer sequence
Probe sequence (5' with 6FAM
and 3' with BHQ1)
GDNF HB 222
TCGTTTGTTCGCGT
AGGTGTC
CGATATAAAACAACA
CCAAACAAACAAC
6FAM-
TCCCATAACTTCATCTTAAAAT
CCCGTCCG-BHQ-1
HOXA
9 HB-516
TTATTAAGAAGGC
GCGGGTTTC
CAATAAAACGACAAT
ACGATTTAACTACTTT
TT
6FAM-
TAAAAAACTCCGCTAACCGCA
CTCGCAC-BHQ-1
MOS HB-515
CGTTTATCGATGG
GGAAAATTC
CTAATCTCTTCATTCA
CTCCAACGA
6FAM-
CCCAACAAAATACGATACCCT
CGCCCCTA-BHQ-1
MT1G HB-204
CGTTTAAGGGATTT
TGTATTTGGTTTAT
CCGCTAAATCCGCAC
CG
6FAM-
CGCGATCCCGACCTAAACTAT
ACGCA-BHQ-1
MTHF
R
f
HB-058
TGGTAGTGAGAGT
TTTAAAGATAGTTC
GA
CGCCTCATCTTCTCCC
GA
6FAM-
TCTCATACCGCTCAAAATCCAA
ACCCG-BHQ-1
OPCM
L
g
HB-209
CGTTTCGAGGCGG
TATCG
CGAACCGCCGAAATT
ATCAT
6FAM-
AACAACTCCATCCCTAACCGC
CACTTTCT-BHQ-1
PITX2 HB-234
GGAGTGACGTGAC
GTTAGTAGAGATT
T
AACCGCGCAACCGAA
CT
6FAM-
CGCCCGCGCGCCACTATACA-
BHQ-1
PTPRN
2 HB-392
MGTTUUAAUAGUT
TMGGGTUUAGTUA
UAAGT
AACTRCKCTTTCTCAR
CKCCTC
6FAM-
TAAAACGACCGCGTACTCGCC
AAAAAA-BHQ-1
TCF21 HB 359
GAGAGTTTTAATT
GCGAGAATGGG
TCTTCTTAATAAACGC
CTTCCTCC
6FAM-
CCAAACCGCCGCGACCCTTCT-
BHQ-1
TNFRS
F25 HB-080
GCGGAATTACGAC
GGGTAGA
ACTCCATAACCCTCCG
ACGA
6FAM-
CGCCCAAAAACTTCCCGACTC
CGTA-BHQ-1

Abstract (if available)

Abstract Lung cancer is the number one cancer killer in the United States. This disease is divided into two sub-types, small cell lung cancer, (10-15% of lung cancer cases), and non-small cell lung cancer (NSCLC

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Development of DNA methylation biomarkers as an early detection tool for human lung adenocarcinoma

PDF

DNA methylation markers for blood-based detection of small cell lung cancer in mouse models

PDF

CpG methylation profiling in lung cancer cell lines, tumors and non-tumors

PDF

DNA methylation changes in the development of lung adenocarcinoma

PDF

The role of DNA methylation in early detection and progression of pancreatic cancer

PDF

DNA methylation as a biomarker in human reproductive health and disease

PDF

Identification of DNA methylation markers in diffuse large B-cell lymphoma

PDF

The expression of human carboxylesterases in normal tissues and cancer cell-lines

PDF

Use of DNA methylation inhibitors for chemotherapy and chemoprevention of cancer

PDF

Modeling lung adenocarcinoma progression in vitro using immortalized human alveolar epithelial cells

PDF

Development of lung cancer specific DNA methylation markers

PDF

Developing a robust single cell whole genome bisulfite sequencing protocol to analyse circulating tumor cells

PDF

Limit of detection analysis for cell-free DNA methylation using targeted bisulfite sequencing

PDF

DNA hypermethylation: its role in colorectal tumorigenesis and potential clinical applications

PDF

Early detection of lung cancer by characterizing circulating rare cells using peripheral blood liquid biopsy

PDF

Anti-Hu immuno-responsivenss in small-cell lung cancer

PDF

Development of immunotherapy for small cell lung cancer using iso-aspartylated antigen

PDF

Development of immunotherapy for small cell lung cancer using novel modified antigens

PDF

Characterization and functional study of a novel human protein SFMBT, and PR-Set7 histone methyltransferase

PDF

LINC00261 induces a G2/M cell cycle arrest and activation of the DNA damage response in lung adenocarcinoma

Asset Metadata

Creator Anglim, Paul P. (author)

Core Title Development of DNA methylation based biomarkers for the early detection of squamous cell lung cancer

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Biochemistry and Molecular Biology

Degree Conferral Date 2008-12

Publication Date 12/08/2008

Defense Date 07/24/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag DNA methylation,early detection,lung cancer,OAI-PMH Harvest,squamous

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Laird-Offringa, Ite A. (committee chair), Johnson, Deborah L. (committee member), Rice, Judd C. (committee member), Siegmund, Kimberly D. (committee member)

Creator Email anglim@usc.edu,paulanglim@verizon.net

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1889

Unique identifier UC1444820

Identifier etd-Anglim-2548 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-142191 (legacy record id),usctheses-m1889 (legacy record id)

Legacy Identifier etd-Anglim-2548.pdf

Dmrecord 142191

Document Type Dissertation

Rights Anglim, Paul P.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu