Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Construction of a surgical survival prediction model of stage IV NSCLC patients-based on seer database
(USC Thesis Other)
Construction of a surgical survival prediction model of stage IV NSCLC patients-based on seer database
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
CONSTRUCTION OF A SURGICAL SURVIV AL PREDICTION MODEL OF STAGE IV
NSCLC PATIENTS-BASED ON SEER DATABASE
by
Chen Liang
A Thesis Presented to the
FACULTY OF THE USC KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS AND EPIDEMIOLOGY)
May 2023
Copyright 2023 Chen Liang
ii
Dedication
I would like to dedicate this paper to my advisor, thanking him for the guidance and support
on my academic journey.
Throughout my graduate studies, my advisor has provided me with selfless help and guidance,
and his expertise and rigorous attitude have had a profound impact on my academic growth. He
not only provided invaluable advice and guidance on the choice of research direction, but also
recommended data resources, suggested methodologies and relevant literature, and provided
insightful comments and feedback on the entire thesis. His unwavering support, patience, and
expertise have been indispensable throughout this process.
I would like to express my deepest gratitude to my advisor for his support and guidance. The
success of this paper would not have been possible without his support and guidance, and he has
been a great mentor and friend on my academic journey. I will always be grateful for his help and
guidance, and will do my best to carry forward the academic spirit.
This paper is a joint achievement, and I am grateful for your companionship throughout the
journey, which has enabled me to achieve such success.
iii
Acknowledgements
I would like to express my deepest gratitude to my advisor, Professor Ming Li, for providing
me with invaluable guidance on the direction of my research, recommending data resources,
suggesting methodologies and relevant literature, as well as providing insightful comments and
feedback on the entire thesis. His unwavering support, patience, and expertise have been
indispensable throughout this process.
I would also like to extend my heartfelt appreciation to Professor Todd A. Alonzo and
Professor Jin Piao for their invaluable suggestions and feedback on the revisions of my thesis.
Their expertise, dedication, and constructive criticism have significantly contributed to the
improvement of the quality of my work.
Finally, I would like to thank my family and friends for their unwavering support and
encouragement throughout this journey.
iv
Table of contents
Dedication ............................................................................................................................................... ii
Acknowledgements ................................................................................................................................ iii
List of Tables ......................................................................................................................................... vii
List of Figures ...................................................................................................................................... viii
Abbreviations ....................................................................................................................................... ixx
Abstract ................................................................................................................................................... x
Chapter One: Introduction ...................................................................................................................... 1
Chapter Two: Methods ............................................................................................................................ 5
2.1 Survival analysis overview .................................................................................................... 5
2.1.1 Start event and end event ................................................................................................ 6
2.1.2 Survival time .................................................................................................................. 6
2.1.3 Complete data and censored data ................................................................................... 6
2.1.4 Censor ............................................................................................................................. 6
2.1.5 Survival probability and survival rate ............................................................................ 7
2.1.6 Probability of death and mortality rate ........................................................................... 7
2.1.7 Nonparametric statistics ................................................................................................. 8
2.1.8 Univariable survival analysis ......................................................................................... 8
2.1.9 Multiple variables survival analysis ............................................................................... 8
2.1.10 Dynamic nomogram ..................................................................................................... 9
2.1.11 Model evaluation index ................................................................................................ 9
2.2 Data source .......................................................................................................................... 11
2.3 Participants .......................................................................................................................... 12
2.3.1 Inclusion criteria ........................................................................................................... 12
2.3.2 Exclusion criteria .......................................................................................................... 12
2.3.3 Covariates ..................................................................................................................... 13
2.4 Follow-up design ................................................................................................................. 14
2.5 Statistical analysis ............................................................................................................... 14
Chapter Three: Results .......................................................................................................................... 16
3.1 Clinical characteristics of patients with stage IV NSCLC ................................................... 16
3.2 Cox univariable and multivariable regression analysis ....................................................... 18
3.3 Survival analysis of patients with stage IV NSCLC ............................................................ 22
3.3.1 Overall survival analysis .............................................................................................. 22
3.3.2 Survival analysis of each baseline characteristics in the training group ...................... 23
v
3.4 Construction and verification of survival prediction model ................................................ 43
3.4.1 Construction of Nomogram prediction model .............................................................. 43
3.4.2 Verification of Nomogram prediction model ............................................................... 44
Chapter Four: Discussion ...................................................................................................................... 51
Chapter Five: Conclusion ...................................................................................................................... 59
References ............................................................................................................................................. 60
vi
List of Tables
Table 1. The baseline characteristics of patients with stage IV NSCLC..............................16
Table 2. Univariate and multiple variables analysis in training group in patients with
stage IV NSCLC.........................................................................................................19
vii
List of Figures
Figure 1. Comparison of OS between different subgroups with the number of
transplanted organs....................................................................................................23
Figure 2. Comparison of OS between different subgroups of age.......................................24
Figure 3. Comparison of OS between different subgroups of sex.......................................25
Figure 4. Comparison of OS between different subgroups of race......................................26
Figure 5. Comparison of OS between different subgroups of marital status.......................27
Figure 6. Comparison of OS between different subgroups of primary site..........................28
Figure 7. Comparison of OS between different subgroups of histological type..................29
Figure 8. Comparison of OS between different subgroups of differential degree...............30
Figure 9. Comparison of OS between different subgroups of T-stage.................................31
Figure 10. Comparison of OS between different subgroups of N-stage..............................32
Figure 11. Comparison of OS between different subgroups of number of
metastatic organs.......................................................................................................33
Figure 12. Comparison of OS between different subgroups of metastatic organ.................34
Figure 13. Comparison of OS between different subgroups of treatment in patients
with single metastatic organ.......................................................................................35
Figure 14. Comparison of OS between different subgroups of treatment in patients
with two metastatic organs.........................................................................................36
Figure 15. Comparison of OS between different subgroups of treatment in patients
with three metastatic organs.......................................................................................37
Figure 16. Comparison of OS between different subgroups of treatment in patients
with lung and brain metastasis...................................................................................38
Figure 17. Comparison of OS between different subgroups of treatment in patients
with lung and bone metastasis....................................................................................39
Figure 18. Comparison of OS between different subgroups of treatment in patients
with lung and liver metastasis....................................................................................40
Figure 19. Comparison of OS between different subgroups of treatment in patients
with brain and bone metastasis..................................................................................41
Figure 20. Comparison of OS between different subgroups of treatment in patients
with brain and liver metastasis...................................................................................42
Figure 21. Comparison of OS between different subgroups of treatment in patients
with bone and liver metastasis...................................................................................43
Figure 22. Survival prediction model for stage IV NSCLC patients in the training group..44
viii
Figure 23. Predicted 3 - and 5-year survival ROC curves in the modeling group................45
Figure 24. Predicted 3 - and 5-year survival ROC curves in the verification group.............46
Figure 25. Predicted 3-year OS calibration curve in the training group..............................47
Figure 26. Predicted 5-year OS calibration curve in the training group..............................48
Figure 27. Predicted 3-year OS calibration curve in the verification group........................49
Figure 28. Predicted 5-year OS calibration curve in the verification group........................50
ix
Abbreviations
Full name Abbreviations
Non-small cell lung cancer NSCLC
Oligo metastases OM
National Comprehensive Cancer Net-Work NCCN
Surveillance, Epidemiology and End Results SEER
Hazard ratio HR
Confidence Interval CI
overall survival OS
concordance index C-index
Receiver operating characteristic ROC
Area Under the Curve AUC
Lung adenocarcinoma LUAD
Lung squamous carcinoma LUSC
x
Abstract
Background: Non-small cell lung cancer (NSCLC), which accounts for about 85% of all
lung cancers, is a major health burden on global health. The treatment and prognosis of
NSCLC usually depend on the stage of NSCLC at the time of diagnosis. Although great
progress has been made in the treatment of advanced lung cancer in recent years, there has
been no significant improvement in overall survival. Hellman and Weichselbaum proposed
the Oligo metastases (OM) theory for the early stage of tumor metastasis, which describes
the appearance of single or small metastases in local organs. The M staging of the 8th
edition of AJCC-TNM staging system for lung cancer described intrathoracic metastasis as
M1a. And extrapural metastasis was further divided into single organ metastasis and
multiple organ metastasis, referred to as M1b and M1c, respectively. The M1b staging is
consistent with the description of OM. The conventional viewpoint is that patients with
stage IV lung cancer prefer palliative systemic treatment to aggressive surgery. However,
in recent years, many studies have confirmed that surgical resection of the primary site and
metastatic site in NSCLC patients with a small number of local organ metastases can
significantly improve the prognosis. The National Comprehensive Cancer Net-Work
(NCCN) guidelines (version 2021.4) also suggest that radical local treatment can prolong
survival for patients with single or oligometastatic NSCLC. To further clarify the surgical
indications of patients with stage IV NSCLC and analyze the prognosis of patients with
stage IV NSCLC, a survival prediction model was constructed using case data extracted
from the Surveillance, Epidemiology and End Results (SEER) database.
xi
Methods: Data for patients diagnosed with stage Ⅳ NSCLC between 2010 and 2015 were
extracted from the SEER database. The included stage Ⅳ NSCLC patients were
randomized using STATA 15.1 at a ratio of 7:3 and divided into training set and validation
set. A Chi-square test was used to compare the differences between the groups. Cox
univariable regression analysis was performed on the training group, and variables with
statistical significance were selected into Cox multivariable analysis. Independent
prognostic factors were identified. The results are shown as Hazard ratio (HR) and 95%
Confidence Interval (CI). The overall 3-year and 5-year cumulative survival rates were
calculated in life table method. Kaplan-Meier method was used to analyze the survival of
each variable, and survival curves were drawn. Differences among curves were tested by
the Log-rank method. A nomogram survival prediction model was constructed to predict
overall survival (OS) in stage IV NSCLC patients. The concordance index (C-index) was
calculated to evaluate the model performance. Receiver operating characteristic (ROC)
curve and calibration curve were combined to evaluate the accuracy of the model. P < 0.05
was considered statistically significant.
Results: The median survival time of all subjects was 6.01 months, 7.05 months in the
single-organ transfer group, and 4.46 months in the multi-organ transfer group. The median
age of all cases was 67 years, 54.5% were male, 77.2% were Caucasian, 51.3% were
married, 71.7% had adenocarcinoma, 56.1% had primary tumor location in the upper lobe
xii
of the lung, most patients had unknown differentiation, 35.0% had stage T4, and 46.1%
had stage N2. Metastasis to a single organ was found in 63.9% of patients. 21.4% of the
patients chose surgical treatment, among which 1039 patients received primary site surgery,
2722 patients received metastatic site surgery, and 291 patients received primary site
surgery and metastatic site surgery. Multivariable Cox analysis in the training group
showed that age of diagnosis, gender, race, marital status, primary tumor location,
histological type of tumor, degree of differentiation of tumor cells, tumor size, T-stage, N
stage of lymph node metastasis, number of distant metastatic organs and treatment were all
independent factors affecting prognosis of stage IV NSCLC patients. A Nomogram
survival prediction model was established based on multivariable Cox analysis. Survival
analysis showed that surgical treatment was an independent prognostic factor for stage IV
NSCLC patients with single organ metastases and lung and brain metastases. The internal
verification of the model showed that the C-index of the training group was 0.714 (95%
CI: 0.710-0.718), and the prediction model showed that the Area Under the Curve (AUC)
of the 3-year OS was 0.848 and the AUC of the 5-year OS was 0.842. In the verification
group, the calculated C index was 0.712 (95% CI:0.706-0.718), and the AUC for 3-year
OS and 5-year OS were 0.846 and 0.841, respectively. The calibration curves of both the
training group and the verification group were close to 45 degrees, which ensured the
accuracy and reliability of the prediction model.
Conclusion: Multivariable Cox analysis showed that age at the first diagnosis, sex, race,
xiii
marital status, primary tumor location, histological type of tumor, degree of differentiation
of tumor cells, tumor size, T-stage, N-stage of lymph node metastasis, number of distant
metastatic organs and treatment methods were all independent factors affecting prognosis
of patients with stage IV NSCLC. The OS of patients with multiple organ metastasis was
significantly lower than that of patients with single organ metastasis. The single organ
metastasis group could be divided into M1a (single lung metastasis) and M1b (single bone,
brain and liver metastasis) according to the 8
th
version of TNM staging, and the OS of
patients in group M1a was higher than that in group M1b. The 3-year, and 5-year survival
rates of patients with stage IV NSCLC can be predicted by the line graph model. Surgical
treatment of patients with single organ metastasis and those with two-organ metastasis of
the lung and brain can significantly improve the prognosis. Surgical excision was an
independent factor for improvement in patients with oligometastatic stage IV NSCLC.
Keywords: Non-small cell lung cancer, metastasis, Surgery, Survival prediction model,
SEER database
1
Chapter One: Introduction
Lung cancer is a major burden of global health and the second most diagnosed cancer
in both men and women in the United States (Siegel et al., 2021). Lung cancer accounted
for about a quarter of all cancer deaths in 2021. Among all types of Lung cancer, non-small
cell Lung cancer (NSCLC) accounts for about 85%. The most common subtypes of
NSCLC are Lung adenocarcinoma (LUAD) and Lung squamous carcinoma (LUSC)
(Herbst et al., 2018). The treatment and prognosis of NSCLC usually depend on the stage
of progression at the time of diagnosis. The determination of clinical staging is the most
important basis for determining whether a patient with NSCLC has the option of surgical
treatment and is inoperable but may benefit from chemotherapy and/or radiotherapy.
Surgical treatment is an important treatment for patients with NSCLC. It is not only the
first choice for patients with stage I or II NSCLC, but also an acceptable treatment for
patients with stage IIIB and IV NSCLC (Hoy et al., 2019). Although great progress has
been made in the treatment of advanced lung cancer in recent years, there has been no
significant improvement in overall survival. In 2014, the UK Office for National Statistics
reported that the 1-year OS of stage IV lung cancer patients was only 15-19%, while that
of stage I patients was 81-85% (Blandin Knight et al., 2017). The 5-year overall survival
rate for stage IV NSCLC patients was less than 5% until the last decade (Arbour & Riely,
2019). The main reason for the significant difference in survival rate is that about 57% of
lung cancer patients were advanced (stage Ⅳ) when first diagnosed (Simeone et al., 2019).
The decline in lung cancer deaths accounted for nearly half of the total decline in cancer
2
deaths from 2014 to 2018, and the rapid decline likely reflects improvements in treatment.
Hellman and Weichselbaum first proposed the theory of oligometastasis (OM) in 1995,
describing the state of single or small amount of metastasis in local organs (Hellman &
Weichselbaum, 1995). OM means that the tumor is in the early stable stage of metastasis,
and the number and location of metastatic lesions are limited. In 2019, the European
consensus proposed that OM-NSCLC has a maximum of five metastatic lesions, among
which three different organs are involved at most (Dingemans et al., 2019). However, the
location and number of OM have not been clearly defined in clinical practice. The AJCC
defines malignancy stage as "the severity of an individual's cancer based on the size of the
primary tumor and how far the cancer has spread in the body." In the 8
th
edition of AJCC-
TNM stage of lung cancer, intrathoracic metastasis was described as M1a in the M stage,
and extracranial metastasis was further divided into single organ metastasis and multiple
organ metastasis, which were described as M1b and M1c, respectively. Among of these,
the M1b stage was consistent with the description of OM. Although the difference in
survival between M1a and M1b is small, AJCC prefers to record OM disease separately
because patients with OM disease can receive more aggressive and radical local therapy to
improve prognosis in addition to systemic treatment (Gregor et al., 2020).
The survival rate of patients with stage IV NSCLC is extremely low, with a median
survival of only 8-11 months. Currently, the gold standard of treatment is still
multidisciplinary systematic therapy rather than surgical treatment. About 7% of patients
with stage IV NSCLC show only a single or small number of metastases after
3
comprehensive examination (Patrini et al., 2018). It is believed that patients with stage IV
lung cancer are more inclined to take palliative chemotherapy than aggressive surgical
treatment (Alexander et al., 2020). In recent years, many studies have confirmed that the
prognosis of NSCLC patients with distant metastases is not absolutely poor, and the
prognosis of NSCLC patients with a small number of local organ metastases can be
significantly improved after surgical resection of the primary site and metastatic site (David
et al., 2019; Hao et al., 2021; C. F. J. Yang et al., 2018; Zhang et al., 2019). The 4
th
Edition
of NCCN guidelines also proposed that radical local treatment for NSCLC patients with
single metastasis or OM can prolong survival time (Ettinger et al., 2022). A cohort study
of intracranial OM-NSCLC showed that neurosurgical resection or stereotactic
radiosurgery and pulmonary resection could significantly increase the survival time of
patients, which further confirming the benefit of radical local treatment for survival of
intracranial OM-NSCLC. For extracranial OM-NSCLC, Buero A et al. confirmed that
NSCLC patients with single adrenal metastasis can benefit long-term survival after lung
resection and adrenal resection (Congedo et al., 2012). Surgical treatment of stage IV
NSCLC may be beneficial.
Although some stage IV NSCLC may benefit from surgery, it is not clear whether all
patients benefit from surgery. Some studies have proposed that systemic inflammation
caused by surgical trauma will circulate the adhesion of tumor cells to vascular
endothelium of distant organs, which is easy to lead to cancer recurrence (Nojiri et al.,
2017). Takenaka M et al. retrospectively analyzed the prognosis of 105 patients who
4
received secondary surgery for lung cancer and found that the recurrence rate of advanced
lung cancer was relatively high, and the application of surgery in advanced lung cancer
should be limited (Takenaka et al., 2021). A study on radical local treatment of elderly
patients with stage IV NSCLC found that surgical treatment reduced OS in patients aged
70 years and older with stage IV NSCLC (Qiu et al., 2020). To sum up, it is not clear
clinically which stage IV NSCLC patients are suitable for surgical treatment. Therefore, in
order to further clarify the surgical indications of stage IV NSCLC patients and analyze the
prognosis of stage IV NSCLC patients, we collected a large number of clinical data in the
SEER database for statistical analysis. To provide a reliable reference for the clinical
treatment of stage IV NSCLC patients.
5
Chapter Two: Methods
2.1 Survival analysis overview
Survival analysis which is also called event time analysis is mainly used to deal with
dichotomous outcome variables with its time of occurrence. This method suit for data
combined with the status of event occurrence and the length of the occurrence time.
Generally, survival analysis includes outcome event, such as death or disease recurrence
and censored data. There are five features of data for survival analysis. Firstly, it includes
event outcome and corresponding survival time. Secondly, the distribution of survival time
does not conform to the normal distribution. Thirdly, event outcome is mutually exclusive
event. Fourthly, the data of survival analysis are generally collected through follow-up trial
with a certain start time and a specific end time. Fifthly, the data contain censored data
which means other events occur except for the outcome event of research, resulting in
uncertain survival time. Survival analysis is mainly used to predict the prognosis and
survival of patients.
There are four basic requirements for data of survival analysis. First, the number of
samples and the proportion of the outcome events should not be too small. Second, the
proportion of completed data should not be too small which means the censored data should
not be too much. Third, the reason of censored data should be unbiased. Fourth, survival
time is as accurate as possible.
6
2.1.1 Start event and end event
The start event refers to the initial characteristics event that reflects the survival
process of the research object. The end event refers to a specific outcome that researcher
study, such as death.
2.1.2 Survival time
Survival time is the span of time between the start event and the end event, which does
not conform to a normal distribution. It may satisfy exponential distribution, Weibull
distribution, loglogistic distribution, or any distribution at all.
2.1.3 Complete data and censored data
Completed data refers to research object experienced the start event and outcome event,
with exact survival time. Censored data refers to the end event does not occur due to some
reason which resulting missing survival time. For example, censored data generated when
patients were lost to follow-up (lost contact with the investigator), dropped out due to
research unrelated reason or terminated (reached the study endpoint without end event).
2.1.4 Censor
Censor include left censoring, right censoring and interval censoring which could
generate censored data. Left censoring refers to the knowledge that patients occur end event
before the start of follow-up, which cannot be located to the specific time point. Right
7
censoring refers to knowing that the end event will occur after the outcome event. Generally,
it means that the end event is not observed at the end of follow-up, but at some time in the
future, the end event will still occur. Interval censoring refers to knowing that the end event
will occur in a certain time period.
2.1.5 Survival probability and survival rate
Survival probability refers to the probability that an individual alive at the beginning
and the end of the study period which is the number of people living for a full study period
divided by the number of people observed at the beginning of cohort. If there is a censored
data, the population needs to be corrected, which is equal to the number of observations at
the beginning of the cohort minus half of the number of censored dases.
Survival rate which is denoted as S(t) refers to the probability of surviving after
multiple time periods. It was calculated by the number of cases still alive after multiple
time periods divided by the total number of cases. The segmented survival probability
should be calculated if there is censored data.
2.1.6 Probability of death and mortality rate
The probability of death refers to the probability that an individual alive at the
beginning of a study period will die at the end of that study period.
Mortality rate is the probability of dying after multiple study period.
8
2.1.7 Nonparametric statistics
Nonparametric statistics include kaplan-Meier method and life table method. kaplan-
Meier method, also known as the product-limit method, is the most commonly used method
for estimating survival rates. It was suitable for large or small samples with available
original data. Life table method refers to the survival data which has been summarized into
a frequency table with a number of time period. While the original survival data wasn’t
available.
2.1.8 Univariable survival analysis
Univariable survival analysis includes log-rank test and Breslow test. Both of them
were non-parametric statistical methods to test whether two or more groups have the same
distribution of overall survival time. Breslow test was more sensitive to early differences.
While log-rank test was more sensitive to late differences.
2.1.9 Multiple variables survival analysis
Multiple variables survival analysis could be used to estimate and compare the survival
rate of groups after adjusting multiple influential factors. Before fitting Cox regression
model, the equal proportional risk should be satisfied and proportional linearity should be
tested. The sample size requirement is similar to logistic regression, requiring at least 10
times the number of independent variables.
The basic form of Cox proportional risk regression model is as follows: the risk of
9
failure event of individual t at a certain time point is divided into two parts: h0(t) and h(t1X).
The factor X increase the risk function from h0(t) by exp(βiXi) to h0(t)*exp(βiXi). The
model influenced by multiple factors is h0(t)*exp(β1X1+... + beta iXi).
The segmented Cox regression model is used when the proportional risk assumption
is established. Although the equal proportional assumption is not established during the
whole study, it is effective in a short period. Therefore, the whole study period can be
divided into several short periods to match with the equal risk proportional model. The
final fitting could divide the data into several segment or establish a unified model using
the interaction of time and covariables (time dependent covariables).
2.1.10 Dynamic nomogram
The dynamic nomogram is a kind of graph based on multiple variables Cox regression
model. Various prediction indicators are listed and scored on a graph in a certain proportion
with a graduated line segment. A total score which is used to predict the final outcome was
obtained by adding each score of all variables together.
2.1.11 Model evaluation index
C-index, called Harrell’s concordance index, can be used to evaluate the predictive
ability of the model. It was proposed by a biostatistics professor in 1996 and mainly used
to calculate the differentiation between the predicted value and the real value of the Cox
model in survival analysis. Specifically, it refers to randomly forming pairs of research
10
objects in research data, and judging the proportion of pairs whose predicted results are
consistent with the actual results. The C-index is between 0.5 and 1. A random model with
no predictive effect was presumed if C-index was 0.5. While an ideal model was affirmed
if C-index was 1, indicating that the predicted results of the model are completely
consistent with the reality. It was believed that a low accuracy was determined when C-
index was between 0.50 to 0.70; a medium accuracy for 0.71-0.90; a high accuracy for
higher than 0.90.
The ROC curve was used to assess the predictive power of an investigated factor. The
standard ROC curve considered the individual disease status and influential factors as fixed
value which did not take the time factor into consideration in the analysis. While, the
disease status and baseline characteristics of patients will change with the study proceed.
Therefore, time-dependent ROC is more suitable. The time-dependent ROC curve was
mainly utilized in the following situation: first, it was used for investigating the optimal
cut-off value of continuous variables in survival data. Second, it was used to evaluate the
predictive efficacy of variables for outcomes with survival data. Third, it was used for
evaluating the influence of a certain variables on the outcomes over time. The horizontal
coordinate of the curve 1-specificity which is false positive rate. The ordinate represents
sensitivity which is the true positive rate. The relationship between the false positive rate
and the true positive rate can be shown by ROC curve. The larger the area under the curve,
the higher the accuracy of diagnosis. Conversely, the smaller the area under the curve, the
lower the accuracy of diagnosis.
11
C-index and AUC confirm the discriminant ability of the model. Calibration curve can
well evaluate the calibration degree of the model. The calibration curve is a scatter plot of
actual and predicted incidence. It is the visualization of the results of the Hosmer-
Lemeshow goodness of fit test. However, neither of them takes into account whether
patients receive benefits or not. There may be two extreme cases if the benefits of patients
are not taken into account. The first is that all patients do not receive additional intervention
with receiving none additional benefits. The second is that additional interventions are used
in all patients which are anticipated an outcome event. While only some patients benefit
from the intervention. The decision analysis curve is a good way to evaluate whether the
clinical benefit of using the clinical prediction model.
2.2 Data source
All data included in this study were collected from the SEER database
(https://seer.cancer.gov). The SEER database covers about 35 percent of the U.S.
population and is considered representative of the whole population. To date, the SEER
database is the only comprehensive source of data on patient demographics, tumor
characteristics, and survival status and time of follow-up.
The SEER database is publicly available, and we use SEER*Stat software (version
8.3.9) to select "Incidence-SEER 18 Regs Custom Data (with additional treatment fields)
in the Data module, Nov 2018 Sub" for data extraction.
12
2.3 Participants
2.3.1 Inclusion criteria
(1) The year of diagnosis was from 2010 to 2015;
(2) The primary Site of tumor was Lung and bronchus (site recode ICD-O-3/WHO 2008 =
Lung and Bronchus);
(3) According to the International Classification of Tumor Diseases, Third Edition (ICD-
O-3), pathologically confirmed non-small cell lung cancer, Including adenocarcinoma
(8140, 8144, 8200, 8250, 8251, 8252, 8254, 8255, 8260, 8310, 8323, 8480, 8481, 8490,
8550), Squamous cell carcinoma (8052, 8070, 8071, 8072, 8073, 8074, 8075, 8083,
8084), Others (8012, 8013, 8014, 8022, 8030, 8031, 8032, 8050, 8082, 8123, 8230,
8240, 8249, 8430, 8560, 8980);
(4) According to the TNM stages Derived AJCC (7th ed), the M1 stages were selected.
2.3.2 Exclusion criteria
(1) Non-primary tumor;
(2) Patients diagnosed at autopsy or at death;
(3) The pathology is not clear;
(4) Patients younger than 18 years of age at diagnosis;
(5) Unavailable clinical staging;
(6) The surgical condition is unknown;
(7) Distant metastasis is unclear.
13
2.3.3 Covariates
Covariates include age at diagnosis, sex, race, marital status, primary location, histological
type, degree of tumor differentiation, T staging, N staging, distant metastatic organ,
primary site surgery, metastatic site surgery, radiotherapy, chemotherapy, survival status,
survival time (i.e., follow-up time) and cause of death (tumor-related death or non-tumor-
related death). According to the SEER database, age groups were divided into < 45 years
old, 45-59 years old, 60-74 years old, and ≥75 years old. The population is ethnically
divided into White, black, Asian, Pacific Islander, and other races. Marital status is grouped
into unmarried, married, single (including divorced, separated, widowed) and other status.
Primary tumor sites were grouped into the main bronchus, upper lobe, middle lobe, lower
lobe, overlapping lung lesions, and lung, NOS (lung, not specifically referred to).
Histological types were divided into adenocarcinoma, squamous cell carcinoma and others.
T stages of tumor size were divided into T1, T2, T3 and T4. The N stages of lymph node
metastasis were N0, N1, N2 and N3. The metastatic organs were divided into 1, 2, 3 and 4
organs. The treatment methods were divided into primary surgery (including radiotherapy
or chemotherapy combined with or without radiotherapy or/chemotherapy), metastatic
surgery, primary and metastatic surgery, and non-surgical treatment (including radiotherapy,
chemotherapy, radiotherapy combined with chemotherapy, palliation).
14
2.4 Follow-up design
The end point of follow-up was set as December 31, 2015. The outcome event of the
study is the overall survival time (OS time) of the study object. OS refers to time from a
precise diagnosis of NSCLC in a patient to all-cause mortality. Follow-up time and survival
time ranged from diagnosis of NSCLC to study endpoint or death.
2.5 Statistical analysis
In this study, STATA 15.1 was used to process, analyze data and generate figures. Firstly,
stage Ⅳ NSCLC patients included in this study were randomized at a ratio of 7:3, and
divided into training group and Verification group. The differences in baseline
characteristics between the groups of variables were compared by Chi-square test. The
overall survival and 3-year and 5-year cumulative survival rates were calculated using the
life table method. Cox univariable regression analysis was performed on the training group,
and variables with P<0.05 were introduced into Cox Multivariable analysis. The results
were shown as Hazard ratio (HR) and 95% Confidence Interval (CI). The Kaplan-Meier
method was used to analyze the survival of each variable, and survival curves were drawn.
Differences among curves were tested by Log-rank method and nomogram survival
prediction model was constructed. The model performance was verified internally with C-
index, ROC curve and AUC to evaluate the accuracy of the model's predictive ability. The
calibration curve was drawn to evaluate the consistency of the prediction results. The model
risk score was calculated, and the median score was used as cut-off value to divide the
15
patients into high-risk group and low-risk group, and survival analysis was conducted for
risk grouping and transfer surgery. P<0.05 was considered statistically significant.
16
Chapter Three: Results
3.1 Clinical characteristics of patients with stage IV NSCLC
According to the above inclusion and exclusion criteria, a total of 35,350 patients were
included in this study from 2010 to 2015 stage IV NSCLC patients with intermediate
diagnosis. All patients were randomized in a 7:3 ratio (Table 1).
Table 1. The baseline characteristics of patients with stage IV NSCLC.
SEER variables Total (n=35350) Training group
(n=24746)
Verification group
(n=10604)
P value
Age (%), years 0.097
< 45 770 (2.2) 555 (2.2) 215 (2.0)
45-59 8476 (24.0) 6009 (24.3) 2467 (23.3)
60-74 16960 (48.0) 11813 (47.7) 5147 (48.5)
≥ 75 9144 (25.9) 6369 (25.7) 2776 (26.2)
Sex (%) 0.750
Male 19254 (54.5) 13492 (54.5) 5762 (54.3)
Female 16096 (45.5) 11254 (45.5) 4842 (45.7)
Race (%) 0.789
White 27274 (77.2) 19071 (77.1) 8203 (77.4)
Black 4673 (13.2) 3293 (13.3) 1380 (13.0)
Asian 3172 (9.0) 2225 (9.0) 947 (8.9)
Other 231 (0.7) 157 (0.6) 74 (0.7)
Marital status (%) 0.898
Married 5574 (15.8) 3914 (15.8) 1660 (15.7)
Single 18142 (51.3) 12674 (51.2) 5468 (51.6)
Divorced/separated/widowed 10087 (28.5) 7081 (28.6) 3006 (28.3)
Other 1547 (4.4) 1077 (4.4) 470 (4.4)
Primary site (%) 0.760
Main bronchus 1500 (4.2) 1067 (4.3) 433 (4.1)
Superior lobe 19816 (56.1) 13858 (56.0) 5958 (56.2)
Middle lobe 1475 (4.2) 1041 (4.2) 434 (4.1)
Inferior lobe 9743 (27.6) 6830 (27.6) 2913 (27.5)
Multiple lesions 312 (0.9) 222 (0.9) 90 (0.8)
Lung, NOS 2504 (7.1) 1728 (7.0) 776 (7.3)
Histological type (%) 0.442
Adenocarcinoma 25351 (71.7) 17783 (71.9) 7568 (71.4)
Squamous cell carcinoma 8146 (23.0) 5688 (23.0) 2458 (23.2)
Unclear 1853 (5.2) 1275 (5.2) 578 (5.5)
Differentiated degree (%) 0.793
Grade I 981 (2.8) 672 (2.7) 309 (2.9)
Grade II 4762 (13.5) 3339 (13.5) 1423 (13.4)
Grade III 9302 (26.3) 6498 (26.3) 2804 (26.4)
Grade IV 342 (1.0) 245 (1.0) 97 (0.9)
Unclear 19963 (56.5) 13992 (56.5) 5971 (56.3)
T stage (%) 0.114
T1 4360 (12.3) 2999 (12.1) 1361 (12.8)
T2 9242 (26.1) 6521 (26.4) 2721 (25.7)
T3 9364 (26.5) 6597 (26.7) 2767 (26.1)
T4 12384 (35.0) 8629 (34.9) 3755 (35.4)
N stage (%) 0.456
N0 8505 (24.1) 5901 (23.8) 2604 (24.6)
N1 2961 (8.4) 2061 (8.3) 900 (8.5)
17
N2 16291 (46.1) 11448 (46.3) 4843 (45.7)
N3 7593 (21.5) 5336 (21.6) 2257 (21.3)
Number of metastatic organs (%) 0.775
One 22605 (63.9) 15797 (63.8) 6808 (64.2)
Two 9147 (25.9) 6417 (25.9) 2730 (25.7)
Three 3014 (8.5) 2129 (8.6) 885 (8.3)
Four 584 (1.7) 403 (1.6) 181 (1.7)
Treatment (%) 0.216
Primary site surgery 1039 (2.9) 707(2.9) 332(3.1)
Metastatic site surgery 2722(7.7) 1948(7.9) 774(7.3)
Combined surgery 291(0.8) 196(0.8) 95(0.9)
Radiotherapy 5718(16.2) 3957(16.0) 1761(16.6)
Chemotherapy 7639(21.6) 5344(21.6) 2295(21.6)
Chemoradiotherapy 10191(28.8) 7140(28.9) 3051(28.8)
Palliative treatment 7750(21.9) 5454(21.9) 2296(21.7)
The mean age of all patients was 66.79 years old, and the median age was 67 years old
(18-100 years old). Most of the study population were elderly patients. Male patients
accounted for 54.5% of the total patient population, a total of 19,254 cases. The largest
proportion of patients in the study population were Caucasian and married (77.2% and
51.3%). The most common primary location of NSCLC is the upper lobe of the lung,
accounting for 56.1%, followed by the lower lobe of the lung, accounting for 27.6%. There
were 25351 patients (71.7%) with adenocarcinoma, 8146 patients (23.0%) with squamous
cell carcinoma, and 1853 patients (5.2%) with other histological types. The degree of
differentiation of the vast majority of patients was unknown, accounting for 56.5%,
followed by the poorly differentiated patients (9302 cases, accounting for 26.3%). The
subjects of this study were stage IV NSCLC patients. For T stage, there were 4360 patients
(12.3%) in T1 stage, 9242 patients (26.1%) in T2 stage, 9364 patients (26.5%) in T3 stage,
and 12384 patients (35.0%) in T4 stage. For N stage, the majority of patients with lymph
node metastasis, 26845 (75.9%). 16291 patients (46.1%) were in stage N2, followed by
stage N3 (7593, 21.5%) and stage N1 (2961, 8.4%). A total of 8505 patients (24.1%) did
not develop lymph node metastasis. A total of 22605 patients (63.9%) had only one
18
metastatic site and 12745 patients (36.1%) had multiple metastatic sites. The number of
patients with bone, brain, lung and liver metastases was 7854, 5993, 6882 and 1876,
respectively. The number of patients with two, three, and four metastatic sites was 9147,
3014, and 584, respectively. 21.4% of the patients underwent surgery, of which 1039
underwent primary site surgery, 2722 underwent metastatic site surgery, and 291 patients
underwent both primary site and metastatic site surgery. Non-surgical treatment accounted
for a large proportion of patients (23,548 cases (66.6%)), followed by palliative care (7750
cases (21.9%), and surgical treatment was the least (1330 cases (3.8%). For patients with
single organ metastasis, 2942 cases (8.3%) chose surgery, while for patients with multiple
organ metastasis, only 1110 cases (3.1%) chose surgery. However, no data was available to
record specific surgical interventions for each patient.
Chi-square test showed that there was no significant difference between the two groups
in each baseline features (P > 0.05), which indicates that patients between two groups were
comparable (Table 1).
3.2 Cox univariable and multivariable regression analysis
The demographic characteristics, clinicopathological characteristics and treatment
methods of the patients in the training group were included in the univariable Cox
regression analysis. The results showed that the prognostic factors of stage IV NSCLC
patients included age of diagnosis, sex, race, marital status, primary location, histological
type, degree of tumor differentiation, T stage, N stage, number of metastatic organs and
19
treatment methods. All P values were less than 0.05 (Table 2).
Table 2. Univariate and multiple variables analysis in training group in patients
with stage IV NSCLC.
Variables Univariate analysis Multiple variables analysis
HR (95%CI) P value HR (95%CI) P value
Age, years
< 45 Reference - Reference -
45-59 1.512(1.369-1.669) <0.001 1.383 (1.253-1.528) <0.001
60-74 1.766(1.603-1.945) <0.001 1.558(1.413-1.718) <0.001
≥ 75 2.236(2.026-2.467) <0.001 1.703(1.540-1.883) <0.001
Sex
Male Reference - Reference -
Female 0.781(0.760-0.802) <0.001 0.806(0.784-0.829) <0.001
Race
White Reference - Reference -
Black 1.020(0.981-1.061) 0.314 0.983 (0.944-1.022) 0.386
Asian 0.664(0.632-0.697) <0.001 0.678(0.645-0.713) <0.001
Other 0.870(0.731-1.034) 0.114 0.818(0.687-0.972) 0.023
Marital status
Single Reference - Reference -
Married 0.864(0.832-0.897) <0.001 0.924(0.888-0.961) <0.001
Divorced/separated/widowed 1.038(0.996-1.082) 0.074 1.003(0.961-1.046) 0.905
Other 0.938(0.874-1.007) 0.079 0.994(0.926-1.068) 0.874
Primary site
Main bronchus Reference - Reference -
Superior lobe 0.790(0.741-0.843) <0.001 0.849(0.796-0.907) <0.001
Middle lobe 0.695(0.635-0.761) <0.001 0.776(0.709-0.850) <0.001
Inferior lobe 0.803(0.751-0.859) <0.001 0.864(0.807-0.925) <0.001
Multiple lesions 0.916(0.787-1.065) 0.255 0.966(0.830-1.124) 0.652
Lung, NOS 0.882(0.814-0.955) 0.002 0.917(0.846-0.994) 0.034
Histological type
Adenocarcinoma Reference - Reference -
Squamous cell carcinoma 1.378(1.336-1.422) <0.001 1.174(1.136-1.213) <0.001
Unclear 1.089(1.025-1.157) 0.006 1.043(0.980-1.109) 0.187
Differentiated degree
Grade I Reference - Reference -
Grade II 1.362(1.241-1.494) <0.001 1.329(1.211-1.460) <0.001
Grade III 1.796(1.643-1.963) <0.001 1.741(1.591-1.905) <0.001
Grade IV 2.086(1.787-2.436) <0.001 1.827(1.563-2.136) <0.001
Unclear 1.697(1.556-1.852) <0.001 1.619(1.483-1.768) <0.001
T stage
T1 Reference - Reference -
T2 1.154(1.102-1.209) <0.001 1.105(1.054-1.157) <0.001
T3 1.248(1.191-1.307) <0.001 1.121(1.069-1.175) <0.001
T4 1.258(1.203-1.315) <0.001 1.084(1.035-1.135) <0.001
N stage
N0 Reference - Reference -
N1 1.108(1.050-1.168) <0.001 1.198(1.135-1.264) <0.001
N2 1.264(1.222-1.307) <0.001 1.313 (1.268-1.359) <0.001
N3 1.227(1.179-1.276) <0.001 1.345 (1.290-1.402) <0.001
Number of metastatic organs
One Reference - Reference -
Two 1.322(1.283-1.363) <0.001 1.374(1.332-1.418) <0.001
Three 1.521(1.451-1.594) <0.001 1.733(1.651-1.819) <0.001
Four 1.622(1.464-1.796) <0.001 1.908(1.720-2.117) <0.001
Treatment
Primary site surgery Reference - Reference -
Metastatic site surgery 1.879(1.700-2.077) <0.001 1.594(1.439-1.765) <0.001
Combined surgery 0.828(0.682-1.006) 0.057 0.837(0.689-1.017) 0.073
20
Radiotherapy 4.285(3.900-4.709) <0.001 3.438(3.125-3.784) <0.001
Chemotherapy 1.508(1.374-1.654) <0.001 1.221(1.111-1.342) <0.001
Chemoradiotherapy 1.724(1.573-1.890) <0.001 1.342(1.223-1.474) <0.001
Palliative treatment 4.887(4.454-5.361) <0.001 3.988(3.630-4.383) <0.001
In order to further exclude the interference of confounding factors on univariable
analysis, the variables mentioned above were included in multivariable analysis. The
results showed that the risk of death increased with the increase of age. Patients <45 years
old were taken as reference, and the HR of 45-59 years old patients was 1.383
(95%CI:1.253-1.528, P<0.001). The HR (95%CI) of patients aged 60-74 was 1.558 (1.413-
1.718, P<0.001), and the prognosis of patients aged ≥ 75 was the worst (HR=1.703, 1.540-
1.883; P<0.001). The prognosis of female patients was better than those of male patients.
Using male patients as reference, the HR of female patients was 0.806 (95%CI: 0.784-
0.829; P<0.001). HR=0.678 (95%CI:0.645-0.713; P<0.001) for Asian or Pacific Islanders
and 0.818 (95%CI:0.687-0.972; P=0.023) for other races compared with Caucasians.
Analysis of marital status showed that the HR (95%CI) of married patients was 0.924
(0.889-0.961; P<0.001) compared with unmarried patients. The primary location showed
that the HR of patients with primary tumor in the upper lobe of the lung was 0.849
(95%CI:0.796-0.907; P<0.001) and that of patients with primary tumor in the middle lobe
of the lung was 0.776 (95%CI: 0.799-0.850; P<0.001), patients with primary tumor in the
inferior lobe of the lung was 0.864 (95%CI: 0.807-0.925; P<0.001), patients with NOS
(lung, not specifically reference) was 0.917 (95%CI: 0.846-0.994; P<0.001), compared
with patients with primary tumor in the main bronchus. Histological types showed that
21
lung squamous cell carcinoma had the worst prognosis compared with lung
adenocarcinoma, HR=1.174 (95%CI:1.136-1.213; P<0.001). The risk of death increased
with the degree of differentiation. The risk of death in grade Ⅱ differentiation was 1.329
times higher than that in grade Ⅰ differentiation, increasing by grade Ⅲ (1.741 times) and
grade Ⅳ (1.827 times). T3 stage had the worst prognosis, HR=1.121 (95%CI: 1.069-1.175;
P<0.001), and T2 and T4 stages had an increased risk of death compared with T1 stage (T1:
HR=1.105 (95%CI: 1.054-1.157; T2: HR=1.084 (95%CI: 1.0535-1.135; P<0.001). The
risk of death increased with the increase of lymph node metastasis, and the HR (95%CI)
for stage N1, N2 and N3 were 1.198(1.135-1.264), 1.313 (1.268-1.359) and 1.345 (1.290-
1.402), respectively. The number of metastatic sites also predicted a patient's risk of death,
which was significantly increased in patients with multiple organ metastases compared to
patients with single organ metastases. The HR (95%CI) of 2, 3 and 4 metastatic organs
were 1.374 (1.332-1.418), 1.733 (1.651-1.819) and 1.908 (1.720-2.117), respectively. In
terms of treatment means, patients receiving surgical treatment had a better prognosis than
non-surgical treatment. The HR (95%CI) of metastatic site surgery was 1.594(1.439-1.765;
P<0.001) based on primary site surgery. In patients who did not receive surgery, the
prognosis of chemotherapy was better than that of receiving chemotherapy combined with
radiotherapy, and radiotherapy.
The results of multivariable analysis in the training group showed that aging, male,
squamous cell carcinoma of lung, degree of differentiation, T stage, N stage, increased
number of metastatic organs, and no surgical treatment were independent risk factors for
22
the prognosis of stage IV NSCLC patients. Race of Asian or Pacific Islander or other race,
being married, having primary disease in the upper, middle, lower lobes and lungs (not
specifically referred to), and receiving surgical treatment were independent factors for
favorable prognosis in patients with stage IV NSCLC.
3.3 Survival analysis of patients with stage IV NSCLC
3.3.1 Overall survival analysis
Combined with univariable and multivariable analysis results, Cox regression model
was used to screen out the risk factors in the disease progression of stage Ⅳ NSCLC
patients. Kaplan-Meier method was used to compare the effects of different clinical risk
factors on the prognosis of stage Ⅳ NSCLC patients. The survival analysis curve is shown
below.
A total of 31,591 patients died from various causes during the follow-up period,
including 28,334 from lung cancer and 3,257 from other causes, 919 from multiple
myeloma, 190 from COPD, 501 from heart disease and 101 from cerebrovascular disease.
They were divided into two subgroups according to the number of metastatic organs.
The median OS of patients with single organ transfer was 7.05 months, and the 1-year, 2-
year and 3-year survival rates were 32.64%, 17.35% and 10.84%, respectively (Figure 1).
The median OS of patients with multiple organ metastasis was 4.46 months, and the 1-year,
2-year and 3-year survival rates were 22.05%, 10.03% and 5.40%, respectively. The more
metastatic organs, the worse the prognosis.
23
Figure 1. Comparison of OS between different subgroups with the number of
transplanted organs.
3.3.2 Survival analysis of each baseline characteristics in the training group
There are significant differences in prognosis among different age groups. Patients <45
years old have significantly higher OS than other age groups. And the older the age, the
lower the OS (Figure 2). Female patients had better OS than male patients (Figure 3). There
is no significant difference in the OS of Caucasians and blacks. While the OS of Asian and
Pacific Islanders is significantly higher than that of other ethnic groups (Figure 4). As
24
shown in Figure 5, the OS of the married group is higher than that of other groups. All P
values are less than 0.001.
Figure 2. Comparison of OS between different subgroups of age.
25
Figure 3. Comparison of OS between different subgroups of sex.
26
Figure 4. Comparison of OS between different subgroups of race.
27
Figure 5. Comparison of OS between different subgroups of marital status.
The lowest OS was found in the primary bronchus and the highest OS was found in the
middle lobe of the lung (Figure 6). Patients with adenocarcinoma having significantly
higher OS than patients with other types (Figure 7). The worse the degree of differentiation,
the worse the prognosis (Figure 8). Patients with stage T1 have the highest OS, while
patients with stage T3 and T4 have poor prognosis (Figure 9). Patients with stage N0 have
the highest OS, while patients with stage N2 and stage N3 have poor prognosis (Figure 10).
28
Patients with single organ metastasis have significantly higher OS than patients with 2, 3,
and 4 organ metastasis (Figure 11). The more metastatic sites, the lower OS. All P values
are less than 0.0001.
Figure 6. Comparison of OS between different subgroups of primary site.
29
Figure 7. Comparison of OS between different subgroups of histological type.
30
Figure 8. Comparison of OS between different subgroups of differential degree.
31
Figure 9. Comparison of OS between different subgroups of T-stage.
32
Figure 10. Comparison of OS between different subgroups of N-stage.
33
Figure 11. Comparison of OS between different subgroups of number of metastatic
organs.
The median OS of patients with single organ metastasis in the training group was 7.08
months. And the 1-year, 2-year, and 3-year survival rates were 32.74%, 17.35%, and
10.74%, respectively. Figure 12 shows the effect of different metastatic organs in a single
organ transfer group on OS in the training group. The prognosis of single metastasis in
different organs is significantly different. Patients with lung metastasis had the best
prognosis, followed by brain metastasis, bone metastasis and liver metastasis which is the
34
worst prognosis, with OS 9.49 months, 7.01 months, 6.18 months and 5.34 months,
respectively (P<0.001).
Figure 12. Comparison of OS between different subgroups of metastatic organ.
Figures 13-15 shows the effects of different treatments on patients' OS. Figure 13 shows
the effects of different treatments on OS in patients with single organ metastasis. Combined
surgical treatment of the primary site and metastatic site can significantly benefit the
survival of patients (P<0.001). Patients with two organ metastasis who received surgical
35
treatment at both the primary site and the metastatic site had the greatest survival benefit,
followed by patients who only received primary lesion surgery or metastatic lesion surgery
(Figure 14). And patients who received no surgical treatment had the worst prognosis
(P<0.001). Figure 15 shows that surgery did not improve outcomes in patients with three
organ metastases (P=0.093).
Figure 13. Comparison of OS between different subgroups of treatment in patients
with single metastatic organ.
36
Figure 14. Comparison of OS between different subgroups of treatment in patients
with two metastatic organs.
37
Figure 15. Comparison of OS between different subgroups of treatment in patients
with three metastatic organs.
Figure 16-21 shows the effects of different treatments on OS in patients with two
metastatic organs. Figure 16 shows that combined surgical treatment at the primary site
and metastatic site can significantly benefit the survival of patients with lung and brain
metastases. Surgery at the primary site alone and surgery at metastatic site alone can also
improve the prognosis. And patients without surgery have the worst OS (P<0.001). Figures
17 to 21 showed that surgery did not improve the prognosis of patients with other two organ
metastases, including patients with lung and bone metastases (P=0.13), patients with lung
and liver metastases (P=0.14), patients with brain and bone metastases (P=0.12), patients
38
with brain and liver metastases (P=0.14), and patients with bone and liver metastases
(P=0.079).
Figure 16. Comparison of OS between different subgroups of treatment in patients
with lung and brain metastasis.
39
Figure 17. Comparison of OS between different subgroups of treatment in patients
with lung and bone metastasis.
40
Figure 18. Comparison of OS between different subgroups of treatment in patients
with lung and liver metastasis.
41
Figure 19. Comparison of OS between different subgroups of treatment in patients
with brain and bone metastasis.
42
Figure 20. Comparison of OS between different subgroups of treatment in patients
with brain and liver metastasis.
43
Figure 21. Comparison of OS between different subgroups of treatment in patients
with bone and liver metastasis.
3.4 Construction and verification of survival prediction model
3.4.1 Construction of Nomogram prediction model
A nomogram of prognostic factors was constructed based on the results of the Cox
regression model. The nomogram chart assigns different scores according to different
regression coefficients of each variable, indicating a different degree of influence on
prognosis. Finally, the predicted survival rate of patients can be obtained by summing the
44
score of each variable. As shown in Figure 22, this Nomogram model can be used to predict
1-year, 3-year, and 5-year overall survival in patients with stage IV NSCLC.
Figure 22. Survival prediction model for stage IV NSCLC patients in the training
group.
3.4.2 Verification of Nomogram prediction model
The prediction ability of the Nomogram model was verified by C-index, ROC curve
and calibration curve. The C-index was calculated as 0.714 (95%CI: 0.710-0.718) in the
training group. The model was applied to the verification group for prediction. We found
that the model still had good prediction ability in the verification group, and the calculated
C-index was 0.712 (95%CI: 0.706-0.718).
The risk scores of each variable in the Cox proportional risk model of the training group
were calculated, and the ROC curve was drawn according to the risk scores. The AUC in
the training group model for 3-year OS and 5-year OS was 0.848 and 0.842, respectively.
45
The AUC for 3-year OS prediction and 5-year OS prediction of the Verification group
model were 0.846 and 0.841, respectively (Figures 23-24), indicating that the model has
good predictive power. And the model can be used as an independent factor to judge the
prognosis of patients with stage Ⅳ NSCLC.
Figure 23. Predicted 3 - and 5-year survival ROC curves in the modeling group.
46
Figure 24. Predicted 3 - and 5-year survival ROC curves in the verification group.
The calibration curve of the Nomogram model was drawn to test consistency (Figures
25-28). The calibration curve was close to 45°, indicating good consistency between the
predicted survival rate and the actual survival rate.
47
Figure 25. Predicted 3-year OS calibration curve in the training group.
48
Figure 26. Predicted 5-year OS calibration curve in the training group.
49
Figure 27. Predicted 3-year OS calibration curve in the verification group.
50
Figure 28. Predicted 5-year OS calibration curve in the verification group.
51
Chapter Four: Discussion
The incidence and mortality of lung cancer rank the first among all malignant tumors.
The onset of NSCLC is occult, and early screening is not fully popularized. More than 60%
of patients have reached the advanced stage when they were diagnosed with lung cancer
(Chen et al., 2016). This situation leads to poor prognosis. For patients with advanced
NSCLC, the current treatment mainly adopts the multidisciplinary treatment model of
systemic therapy (Osmani et al., 2018). For metastatic NSCLC, the 8
th
TNM lung cancer
staging system further divides extrapural metastasis (formerly M1b stage) into two sub-
stages: M1b represents isolated extrapural metastasis in a single organ; M1c indicates
multiple extrapural metastases (Eberhardt et al., 2015). Surgical excision used to be
contraindicated in stage IV NSCLC. However, with the in-depth understanding of lung
cancer and the progress of diagnosis and treatment, the residual lesions and isolated
metastases in patients who have received systematic treatment can benefit from surgery. In
recent years, many studies have also confirmed the survival benefits of surgical resection
in patients with isolated lesions or OM-NSCLC (Casiraghi et al., 2020; Divisi et al., 2018;
Johnson et al., 2016; Novoa et al., 2016).
However, it is still controversial how to select patients with stage IV NSCLC for
surgical treatment. What criteria should be used to select appropriate surgical treatment for
metastatic NSCLC lesions? Do the site and number of metastases, the stage of the primary
tumor (T and N stages), histological type, and molecular characteristics influence the
choice of treatment? Therefore, in order to further clarify the surgical indications of patients
52
with stage IV NSCLC and analyze the prognosis of patients with stage IV NSCLC, we
used the SEER database to collect clinical data of 35350 patients with stage IV NSCLC for
large-scale study.
Multivariable Cox analysis results showed that patients' age at diagnosis, gender, race,
marital status, primary tumor location, histological type of tumor, degree of differentiation
of tumor cells, tumor size, T stage, N stage of lymph node metastasis, number of distant
metastatic organs and treatment were all important factors affecting the OS of patients.
According to the analysis of distant metastatic organs, this study found that patients with
single organ metastasis had more OS than patients with multiple organ metastasis. In
patients with single-organ metastases, lung metastases had the longest OS compared to
other single-organ metastases, suggesting that the type and number of metastatic organs
may affect the prognosis of patients with stage IV NSCLC (P<0.001).
In terms of demographic characteristics, studies have shown that NSCLC is
uncommon in patients under 50 years old (Lara et al., 2014). The incidence of lung cancer
increases with age, which may be related to the aging of the world population. The results
of this study also showed this trend. Age was an independent prognostic factor for patients
with stage IV NSCLC. A study on female cancer showed that female patients with lung
cancer had a better prognosis than male patients (Tanoue, 2021), which was consistent with
the results of Barquin M et al. (Barquin et al., 2020) and this study. Morbidity and mortality
rates are consistently higher in males than in females, and the prognosis for male patients
is significantly worse, possibly because biological differences between the sexes have a
53
significant impact on both disease and treatment outcomes. This study found that
Caucasians accounted for 77.2% of all cases studied. However, the SEER database, which
includes cancer patient information collected from 18 separate cancer registries in the
United States, only represents the overall population characteristics of the United States.
In addition, we found that Asians had higher survival rates than other ethnic groups. A
previous meta-analysis (Klugman et al., 2019) also revealed that Hispanics and Asians had
higher survival rates compared to Caucasians, even after adjusting for clinical factors and
smoking status. Race-related differences are the result of complex interactions between
socioeconomic status, occupational exposure, and lifestyle. The analysis of marital status
showed that married patients had the best prognosis compared with those with other
conditions, which was consistent with the results of a Korean analysis on the influence of
demographic characteristics on the further aggressive treatment of patients with NSCLC
brain metastases (Jung et al., 2019). It might relate to family members could support
patients emotionally and financially which prompting patients more willing to accept
further treatment of their disease.
In terms of clinicopathologic features, it was found that patients with primary NSCLC
located in the main bronchus had the shortest OS and the worst prognosis, which was
consistent with the results of previously published large cohort studies (Li et al., 2019; L.
Yang et al., 2018). The primary location in the main bronchus is an independent risk factor
for the prognosis of NSCLC patients, which should be considered in treatment and
prognosis. Our results also show that the primary location of the tumor, including the main
54
bronchus, upper lobe, middle lobe, and lower lobe, is an important factor affecting survival.
Previous studies have found that different tumor primary sites may lead to different
metastasis modes in stage IV NSCLC patients (Shan et al., 2020). This study did not
involve the correlation between primary sites and metastasis modes which are needed in
further studies. Although the proportion of various histological types of NSCLC varies
from country to country, it is consistent that adenocarcinoma is more common than
squamous carcinoma (Cheng et al., 2016). A study comparing the overall survival rate of
LUAD and LUSC at the same stage showed that the 5-year OS of LUAD at different stages
was higher than that of LUSC (Wang et al., 2020), which was consistent with our results.
Studies have shown (Yasukawa et al., 2018) that the degree of differentiation is an
independent prognostic factor for patients with different stages of NSCLC. The worse the
degree of differentiation, the worse the prognosis of patients. Since the vast majority of
cases included in this study did not register clear histological differentiation grade at the
time of diagnosis, accounting for about 56.5%. The conclusion was only available among
the remaining patients with clear differentiation grade. This study also found that the
prognosis of stage IV NSCLC patients is related to the size of the primary tumor and lymph
node metastasis, which is similar to the previously published study on stage IV breast
cancer (Wang et al., 2019).
It is worth noting that the 8
th
edition of TNM staging updated the M staging, defining
M1b disease with single extrinsic chest metastasis and distinguishing it from M1c with
multiple extrinsic chest metastasis (Park et al., 2019). The single organ metastatic cases in
55
this study can be further divided into group M1a (single metastatic cases in lung) and group
M1b (other single organ metastatic cases including brain, bone, and liver single metastatic
cases). Bone and brain are the two common sites of M1b stage disease metastasis, which
is consistent with previously reported findings (Shin et al., 2017; Tufman et al., 2017). Our
study confirmed the newly proposed survival distinction described by the M stage. The
prognosis of patients with M1c stage (multiple organ metastasis) was significantly lower
than that of patients with other M1a and M1b stages. After adjusting for other clinical
prognostic factors, the newly proposed M stage remained consistent. M1b tends to exhibit
better OS than M1c which indicating it is an independent factor affecting the prognosis of
patients with stage IV NSCLC (Shin et al., 2017). Therefore, it is clinically important to
distinguish M1b from M1c. Statistical analysis in this study showed that the 1-year, 2-year
and 3-year cumulative overall survival rates of all patients were 28.82%, 14.72% and
8.90%, respectively. Patients with single-organ metastasis had a better prognosis, and their
1-year survival rates and median OS were 32.64% and 7.05 months. While patients with
multi-organ metastasis had a poor prognosis. The 1-year survival rate and median OS were
32.64% and 7.05 months, 22.05% and 4.46 months, respectively. Kaplan-Meier survival
analysis showed that patients with lung metastases had the best prognosis and patients with
liver metastasis had the worst prognosis. This is consistent with previous studies on
advanced prostate cancer. Currently, palliative treatment is still the first choice for patients
with stage IV NSCLC. While the treatment is still controversial for patients with
oligometastases. This study found that the combined operation of the primary site and the
56
metastatic site had significant benefits for the survival of patients with single metastasis.
Sun Z et al. (Sun et al., 2019) proved that primary tumor resection was associated with
better survival in patients with single external thoracic metastasis of NSCLC. A
retrospective multi-center study by Fuchs J et al. (Fuchs et al., 2021) also proved that brain
metastatic resection in patients with single thoracic metastasis of NSCLC was safe and
could lead to a good prognosis. Sivasanker M et al. (Sivasanker et al., 2018) 's study on
single brain metastases of different primary tumors also showed that surgical resection of
metastatic lesions could significantly improve the prognosis. The study of Karagkiouzis G
et al. (Karagkiouzis et al., 2017) confirmed that primary pulmonary resection combined
with metastasis resection brought survival benefits to NSCLC patients with single
metastasis in different organs. These studies share the same view as the present study:
primary lesion resection and surgical resection of single metastatic lesions can improve
patient survival. The results of this study showed that receiving surgical resection could
improve the prognosis of patients in the two organ metastasis group, which was in line with
the definition of oligometastases proposed by the European consensus in 2019: a maximum
of five metastatic lesions involving a maximum of three different organs (Dingemans et al.,
2019). In order to further distinguish the impact of surgery on the prognosis of patients
with different organ metastases, survival analysis was conducted according to different
organ metastases. The results showed that surgical resection improved the survival rate of
lung and brain metastases in patients with two organ metastases.
By using the clinical data of 35,350 patients with stage IV NSCLC, the results of Cox
57
proportional risk model were used to construct a prognosis histogram. The nomogram chart
is built on the basis of the multi-factor Cox proportional risk model. According to the
influence of each prediction index in the model, the risk score is quantified and the scores
of multiple indicators are integrated to calculate the predicted value of the prognosis. The
nomogram chart is readable and the results are more intuitive, facilitating patient evaluation.
According to the regression coefficients of variables in the Cox regression model, the
influence of various variables on the prognosis of patients with stage IV NSCLC was
visualized in the nomogram chart, which could predict the 1-year, 3-year and 5-year
survival probabilities of patients. The C-index of the training group was calculated to be
0.714(95%CI:0.710-0.718). The AUC of 3-year OS and 5-year OS were 0.707 and 0.726,
respectively. The C-index (95%CI) of the verification group was 0.712 (95%CI: 0.706-
0.718), and the AUC of the 3-year OS and 5-year OS were 0.714 and 0.747, respectively.
The calibration curves of both the training group and the verification group were close to
45°, which ensured the accuracy and reliability of the prediction model. Clinicians can refer
to the histogram model to assign prognostic factors and predict the survival rate of patients
with stage IV NSCLC.
To sum up, OS differs from multiple metastases in patients with stage IV NSCLC with
single metastases. In general, OS differs from multiple metastases in patients with stage IV
NSCLC with single metastases. The fact that surgical resection is an independent factor for
improving outcomes in patients with stage IV NSCLC with single organ metastases and
two organ metastases of lung and brain. This should be incorporated into clinical treatment
58
decisions in patients with stage IV NSCLC.
Advantages of this study: The objects of this study were a large number of stage IV
NSCLC cases registered in the SEER database. It is a large sample retrospective analysis,
and the results are representative. Clinicians can predict the survival rate of stage IV
NSCLC patients according to the survival prediction model.
This study has the following limitations: (1) The SEER database does not provide data
on patients' comorbidity, such as cardiovascular disease and chronic lung disease. Smoking
history and specific treatment methods, including specific surgical methods, chemotherapy
regimen and cycle, radiotherapy site and dose, were also not provided. (2) Stage IV NSCLC
in this study refers to patients with bone, brain, lung, and liver metastases recorded in the
SEER database. While the common metastatic sites of NSCLC, such as adrenal glands, are
not included in the SEER database. Therefore, this study cannot represent all patients with
stage IV NSCLC. A large number of samples need to be collected clinically for verification.
59
Chapter Five: Conclusion
1. Multivariable Cox analysis showed that age of diagnosis, gender, race, marital status,
primary location of tumor, histological type of tumor, degree of differentiation of tumor
cells, T-stage of tumor size, N-stage of lymph node metastasis, number of distant
metastatic organs and treatment methods were all independent factors affecting
prognosis of patients with stage IV NSCLC.
2. The OS of patients with multi-organ metastasis was significantly lower than that of
patients with single-organ metastasis. The single organ metastasis group could be
divided into M1a (single lung metastasis) and M1b (single bone, brain and liver
metastasis) according to the 8
th
version of TNM staging which the OS of patients in
group M1a was higher than that in group M1b.
3. The 1-year, 3-year, and 5-year survival rates of patients with stage IV NSCLC can be
predicted by the nomogram graph model. The prognosis can be significantly improved
by surgical treatment for patients with single organ metastasis and those with two-organ
metastasis of the lung and brain. Surgical excision was an independent factor for
improvement in patients with oligometastatic stage IV NSCLC.
60
References
Alexander, M., Kim, S. Y ., & Cheng, H. (2020). Update 2020: Management of Non-Small Cell Lung Cancer.
Lung, 198(6), 897-907. https://doi.org/10.1007/s00408-020-00407-5
Arbour, K. C., & Riely, G. J. (2019). Systemic Therapy for Locally Advanced and Metastatic Non-Small Cell
Lung Cancer: A Review. JAMA, 322(8), 764-774. https://doi.org/10.1001/jama.2019.11058
Barquin, M., Calvo, V ., Garcia-Garcia, F., Nunez, B., Sanchez-Herrero, E., Serna-Blasco, R., Auglyt, M.,
Carcereny, E., Rodriguez-Abreu, D., Castro, R. L., Guirado, M., Camps, C., Bosch-Barrera, J.,
Massuti, B., Ortega, A. L., del Barco, E., Gonzalez-Larriba, J. L., Aguiar, D., Garcia-Campelo, R., . . .
Provencio, M. (2020). Sex is a strong prognostic factor in stage IV non-small-cell lung cancer
patients and should be considered in survival rate estimation. Cancer Epidemiology, 67.
https://doi.org/10.1016/j.canep.2020.101737
Blandin Knight, S., Crosbie, P. A., Balata, H., Chudziak, J., Hussell, T., & Dive, C. (2017). Progress and
prospects of early detection in lung cancer. Open Biol, 7(9). https://doi.org/10.1098/rsob.170070
Casiraghi, M., Bertolaccini, L., Sedda, G., Petrella, F., Galetta, D., Guarize, J., Maisonneuve, P., De Marinis,
F., & Spaggiari, L. (2020). Lung cancer surgery in oligometastatic patients: outcome and survival.
Eur J Cardiothorac Surg, 57(6), 1173-1180. https://doi.org/10.1093/ejcts/ezaa005
Chen, W., Zheng, R., Baade, P. D., Zhang, S., Zeng, H., Bray, F., Jemal, A., Yu, X. Q., & He, J. (2016). Cancer
statistics in China, 2015. CA Cancer J Clin, 66(2), 115-132. https://doi.org/10.3322/caac.21338
Cheng, T. Y ., Cramb, S. M., Baade, P. D., Youlden, D. R., Nwogu, C., & Reid, M. E. (2016). The International
Epidemiology of Lung Cancer: Latest Trends, Disparities, and Tumor Characteristics. J Thorac
Oncol, 11(10), 1653-1671. https://doi.org/10.1016/j.jtho.2016.05.021
Congedo, M. T., Cesario, A., Lococo, F., De Waure, C., Apolone, G., Meacci, E., Cavuto, S., & Granone, P.
(2012). Surgery for oligometastatic non-small cell lung cancer: long-term results from a single
center experience. J Thorac Cardiovasc Surg, 144(2), 444-452.
https://doi.org/10.1016/j.jtcvs.2012.05.051
David, E. A., Andersen, S. W., Beckett, L. A., Melnikow, J., Clark, J. M., Brown, L. M., Cooke, D. T., Kelly,
K., & Canter, R. J. (2019). Survival benefits associated with surgery for advanced non-small cell
lung cancer. J Thorac Cardiovasc Surg, 157(4), 1620-1628.
https://doi.org/10.1016/j.jtcvs.2018.10.140
Dingemans, A. C., Hendriks, L. E. L., Berghmans, T., Levy, A., Hasan, B., Faivre-Finn, C., Giaj-Levra, M.,
Giaj-Levra, N., Girard, N., Greillier, L., Lantuejoul, S., Edwards, J., O'Brien, M., Reck, M., Smit,
61
E. F., Van Schil, P., Postmus, P. E., Ramella, S., Lievens, Y ., . . . Novello, S. (2019). Definition of
Synchronous Oligometastatic Non-Small Cell Lung Cancer-A Consensus Report. J Thorac Oncol,
14(12), 2109-2119. https://doi.org/10.1016/j.jtho.2019.07.025
Divisi, D., Barone, M., Zaccagna, G., Gabriele, F., & Crisci, R. (2018). Surgical approach in the
oligometastatic patient. Ann Transl Med, 6(5), 94. https://doi.org/10.21037/atm.2018.01.19
Eberhardt, W. E., Mitchell, A., Crowley, J., Kondo, H., Kim, Y . T., Turrisi, A., 3rd, Goldstraw, P., Rami-Porta,
R., International Association for Study of Lung Cancer, S., Prognostic Factors Committee, A. B. M.,
& Participating, I. (2015). The IASLC Lung Cancer Staging Project: Proposals for the Revision of
the M Descriptors in the Forthcoming Eighth Edition of the TNM Classification of Lung Cancer. J
Thorac Oncol, 10(11), 1515-1522. https://doi.org/10.1097/JTO.0000000000000673
Ettinger, D. S., Wood, D. E., Aisner, D. L., Akerley, W., Bauman, J. R., Bharat, A., Bruno, D. S., Chang, J.
Y ., Chirieac, L. R., D'Amico, T. A., DeCamp, M., Dilling, T. J., Dowell, J., Gettinger, S., Grotz, T.
E., Gubens, M. A., Hegde, A., Lackner, R. P., Lanuti, M., . . . Hughes, M. (2022). Non-Small Cell
Lung Cancer, Version 3.2022, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc
Netw, 20(5), 497-530. https://doi.org/10.6004/jnccn.2022.0025
Fuchs, J., Fruh, M., Papachristofilou, A., Bubendorf, L., Hauptle, P., Jost, L., Zippelius, A., & Rothschild, S.
I. (2021). Resection of isolated brain metastases in non-small cell lung cancer (NSCLC) patients -
evaluation of outcome and prognostic factors: A retrospective multicenter study. PLoS One, 16(6),
e0253601. https://doi.org/10.1371/journal.pone.0253601
Gregor, A., Inage, T., Hwangbo, B., & Yasufuku, K. (2020). Lung cancer staging: State of the art in the era
of ablative therapies and surgical segmentectomy. Respirology, 25(9), 924-932.
https://doi.org/10.1111/resp.13827
Hao, Z. X., Liang, H. R., Zhang, Y . C., Wei, W., Lan, Y . T., Qiu, S. X., Lin, G., Wang, R. C., Liu, Y . L., Chen,
Y . Y ., Huang, J., Wang, W., Cui, F., Goto, T., Jeong, J. Y ., Veronesi, G., Lopez-Pastorini, A., Igai, H.,
Liang, W. H., . . . Liu, J. (2021). Surgery for advanced-stage non-small cell lung cancer: lobectomy
or sub-lobar resection? Translational Lung Cancer Research, 10(3), 1408-1423.
https://doi.org/10.21037/tlcr-21-39
Hellman, S., & Weichselbaum, R. R. (1995). Oligometastases. J Clin Oncol, 13(1), 8-10.
https://doi.org/10.1200/JCO.1995.13.1.8
Herbst, R. S., Morgensztern, D., & Boshoff, C. (2018). The biology and management of non-small cell lung
cancer. Nature, 553(7689), 446-454. https://doi.org/10.1038/nature25183
Hoy, H., Lynch, T., & Beck, M. (2019). Surgical Treatment of Lung Cancer. Crit Care Nurs Clin North Am,
62
31(3), 303-313. https://doi.org/10.1016/j.cnc.2019.05.002
Johnson, K. K., Rosen, J. E., Salazar, M. C., & Boffa, D. J. (2016). Outcomes of a Highly Selective Surgical
Approach to Oligometastatic Lung Cancer. Ann Thorac Surg, 102(4), 1166-1171.
https://doi.org/10.1016/j.athoracsur.2016.04.086
Jung, G., Kim, S. H., Kim, T. G., & Kim, Y . Z. (2019). Demographic and Socioeconomic Factors for
Renouncing Further Active Therapy for Patients with Brain Metastasis of Non-Small Cell Lung
Cancer. Brain Tumor Res Treat, 7(2), 112-121. https://doi.org/10.14791/btrt.2019.7.e35
Karagkiouzis, G., Spartalis, E., Moris, D., Patsouras, D., Athanasiou, A., Karathanasis, I., Verveniotis, A.,
Konstantinou, F., Kouerinis, I. A., Potaris, K., Dimitroulis, D., & Tomos, P. (2017). Surgical
Management of Non-small Cell Lung Cancer with Solitary Hematogenous Metastases. In Vivo,
31(3), 451-454. https://doi.org/10.21873/invivo.11082
Klugman, M., Xue, X., & Hosgood, H. D., 3rd. (2019). Race/ethnicity and lung cancer survival in the United
States: a meta-analysis. Cancer Causes Control, 30(11), 1231-1241.
https://doi.org/10.1007/s10552-019-01229-4
Lara, M. S., Brunson, A., Wun, T., Tomlinson, B., Qi, L., Cress, R., Gandara, D. R., & Kelly, K. (2014).
Predictors of survival for younger patients less than 50 years of age with non-small cell lung cancer
(NSCLC): a California Cancer Registry analysis. Lung Cancer, 85(2), 264-269.
https://doi.org/10.1016/j.lungcan.2014.04.007
Li, C., Liu, J., Lin, J. M., Li, Z. X., Shang, X. L., & Wang, H. Y . (2019). Poor survival of non-small-cell lung
cancer patients with main bronchus tumor: a large population-based study. Future Oncology, 15(24),
2819-2827. https://doi.org/10.2217/fon-2019-0098
Nojiri, T., Hamasaki, T., Inoue, M., Shintani, Y ., Takeuchi, Y ., Maeda, H., & Okumura, M. (2017). Long-
Term Impact of Postoperative Complications on Cancer Recurrence Following Lung Cancer Surgery.
Ann Surg Oncol, 24(4), 1135-1142. https://doi.org/10.1245/s10434-016-5655-8
Novoa, N. M., Varela, G., & Jimenez, M. F. (2016). Surgical management of oligometastatic non-small cell
lung cancer. J Thorac Dis, 8(Suppl 11), S895-S900. https://doi.org/10.21037/jtd.2016.08.13
Osmani, L., Askin, F., Gabrielson, E., & Li, Q. K. (2018). Current WHO guidelines and the critical role of
immunohistochemical markers in the subclassification of non-small cell lung carcinoma (NSCLC):
Moving from targeted therapy to immunotherapy. Semin Cancer Biol, 52(Pt 1), 103-109.
https://doi.org/10.1016/j.semcancer.2017.11.019
Park, H., Dahlberg, S. E., Lydon, C. A., Araki, T., Hatabu, H., Rabin, M. S., Johnson, B. E., & Nishino, M.
63
(2019). M1b Disease in the 8th Edition of TNM Staging of Lung Cancer: Pattern of Single
Extrathoracic Metastasis and Clinical Outcome. Oncologist, 24(8), e749-e754.
https://doi.org/10.1634/theoncologist.2018-0596
Patrini, D., Panagiotopoulos, N., Bedetti, B., Mitsos, S., Crisci, R., Solli, P., Bertolaccini, L., & Scarci, M.
(2018). Surgical approach in oligometastatic non-small cell lung cancer. Ann Transl Med, 6(5), 93.
https://doi.org/10.21037/atm.2018.02.16
Qiu, C., Zhang, S., Jin, H., Zhang, X., & Ye, J. (2020). Radical local treatment for stage IV non-small cell
lung cancer in older adults: a propensity-score matched analysis of the SEER database. Transl
Cancer Res, 9(9), 5336-5349. https://doi.org/10.21037/tcr-19-2796
Shan, Q., Li, Z., Lin, J., Guo, J., Han, X., Song, X., Wang, H., & Wang, Z. (2020). Tumor Primary Location
May Affect Metastasis Pattern for Patients with Stage IV NSCLC: A Population-Based Study. J
Oncol, 2020, 4784701. https://doi.org/10.1155/2020/4784701
Shin, J., Keam, B., Kim, M., Park, Y . S., Kim, T. M., Kim, D. W., Kim, Y . W., & Heo, D. S. (2017). Prognostic
Impact of Newly Proposed M Descriptors in TNM Classification of Non-Small Cell Lung Cancer.
J Thorac Oncol, 12(3), 520-528. https://doi.org/10.1016/j.jtho.2016.11.2216
Siegel, R. L., Miller, K. D., Fuchs, H. E., & Jemal, A. (2021). Cancer Statistics, 2021. CA Cancer J Clin,
71(1), 7-33. https://doi.org/10.3322/caac.21654
Simeone, J. C., Nordstrom, B. L., Patel, K., & Klein, A. B. (2019). Treatment patterns and overall survival
in metastatic non-small-cell lung cancer in a real-world, US setting. Future Oncol, 15(30), 3491-
3502. https://doi.org/10.2217/fon-2019-0348
Sivasanker, M., Madhugiri, V . S., Moiyadi, A. V ., Shetty, P., & Subi, T. S. (2018). Surgery for brain metastases:
An analysis of outcomes and factors affecting survival. Clin Neurol Neurosurg, 168, 153-162.
https://doi.org/10.1016/j.clineuro.2018.03.011
Sun, Z., Sui, X., Yang, F., & Wang, J. (2019). Effects of primary tumor resection on the survival of patients
with stage IV extrathoracic metastatic non-small cell lung cancer: A population-based study. Lung
Cancer, 129, 98-106. https://doi.org/10.1016/j.lungcan.2018.11.012
Takenaka, M., Mori, M., & Tanaka, F. (2021). [The Outcome of Multiple Lung Cancer Cases with Two-stage
Surgery]. Kyobu Geka, 74(1), 54-61. https://www.ncbi.nlm.nih.gov/pubmed/33550320
Tanoue, L. T. (2021). Women and Lung Cancer. Clin Chest Med, 42(3), 467-482.
https://doi.org/10.1016/j.ccm.2021.04.007
64
Tufman, A., Kahnert, K., Kauffmann-Guerrero, D., Manapov, F., Milger, K., Muller-Lisse, U., Winter, H.,
Huber, R. M., & Schneider, C. (2017). Clinical relevance of the M1b and M1c descriptors from the
proposed TNM 8 classification of lung cancer. Strahlenther Onkol, 193(5), 392-401.
https://doi.org/10.1007/s00066-017-1118-9 (Klinische Relevanz der M1b- und M1c-Deskriptoren
der neuen TNM-8-Klassifikation des Lungenkarzinoms.)
Wang, B. Y ., Huang, J. Y ., Chen, H. C., Lin, C. H., Lin, S. H., Hung, W. H., & Cheng, Y . F. (2020). The
comparison between adenocarcinoma and squamous cell carcinoma in lung cancer patients. Journal
of Cancer Research and Clinical Oncology, 146(1), 43-52. https://doi.org/10.1007/s00432-019-
03079-8
Wang, R., Zhu, Y ., Liu, X., Liao, X., He, J., & Niu, L. (2019). The Clinicopathological features and survival
outcomes of patients with different metastatic sites in stage IV breast cancer. BMC Cancer, 19(1),
1091. https://doi.org/10.1186/s12885-019-6311-z
Yang, C. F. J., Gu, L., Shah, S. A., Yerokun, B. A., D'Amico, T. A., Hartwig, M. G., & Berry, M. F. (2018).
Long-term outcomes of surgical resection for stage IV non-small-cell lung cancer: A national
analysis. Lung Cancer, 115, 75-83. https://doi.org/10.1016/j.lungcan.2017.11.021
Yang, L., Wang, S., Gerber, D. E., Zhou, Y ., Xu, F., Liu, J., Liang, H., Xiao, G., Zhou, Q., Gazdar, A., & Xie,
Y . (2018). Main bronchus location is a predictor for metastasis and prognosis in lung
adenocarcinoma: A large cohort analysis. Lung Cancer, 120, 22-26.
https://doi.org/10.1016/j.lungcan.2018.03.011
Yasukawa, M., Sawabata, N., Kawaguchi, T., Kawai, N., Nakai, T., Ohbayashi, C., & Taniguchi, S. (2018).
Histological Grade: Analysis of Prognosis of Non-small Cell Lung Cancer After Complete
Resection. In Vivo, 32(6), 1505-1512. https://doi.org/10.21873/invivo.11407
Zhang, C., Wang, L., Li, W., Huang, Z., Liu, W., Bao, P., Lai, Y ., Han, Y ., Li, X., & Zhao, J. (2019). Surgical
outcomes of stage IV non-small cell lung cancer: a single-center experience. J Thorac Dis, 11(12),
5463-5473. https://doi.org/10.21037/jtd.2019.11.30
Abstract (if available)
Abstract
Background: Non-small cell lung cancer (NSCLC) is a major health burden on global health. The surgical indications of patients with stage IV NSCLC were investigated using case data extracted from the SEER database.
Methods: Data for patients diagnosed with stage Ⅳ NSCLC between 2010 and 2015 were extracted from the SEER database. Training set and validation set were divided at a ratio of 7:3. Cox univariable and multivariable regression analysis were performed on the training group to identify independent prognostic factors.
Results: The median survival time of all subjects was 6.01 months. Multivariable Cox analysis in the training group showed that age of diagnosis, gender, race, marital status, primary tumor location, histological type of tumor, degree of differentiation of tumor cells, tumor size, T-stage, N stage of lymph node metastasis, number of distant metastatic organs and treatment were all independent factors affecting prognosis of stage IV NSCLC patients. Survival analysis showed that surgical treatment was an independent prognostic factor for stage IV NSCLC patients with single organ metastases and lung and brain metastases.
Conclusion: The OS of patients with multiple organ metastasis was significantly lower than that of patients with single organ metastasis. The OS of patients in group M1a was higher than that in group M1b. Surgical treatment of patients with single organ metastasis and those with two-organ metastasis of the lung and brain can significantly improve the prognosis. Surgical excision was an independent factor for improvement in patients with oligometastatic stage IV NSCLC.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
An analysis of disease-free survival and overall survival in inflammatory breast cancer
PDF
Predictive factors of breast cancer survival: a population-based study
PDF
The risk estimates of pneumoconiosis and its relevant complications: a systematic review and meta-analysis
PDF
A novel risk-based treatment strategy evaluated in pediatric head and neck non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) patients: a survival analysis from the Children's Oncology Group study...
PDF
Lymph node metastases in non-rhabdomyosarcoma soft tissue sarcoma
PDF
Analysis of factors associated with breast cancer using machine learning techniques
PDF
An assessment of necrosis grading in childhood osteosarcoma: the effect of initial treatment on prognostic significance
PDF
Survival of children and adolescents with low-risk non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) treated with surgery only: an analysis of 234 patients from the Children’s Oncology Group stud...
PDF
The impact of the COVID-19 pandemic on cancer care delivery
PDF
Predicting ototoxicity evaluated by SIOP in children receiving cisplatin
PDF
Need for tissue plasminogen activator for central venous catheter malfunction and its association with occurrence of vVenous thromboembolism
PDF
Risk factors and survival outcome in childhood alveolar soft part sarcoma among patients in the Children’s Oncology Group (COG) Phase 3 study ARST0332
PDF
Incidence and survival rates of the three major histologies of renal cell carcinoma
PDF
Instability of heart rate and rating of perceived exertion during high-intensity interval training in breast cancer patients undergoing anthracycline chemotherapy
PDF
Origins of the gender disparity in bladder cancer risk: a SEER analysis
PDF
The interplay between tobacco exposure and polygenic risk score for growth on birthweight and childhood acute lymphoblastic leukemia
PDF
HIF-1α gene polymorphisms and risk of severe-spectrum hypertensive disorders of pregnancy: a pilot triad-based case-control study
PDF
Predicting hospital length of stay (LOS) using the National Inpatient Sample
PDF
The effect of renal function on toxicity of E7389 (eribulin) among patients with bladder cancer
PDF
Disparities in gallbladder, intra-hepatic bile duct, and other biliary cancers among multi-ethnic populations: a California Cancer Registry study
Asset Metadata
Creator
Liang, Chen
(author)
Core Title
Construction of a surgical survival prediction model of stage IV NSCLC patients-based on seer database
School
Keck School of Medicine
Degree
Master of Science
Degree Program
Applied Biostatistics and Epidemiology
Degree Conferral Date
2023-05
Publication Date
06/12/2023
Defense Date
05/09/2023
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
Construction,OAI-PMH Harvest,prediction model,seer,stage IV,Survival
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Li, Ming (
committee chair
), Alonzo, Todd (
committee member
), Piao, Jin (
committee member
)
Creator Email
cliang27@usc.edu,rorylc88@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC113170479
Unique identifier
UC113170479
Identifier
etd-LiangChen-11949.pdf (filename)
Legacy Identifier
etd-LiangChen-11949
Document Type
Thesis
Format
theses (aat)
Rights
Liang, Chen
Internet Media Type
application/pdf
Type
texts
Source
20230613-usctheses-batch-1055
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
prediction model
seer
stage IV