University of Southern California Dissertations and Theses

Association of comorbidity with prostate cancer tumor characteristics in African American men

(USC Thesis Other)

Association of comorbidity with prostate cancer tumor characteristics in African American men

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Association of comorbidity with Prostate cancer tumor characteristics in African
American men

by

Weijia Hu

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

December 2021

Copyright 2021 Weijia Hu

ii

Dedication
I dedicate this study to my parents, Zhiyong Hu and Wei Liu, who have always loved me
unconditionally and encouraged me to believe in myself.
I also dedicate this study to the memory of my maternal grandfather, Wenguang Liu, for being
my first teacher. Although he was my inspiration to pursue my dreams, he was unable to see my
graduation. This is for him.

iii

Acknowledgement
I would first thank my mentor, Dr. David Conti for support of my master thesis. As a student
who just finished the first year of my graduate program and had no experience in biostatistics,
Dr. David Conti gave me an opportunity to work on the RESPOND project. He taught me
practical knowledge about data analysis. Without his guidance and generous help, this study
would not have been possible.
My sincere thanks also go to my master thesis committee, Dr. Ann Hamilton and Dr. Christopher
Haiman. They gave me the inspiration for this study and always supported me.
I would also like to thank Rich Pinder. He helped me understand the NACCR data dictionary and
provided updated data for my analysis. I thank Prof. Jun Wang for her help with my analysis.
I thank Dr. Meredith Franklin and Dr. Mariana Stern for always providing me academic
instructions and helping me finish my master’s program. I also thank my advisor, Renee Stanley
for her help and patience.
Last, I am grateful to my parents, Zhiyong Hu and Wei Liu, for their love and caring. During this
difficult pandemic time, they gave me endless encouragement and support even I was far away
from home. Thank you.

iv

Table of Contents
Dedication ..................................................................................................................................... ii
Acknowledgements ..................................................................................................................... iii
List of Tables ................................................................................................................................ v
List of Figures .............................................................................................................................. vi
Abstract ...................................................................................................................................... vii
Introduction .................................................................................................................................. 1
Methods ......................................................................................................................................... 3
Results ........................................................................................................................................... 6
Discussion ................................................................................................................................... 16
Supplement ................................................................................................................................. 18
Reference .................................................................................................................................... 20
Appendix ..................................................................................................................................... 24
v

List of Tables
Table 1:
a. Descriptive statistics by PSA levels groups ....................................................................... 8
b. Descriptive statistics by clinical stage groups ................................................................... 9
c. Descriptive statistics by Gleason’s score groups ............................................................. 10
d. Descriptive statistics by aggressiveness groups ……………...…………….…………... 11
Table 2:
a. The results of multiple linear regression analysis for PSA level ………………...…...... 13
b. The results of logistic regression analysis for clinical stage ...………….………...…..... 13
c. The results of logistic regression analysis for Gleason score …………….…………..... 14
d. The results of multinomial logistic regression analysis for aggressiveness ………….... 15
Appendix Table 1. Comorbid condition distribution …………………………...………..…… 25

vi

List of Figures
Figure 1. The histogram of PSA levels ......................................................................................... 6
Appendix Figure 1. Examples of R Markdown outputs ............................................................ 24

vii

Abstract
Prostate cancer (PCa) has a high incidence rate and mortality among men in the United States,
especially among African American (AA) men. Several studies have investigated the potential
explanations for the disparity of PCa in the AA men population including socioeconomic status,
the clinical stages at diagnosis, quality of care they received and genetic susceptibility.
Additionally, there is evidence showing that AA men are more likely to have comorbidity
conditions such as cardiovascular disease. Here, we assess the effect of comorbidity on PCa
tumor characteristics, including prostate-specific antigen (PSA) levels, clinical stage, Gleason
score, and tumor aggressiveness, among 3265 AA men from the RESPOND project. The tumor
aggressiveness categories (low, intermediate, and high) are generated by the other three
characteristics based on the D'Amico criteria. The comorbidity score is generated by using the
modified Charlson comorbidity index. Descriptive statistics tables are used to describe the
distribution of each PCa characteristic stratified by different comorbidity groups. The linear
regression model is used to evaluate the association between comorbidity and PSA levels. For
the effects of comorbidity on clinical stage and Gleason score, we use multiple logistic
regression models. We use a multinomial model to analyze three levels of aggressiveness, low,
intermediate, and high. We also adjust for lifestyle factors based on the above models. There is a
significantly negative association between PSA level and comorbidity after adjusting for lifestyle
(p = 0.034). Other associations between comorbidity score and clinical stage, Gleason score and
aggressiveness are all negative, albeit not significant. Individuals with more comorbid conditions
were less likely to have high values of PCa characteristics compared to those with no
comorbidities. This is an interim analysis, and the sample population may not be representative.
Also, selection, screening and survival bias are possibly introduced in our study.
1

Introduction
Prostate cancer (PCa) is the most common cancer diagnosed among males and the second
common cause of male deaths from cancer in the United States in 2021, which accounts for 26%
of all new cancer cases and 11% of cancer-related deaths
1
. African American (AA) men are more
likely to develop and die of PCa than men in other races. The risk of developing PCa is 1.4 times
higher and the risk of dying from PCa is two to three times higher in AA males compared to
Caucasian males
2, 3
. This disparity in the AA population is a leading health concern, however,
the reasons for it are unclear
4
. Some potential factors are socioeconomic status
5
, the clinical
stages at diagnosis
6
, quality of care they received
7
and genetic susceptibility
8
.
In this study, we have assessed the association of comorbidity with PCa tumor characteristics
that have been used to create a measure of tumor aggressiveness. Comorbid conditions can affect
PCa patients’ health, survival rates, and treatment decisions
9
. AA males are more likely to have
comorbid conditions such as cardiovascular disease
10
, which is negatively associated with health
outcomes
3
. However, the reasons why comorbidity may be related to PCa tumor characteristics
remain unclear. It’s essential to understand the distribution of comorbid conditions and the
association between them and PCa characteristics, such as prostate-specific antigen (PSA) level,
clinical stage, and Gleason score, among the AA population. This understanding could
potentially assist us in reducing PCa disparities.
Our study utilized data from the Research on PCa in Men of African Ancestry: Defining the
Roles of Genetics, Tumor Markers and Social Stress (RESPOND) study
11
. The RESPOND study
is a nation-wide study on AA men diagnosed with PCa and is aimed at figuring out potential
factors associated with aggressive PCa among AA men. The data analyzed in this report is based
2

on two sources cancer registry data on tumor characteristics and self-report data on comorbidities
and lifestyle behaviors from a survey completed by participants. For this investigation, tumor
characteristics were obtained from the cancer registry dataset and the assessment of comorbidity
utilized 17 items pertaining to comorbid conditions from the survey questionnaires plus the
lifestyle variables on smoking, alcohol use, and obesity.

3

Methods
Our study involved a population-based cohort of AA men diagnosed with PCa in the United
States on January 1, 2010, or later, from different cancer registries in the United States including
registries of the National Cancer Institute Surveillance, Epidemiology and End Results (SEER)
program and the National Program of Cancer Registries (NPCR). Clinical information was
collected, including tumor characteristics, date of birth, and age at diagnosis. Comorbid
conditions, body mass index (BMI) and lifestyle information, including smoking history and
drinking history, were collected in the survey questionnaire. This report is based 3265
participants who completed the baseline questionnaire and were linked to cancer registry data
within 7 registry sites: Greater California; Georgia; Northern California; Louisiana; Detroit;
Michigan; and Los Angeles County.
We assessed participants’ tumor characteristics using four characteristics: PSA levels, clinical
stage, Gleason scores, and Aggressiveness, which was based on a combination of the first three
characteristics.
The first three were directly collected from the cancer registry dataset, in which we used the
variable of “CS Site-Specific Factor 1” for PSA level, the variable of “TNM CLIN T” for
clinical stage, and the variable “CS Site-Specific Factor 8” for Gleason score
12
.
We generated an aggressiveness risk variable using the first three characteristics in order to
categorize risk level of tumor based on the D'Amico criteria, which is widely used currently
13
.
The three categories are defined as: low risk (PSA < 10, Clinical Stage ≤ T2a, and Gleason score
≤ 6), intermediate risk (10 ≤ PSA ≤ 20, Clinical Stage = T2b or Gleason score = 7), and high risk
(PSA > 20, Clinical Stage ≥ T2c or Gleason score ≥ 8).
4

We generated a variable called comorbidity score to measure comorbidity. The commonly used
method to assess comorbidity is the Charlson comorbidity index
14
. We adjusted the Charlson
comorbidity index by using a total of 17 comorbid conditions to calculate a comorbidity score
for each participant. These comorbid conditions included heart attack, congestive heart attack,
stroke, hypertension, peripheral arterial disease, high cholesterol, asthma, stomach ulcers,
Crohn’s disease, diabetes, kidney problems, cirrhosis, arthritis, Alzheimer’s disease, depression,
AIDS, and other cancer (besides PCa). The information was collected in the survey by
participants answering “yes” or “no” for each corresponding comorbid condition. Each “yes”
adds one to the comorbidity scores, while a “no” or missing answer adds zero to the comorbidity
scores. The comorbidity score ranged from 0 to 17. We categorized the comorbidity score into
five groups: comorbidity score of 0, 1, 2, 3, and 4 or higher.
We stratified on five variables to investigate effect heterogeneity: age at diagnosis and current
age, study site, BMI, and lifestyle (smoking and alcohol use). The age variables, including
participants’ current age and age at diagnosis, were calculated based on the date of birth in the
cancer registry dataset. Also, the study site variable was directly obtained from the cancer
registry dataset. In the current data set, there were 7 sites, Greater CA, Northern CA, Georgia,
Louisiana, Detroit, Michigan, and Los Angeles County. BMI values were calculated by weight
and height, which were collected from the survey questionnaire, using the following formula:
𝐵𝑀𝐼 =
𝑤 𝑒𝑖𝑔 ℎ 𝑡 ( 𝑙 𝑏 𝑠 )
[ ℎ 𝑒𝑖𝑔 ℎ 𝑡 ( 𝑖𝑛 ) ]
2
× 703
We classified BMI into four groups, normal (BMI < 25), overweight (25 ≤ BMI < 30), and
obesity (BMI ≥ 30)
15
. The lifestyle variables, including smoking history and drinking habits,
were collected in the survey questionnaire. Participants were classified into “smokers” if they
5

answered “yes” to the question “Have you ever smoked at least 100 cigarettes during the
lifetime” and into “nonsmoker” if they answered “no” to this question. To capture drinking
habits, there are 3 levels, “heavy”, “modest”, and “light or none” and they were determined by
the question “In the past month, how often did you have at least one drink of alcoholic
beverage”. Participants were categorized as “heavy” drinkers with answers of “everyday” and
“5-6 times per week”; “modest” drinkers with answers of “3-4 times per week” and “1-2 times
per week”; and in the “light or none” group with answers of “fewer than once per week” and
“did not drink”.
First, descriptive statistics tables were generated for each tumor characteristic. PSA levels were
classified into three categories: PSA < 10; 10 ≤ PSA ≤ 20; and PSA > 20. Clinical stage was
classified into four categories: T0 or TX; T1; T2 and T3. Gleason score was classified into three
categories: Gleason score ≤ 6; Gleason score = 7; Gleason score ≥ 8.
We developed a basic regression model for each PCa tumor characteristic as the outcome
variable. All regression models were adjusted by registry site, age at diagnosis, and BMI. We
treated PSA level as a numeric variable and performed linear regression models on the log scale
to assess the association between comorbidity score and PSA level. For other categorical tumor
characteristic outcome variables, including clinical stage and Gleason score, since they had
multiple categories, we first dichotomized them and conducted logistic regression to evaluate the
association of comorbidity. Clinical stage was dichotomized between stage ≤ T2c vs. > T2c.
Gleason score was dichotomized between ≤ 7 and > 7. Finally, we used a multinomial model to
analyze factors associated with three levels of aggressiveness. Next, adjusting for smoking
history and drinking habit was conducted on mentioned basic models.

6

Results
Descriptive statistics for the study population are shown in Table 1 including four tables
corresponding to PCa tumor characteristics. 19.9% of participants had a comorbidity score of 3
and 23.6% of participants had a comorbidity score of 4 or higher. The median current age at
survey response was 69 years old and the median age at diagnosis was 63 years old. For BMI,
the percentages of participants in overweight and obesity groups were 34.7% and 31.5%
respectively. 47% of participants, whose smoking history records were not missing, were
smokers (smoked at least 100 cigarettes in their lifetime). As for drinking habits, most
participants (63.1%) were light drinkers or did not drink. Only 8.6% of participants were heavy
drinkers.

Figure 1. The histogram of PSA levels
PSA levels were categorized into three categories (PSA levels < 10, 10 ≤ PSA levels ≤ 20, and
PSA levels > 20). Figure 1 shows the skewed distribution of PSA levels. As shown in the Table
1.a, 2027 of 3265 (62.1%) participants had PSA levels less than 10 ng/ml. Across all PSA levels
categories, the distributions of percentages of men in each comorbidity score group had similar
7

patterns. The median age at diagnosis was similar among three PSA levels groups. The only
variable significantly associated with PSA levels was BMI. The proportion of those who were
overweight or obese decreased with the increase of PSA level, 36.2%, 32.1%, and 30.1%; 33.1%,
31.0%, and 25.8%; respectively.
Clinical stage was categorized into four categories (T0 or TX, T1, T2, and T3). Table 1.b shows
that most participants (66%) were at stage T1. There were no significant associations between
comorbidity group, BMI groups and the lifestyle variables and clinical stage. However, the
median current age and age at diagnosis in T3 stage (71 years old and 68 years old) was greater
than that in the sample population (69 years old and 63 years old).
Gleason score was classified into three categories < 7, = 7, and> 7. Similar to clinical stage, the
only variable associated with Gleason score categories was age. There was a slightly increasing
trend in current age and age at diagnosis with increasing of Gleason score (Table 1.c).
Table 1.d indicated that 1429 of 3265 (43.8%) participants had intermediate risk of PCa in this
sample. The only variables associated with aggressiveness were age and drinking habits. Slightly
older age at diagnosis was associated with higher risk diseases, while heavier drinking was
associated with lower risk disease.
8

Table 1.
a. Descriptive statistics by PSA level groups
Variables
a
PSA < 10
(N=2027)
10 ≤ PSA ≤ 20
(N=533)
PSA > 20
(N=365)
Overall
(N=3265)
P-value
b
Comorbidity score

0.300
0 214 (10.6%) 60 (11.3%) 56 (15.3%) 382 (11.7%)

1 407 (20.1%) 103 (19.3%) 83 (22.7%) 657 (20.1%)

2 493 (24.3%) 147 (27.6%) 86 (23.6%) 807 (24.7%)

3 420 (20.7%) 97 (18.2%) 65 (17.8%) 649 (19.9%)

4 or more 493 (24.3%) 126 (23.6%) 75 (20.5%) 770 (23.6%)
Current age

68.0 (6.78) 68.9 (6.85) 68.2 (6.65) 68.2 (6.74) 0.061
Age at diagnosis

62.3 (6.79) 63.2 (6.92) 62.5 (6.58) 62.6 (6.76) 0.056
BMI

0.023
Normal 312 (15.4%) 98 (18.4%) 76 (20.8%) 542 (16.6%)
Overweight 733 (36.2%) 171 (32.1%) 110 (30.1%) 1134 (34.7%)
Obesity 670 (33.1%) 165 (31.0%) 94 (25.8%) 1027 (31.5%)
Missing 312 (15.4%) 99 (18.6%) 85 (23.3%) 562 (17.2%)

Smoking history

0.390
Nonsmoker 1019 (50.3%) 258 (48.4%) 167 (45.8%) 1627 (49.8%)

Smoker 885 (43.7%) 253 (47.5%) 170 (46.6%) 1438 (44.0%)

Missing 123 (6.1%) 22 (4.1%) 28 (7.7%) 200 (6.1%)

Drinking habit

0.183
Light or none 1240 (61.2%) 342 (64.2%) 249 (68.2%) 2061 (63.1%)

Modest 537 (26.5%) 138 (25.9%) 77 (21.1%) 834 (25.5%)

Heavy 194 (9.6%) 42 (7.9%) 28 (7.7%) 281 (8.6%)

Missing 56 (2.8%) 11 (2.1%) 11 (3.0%) 89 (2.7%)

a. Values are mean (SD) for continuous variables and frequency (percentage) for categorical variables.
b. P values for continuous variables are based on ANOVA tests. P values for categorical variables are based on
Pearson’s chi-square tests.

9

b. Descriptive statistics by clinical stage groups
Variables
a

T0 or TX
(N=102)
T1
(N=2155)
T2
(N=480)
T3
(N=92)
Overall
(N=3265)
P-value
b

Comorbidity score

0.504
0 16 (15.7%) 243 (11.3%) 55 (11.5%) 14 (15.2%) 382 (11.7%)

1 24 (23.5%) 427 (19.8%) 93 (19.4%) 19 (20.7%) 657 (20.1%)

2 14 (13.7%) 563 (26.1%)
107
(22.3%)
18 (19.6%) 807 (24.7%)

3 24 (23.5%) 414 (19.2%)
107
(22.3%)
20 (21.7%) 649 (19.9%)

4 or more 24 (23.5%) 508 (23.6%)
118
(24.6%)
21 (22.8%) 770 (23.6%)
Current age

67.6 (6.48)
68.0
(6.73)
68.7 (6.98) 70.4 (6.20) 68.2 (6.74) 0.007
Age at diagnosis

62.2 (6.56)
62.4
(6.72)
63.1 (7.00) 64.8 (6.06) 62.6 (6.76) 0.007
BMI

0.509
Normal 25 (24.5%) 341 (15.8%) 87 (18.1%) 13 (14.1%) 542 (16.6%)

Overweight 30 (29.4%) 756 (35.1%)
163
(34.0%)
28 (30.4%)
1134
(34.7%)

Obesity 32 (31.4%) 690 (32.0%)
145
(30.2%)
31 (33.7%)
1027
(31.5%)

Missing 15 (14.7%) 368 (17.1%) 85 (17.7%) 20 (21.7%) 562 (17.2%)

Smoking history

Nonsmoker 59 (57.8%)
1073
(49.8%)
228
(47.5%)
40 (43.5%)
1627
(49.8%)

Smoker 36 (35.3%) 951 (44.1%)
227
(47.3%)
44 (47.8%)
1438
(44.0%)
0.227
Missing 7 (6.9%) 131 (6.1%) 25 (5.2%) 8 (8.7%) 200 (6.1%)

Drinking habit

0.851
Light or none 65 (63.7%)
1361
(63.2%)
300
(62.5%)
61 (66.3%)
2061
(63.1%)

Modest 22 (21.6%) 542 (25.2%)
132
(27.5%)
19 (20.7%) 834 (25.5%)

Heavy 11 (10.8%) 195 (9.0%) 38 (7.9%) 6 (6.5%) 281 (8.6%)

Missing 4 (3.9%) 57 (2.6%) 10 (2.1%) 6 (6.5%) 89 (2.7%)

a. Values are mean (SD) for continuous variables and frequency (percentage) for categorical variables.
b. P values for continuous variables are based on ANOVA tests. P values for categorical variables are based on
Pearson’s chi-square tests.

10

c. Descriptive statistics by Gleason score groups
Variables
a

< 7
(N=984)
= 7
(N=1512)
> 7
(N=699)
Overall
(N=3265)
P-value
b

Comorbidity score

0.852
0 120 (12.2%) 170 (11.2%) 82 (11.7%) 382 (11.7%)

1 209 (21.2%) 291 (19.2%) 141 (20.2%) 657 (20.1%)

2 235 (23.9%) 376 (24.9%) 184 (26.3%) 807 (24.7%)

3 205 (20.8%) 309 (20.4%) 122 (17.5%) 649 (19.9%)

4 or more 215 (21.8%) 366 (24.2%) 170 (24.3%) 770 (23.6%)
Current age

67.6 (6.61) 68.1 (6.86) 69.2 (6.55) 68.2 (6.74) <0.001
Age at diagnosis

61.9 (6.66) 62.4 (6.85) 63.6 (6.52) 62.6 (6.76) <0.001
BMI

0.738
Normal 175 (17.8%) 235 (15.5%) 117 (16.7%) 542 (16.6%)

Overweight 340 (34.6%) 538 (35.6%) 231 (33.0%) 1134 (34.7%)

Obesity 295 (30.0%) 489 (32.3%) 221 (31.6%) 1027 (31.5%)

Missing 174 (17.7%) 250 (16.5%) 130 (18.6%) 562 (17.2%)

Smoking History

Nonsmoker 509 (51.7%) 737 (48.7%) 353 (50.5%) 1627 (49.8%)

Smoker 414 (42.1%) 681 (45.0%) 309 (44.2%) 1438 (44.0%) 0.517
Missing 61 (6.2%) 94 (6.2%) 37 (5.3%) 200 (6.1%)

Drinking habit

Light or none 590 (60.0%) 952 (63.0%) 470 (67.2%) 2061 (63.1%)

Modest 279 (28.4%) 381 (25.2%) 156 (22.3%) 834 (25.5%)

Heavy 89 (9.0%) 135 (8.9%) 56 (8.0%) 281 (8.6%) 0.124
Missing 26 (2.6%) 44 (2.9%) 17 (2.4%) 89 (2.7%)

a. Values are mean (SD) for continuous variables and frequency (percentage) for categorical variables.
b. P values for continuous variables are based on ANOVA tests. P values for categorical variables are based on
Pearson’s chi-square tests.

11

d. Descriptive statistics by aggressiveness groups
a

Variables
b
Low risk
(N=800)
Intermediate risk
(N=1429)
High risk
(N=977)
Overall
(N=3265)
P-value
c
Comorbidity score

0.915
0 94 (11.8%) 156 (10.9%) 125 (12.8%) 382 (11.7%)

1 169 (21.1%) 274 (19.2%) 201 (20.6%) 657 (20.1%)

2 192 (24.0%) 360 (25.2%) 245 (25.1%) 807 (24.7%)

3 168 (21.0%) 287 (20.1%) 183 (18.7%) 649 (19.9%)

4 or more 177 (22.1%) 352 (24.6%) 223 (22.8%) 770 (23.6%)
Current age

67.5 (6.66) 68.2 (6.87) 68.8 (6.54) 68.2 (6.74) <0.001
Age at diagnosis

61.8 (6.70) 62.5 (6.88) 63.2 (6.51) 62.6 (6.76) <0.001
BMI

0.252
Normal 143 (17.9%) 216 (15.1%) 172 (17.6%) 542 (16.6%)

Overweight 279 (34.9%) 527 (36.9%) 306 (31.3%) 1134 (34.7%)

Obesity 246 (30.8%) 453 (31.7%) 308 (31.5%) 1027 (31.5%)

Missing 132 (16.5%) 233 (16.3%) 191 (19.5%) 562 (17.2%)

Smoking History

Nonsmoker 421 (52.6%) 702 (49.1%) 478 (48.9%) 1627 (49.8%)

Smoker 329 (41.1%) 639 (44.7%) 444 (45.4%) 1438 (44.0%) 0.301
Missing 50 (6.3%) 88 (6.2%) 55 (5.6%) 200 (6.1%)

Drinking

0.039
Light or none 470 (58.8%) 897 (62.8%) 655 (67.0%) 2061 (63.1%)

Modest 232 (29.0%) 363 (25.4%) 221 (22.6%) 834 (25.5%)

Heavy 76 (9.5%) 127 (8.9%) 77 (7.9%) 281 (8.6%)

Missing 22 (2.8%) 42 (2.9%) 24 (2.5%) 89 (2.7%)

a. Aggressiveness was categorized as low (PSA < 10, Clinical Stage ≤ T2a, and Gleason score ≤ 6), intermediate (10
≤ PSA ≤ 20, Clinical Stage = T2b or Gleason score = 7), and high (PSA > 20, Clinical Stage ≥ T2c or Gleason
score ≥ 8).
b. Values are mean (SD) for continuous variables and frequency (percentage) for categorical variables.
c. P values for continuous variables are based on ANOVA tests. P values for categorical variables are based on
Pearson’s chi-square tests.

Results from the regression analysis for each PCa tumor characteristic are presented in Table 2.
The association between comorbidity score categories and PSA levels was investigated using
linear regression. Comorbidity score was negatively associated with PSA levels (p = 0.009) after
12

adjustment for other variables. We used the following formula to calculate the change in
coefficient between two models:
𝛽 ( 𝑈 𝑛𝑎𝑑 𝑗𝑢𝑠 𝑡 𝑒𝑑 ) − 𝛽 ( 𝐴 𝑑 𝑗𝑢𝑠 𝑡 𝑒𝑑 )
𝛽 ( 𝑈 𝑛𝑎𝑑 𝑗𝑢𝑠 𝑡 𝑒𝑑 )
∙ 100
Men with a single comorbid condition had a 30% increase in PSA levels compared to no
conditions after adjustment for lifestyle characteristics. Table 2.a indicated that men with a
comorbidity score of 4 or more had a significant 14.4% decrease on the mean PSA levels
compared to men with comorbidity scores of 0 after adjusting for registry site, age, BMI, and
lifestyle (p = 0.013). Interestingly, lifestyle characteristics (smoking history and drinking habit)
did not have a substantial impact as potential confounders. The overall regression was significant
after adjusting for registry site, age at diagnosis and BMI (p = 0.006).
Logistic regression was used when assessing the association of comorbidity scores on stage and
Gleason score. The coefficients in these models indicated that having one or more comorbidities
leads to lower stage and Gleason scores. However, those results were not statistically significant.
Multinomial regression was utilized to evaluate the relationship between comorbidity scores and
PCa aggressiveness, since there were three levels for aggressiveness. Similar to the above results
there was a negative relationship between comorbidity and PCa aggressiveness, although the
tests of the corresponding terms were not statistically significant.

13

Table 2.
a. The results of multiple linear regression analysis for PSA level
a

Model 1
b
Model 2
c

Variables Log Beta
d
Beta
e
P-value Log Beta
d
Beta
e
P-value
Comorbidity score 1 -0.046 0.955 0.445 -0.032 0.969 0.612
2 -0.097 0.908 0.102 -0.102 0.903 0.094
3 -0.145 0.865 0.019 -0.141 0.868 0.027
4 or more -0.169 0.845 0.005 -0.155 0.856 0.013
Registry site Georgia -0.006 0.994 0.914 -0.017 0.983 0.777
Northern CA 0.007 1.007 0.931 0.015 1.015 0.858
Louisiana -0.032 0.969 0.621 -0.047 0.954 0.481
Detroit -0.057 0.945 0.504 -0.089 0.915 0.309
Michigan 0.078 1.081 0.524 0.106 1.112 0.413
LA County 0.069 1.071 0.337 0.056 1.058 0.446
Age at diagnosis 0.005 1.005 0.043 0.004 1.004 0.146
BMI Overweight -0.127 0.881 0.004 -0.102 0.903 0.026
Obesity -0.118 0.889 0.009 -0.107 0.899 0.022
Smoking history Smoker − − − 0.048 1.049 0.162
Drinking habit Modest − − − -0.109 0.897 0.005
Heavy − − − -0.086 0.918 0.138

Model 1
F-statistic: 2.268
p-value: 0.006
R
2
: 0.012
Adjusted R
2
: 0.007
Model 2
F-statistic: 2.212
p-value: 0.004
R
2
: 0.015
Adjusted R
2
: 0.008
a. The dependent variable was PSA levels with log transformation.
b. Model 1 was adjusted by Registry site, age at diagnosis, and BMI.
c. Model 2 was additionally adjusted by smoking history and drinking habit based on Model 1.
d. Values were on the log scale.
e. Values were exponentiated.

b. The results of logistic regression analysis for clinical stage
Model 1
a
Model 2
b
Variables OR
c
95% CI
d
P-value OR
c
95% CI
d
P-value
Comorbidity
score
1 0.56 (0.26, 1.26) 0.168 0.62 (0.28, 1.39) 0.246
2 0.49 (0.23, 1.06) 0.070 0.50 (0.23, 1.12) 0.092
3 0.48 (0.21, 1.09) 0.081 0.51 (0.22, 1.17) 0.113
4 or more 0.49 (0.23, 1.08) 0.077 0.50 (0.22, 1.12) 0.091
Registry site Georgia 1.16 (0.45, 3.00) 0.763 1.11 (0.43, 2.89) 0.828
Northern CA 1.49 (0.44, 5.03) 0.519 1.52 (0.45, 5.17) 0.498
Louisiana 1.22 (0.44, 3.39) 0.697 1.23 (0.44, 3.43) 0.686
Detroit 0.93 (0.93, 3.56) 0.921 0.97 (0.25, 3.70) 0.960
Michigan 9.38×10
-7
(0, ∞) 0.981 9.15×10
-7
(0, ∞) 0.982
LA County 0.60 (0.16, 2.28) 0.453 0.57 (0.15, 2.18) 0.414
Age at diagnosis 1.06 (1,02, 1,10) 0.004 1.06 (1.02, 1.10) 0.005
BMI Overweight 1.10 (0.56, 2.15) 0.784 1.20 (0.60, 2.41) 0.601
Obesity 1.49 (0.76, 2.92) 0.242 1.60 (0.80, 3.21) 0.182
Smoking history Smoker − − − 1.19 (0.73, 1.93) 0.488
Drinking habit Modest − − − 0.66 (0.36, 1.22) 0.184
Heavy − − − 1.77 (0.32, 1.85) 0.565
a. Model 1 was adjusted by Registry site, age at diagnosis, and BMI.
b. Model 2 was additionally adjusted by smoking history and drinking habit based on Model 1.
c. OR = Odds ratio
d. 95% CI = 95% confidence interval

14

c. The results of logistic regression analysis for Gleason score
Model 1
a
Model 2
b
Variables OR
c
95% CI
d
P-value OR
c
95% CI
d
P-value
Comorbidity score 1 0.96 (0.67, 1.36) 0.804 1.00

(0.69, 1.44) 0.991
2 1.08 (0.77, 1.51) 0.671 1.10 (0.77, 1.55) 0.621
3 0.77 (0.53, 1.11) 0.162 0.78 (0.53, 1.14) 0.202
4 or more 0.95 (0.67, 1.35) 0.770 1.00 (0.70, 1.44) 0.993
Registry site Georgia 0.94 (0.67, 1.31) 0.703 0.94 (0.67, 1.33) 0.744
Northern CA 0.47 (0.27, 0.82) 0.008 0.51 (0.29, 0.89) 0.017
Louisiana 1.09 (0.75, 1.57) 0.655 1.08 (0.74, 1.57) 0.702
Detroit 0.97 (0.60, 1.58) 0.917 0.91 (0.55, 1.51) 0.721
Michigan 0.06 (0.01, 0.47) 0.007 2.30×10
-7
(8.45×10
-294
, 6.25×10
279
) 0.965
LA County 0.82 (0.54, 1.24) 0.347 0.82 (0.54, 1.26) 0.365
Age at diagnosis 1.03 (1.02, 1.05) < 0.001 1.03 (1.02, 1.05) < 0.001
BMI Overweight 0.94 (0.73, 1.21) 0.614 1.05 (0.80, 1.37) 0.731
Obesity 1.06 (0.82, 1.38) 0.646 1.14 (0.86, 1.49) 0.359
Smoking history Smoker − − − 0.99 (0.82, 1.21) 0.954
Drinking habit Modest − − − 0.82 (0.66, 1.03) 0.094
Heavy − − − 0.81 (0.57, 1.15) 0.245
a. Model 1 was adjusted by Registry site, age at diagnosis, and BMI.
b. Model 2 was additionally adjusted by smoking history and drinking habit based on Model 1.
c. OR = Odds ratio
d. 95% CI = 95% confidence interval

15

d. The results of multinomial logistic regression analysis for aggressiveness
Model 1
a
Model 2
b
Aggressiveness
Variables OR
c
95% CI
d
P-value
e
OR
c
95% CI
d
P-value
e
Intermediate Comorbidity score 1 0.80 (0.56, 1.15) 0.228 0.79 (0.55, 1.15) 0.223
2 0.93 (0.66, 1.33) 0.704 0.91 (0.63, 1.30) 0.600
3 0.83 (0.58, 1.19) 0.314 0.77 (0.53, 1.11) 0.158
4 or more 0.97 (0.68, 1.39) 0.862 0.93 (0.64, 1.34) 0.691
Registry site Georgia 1.30 (0.92, 1.84) 0.138 1.26 (0.88, 1.80) 0.206
Northern CA 1.05 (0.65, 1.70) 0.833 0.96 (0.59, 1.58) 0.878
Louisiana 1.32 (0.90, 1.96) 0.158 1.23 (0.83, 1.84) 0.308
Detroit 2.46 (1.42, 4.26) 0.001 2.21 (1.27, 3.87) 0.005
Michigan 1.54 (0.79, 2.99) 0.207 1.51 (0.75, 3.04) 0.253
LA County 1.23 (0.80, 1.88) 0.345 1.13 (0.73, 1.75) 0.573
Age at diagnosis 1.02 (1.01, 1.04) < 0.001 1.02 (1.00, 1.03) 0.023
BMI Overweight 1.32 (1.01, 1.70) 0.041 1.33 (1.02, 1.74) 0.034
Obesity 1.29 (0.99, 1.69) 0.059 1.28 (0.97, 1.68) 0.080
Smoking history Smoker − − − 1.09 (0.89, 1.34) 0.389
Drinking habit Modest − − − 0.79 (0.63, 0.98) 0.033
Heavy − − − 0.85 (0.61, 1.19) 0.342
High
Comorbidity score 1 0.83 (0.56, 1.22) 0.342 0.87 (0.58, 1.30) 0.497
2 0.91 (0.62, 1.33) 0.635 0.92 (0.62, 1.36) 0.666
3 0.69 (0.47, 1.03) 0.069 0.69 (0.46, 1.03) 0.071
4 or more 0.84 (0.57, 1.24) 0.370 0.86 (0.57, 1.28) 0.458
Registry site Georgia 0.93 (0.65, 1.33) 0.684 0.90 (0.62, 1.30) 0.572
Northern CA 0.67 (0.39, 1.14) 0.131 0.65 (0.38, 1.11) 0.112
Louisiana 1.09 (0.73, 1.64) 0.672 1.02 (0.67, 1.54) 0.940
Detroit 1.68 (0.94, 3.01) 0.082 1.49 (0.82, 2.71) 0.186
Michigan 0.30 (0.11, 0.82) 0.019 0.28 (0.09, 0.80) 0.018
LA County 0.90 (0.57, 1.40) 0.631 0.84 (0.53, 1.34) 0.470
Age at diagnosis 1.04 (1.02, 1.05) < 0.001 1.03 (1.02, 1.05) < 0.001
BMI Overweight 0.94 (0.71, 1.25) 0.687 1.04 (0.78, 1.39) 0.790
Obesity 1.15 (0.86, 1.52) 0.349 1.19 (0.88, 1.59) 0.256
Smoking history Smoker − − − 1.16 (0.94, 1.45) 0.173
Drinking habit Modest − − − 0.69 (0.54, 0.88) 0.003
Heavy − − − 0.69 (0.47, 1.00) 0.050
a. Model 1 was adjusted by Registry site, age at diagnosis, and BMI.
b. Model 2 was additionally adjusted by smoking history and drinking habit based on Model 1.
c. OR = Odds ratio
d. 95% CI = 95% confidence interval
e. P-values were based on 2-sided Wald’s test.

16

Discussion
In this cohort of 3265 AA men, the associations between comorbidity score and four PCa tumor
characteristics were assessed using different regression models. PSA levels were statistically
significantly associated with comorbidity scores after adjusting for registry site, age at diagnosis,
BMI, and lifestyle. However, the association was negative, which means individuals with more
comorbid conditions were less likely to have high PSA levels than those with 0 comorbidities.
As for other tumor characteristics, their associations with comorbidity were also all negative,
albeit not significant.
In the current study we assess cancer severity or aggressiveness which were obtained from the
results of PCa screening. The leading objective of cancer screening is to reduce cancer-specific
mortality
16
. The effect of comorbidity among PCa patients has been controversial. A recent
observational study on the effect of comorbidity on PCa-specific mortality indicated that the
association between them was not significant after adjusting for patient and tumor characteristics
and received treatment among Sweden males
7
. In this study, the investigators did not add race
into the Cox proportional hazards regression models. However, another cohort study based on
the National Cancer Institute SEER program showed that there was a significant association
between comorbidity and other-cause mortality among PCa patients

and race was accounted for
in their analysis adjustment
9
. There studies used PCa-specific or other mortality. Moreover, some
comorbid conditions would increase the risk of biochemical recurrence among PCa patients,
such as hypertension
17
.
There are some limitations to our study. First, we were not using the full population of the
RESPOND project. It is expected to enroll at least 10,000 AA men for this project; however, as
17

this is an interim analysis, we only included 3265 AA men for this study. In addition, it is
possible that the sample population probably is not representative of the entire AA population.
Although cancer registry sites were important in the analysis, we only used data from 7 cancer
registries in this study, which could potentially affect the results. Second, the comorbidity
conditions were self-reported and did not compare with the medical record. It is possible for
some participants not to include all comorbidity conditions in the questionnaire. Additionally, we
used a modified version of the Charlson comorbidity index by using a sum of the number of
comorbid conditions. Some conditions might be highly related to PCa, while others might be not
thus diluting our ability to observe an association. In the future, we could potentially modify the
comorbidity score to include a weight specific for PCa studies. Moreover, screening bias might
exist in our study since individuals with more comorbidity conditions are more likely to have
symptoms and do the blood tests and biopsies early, which leads to lower PSA levels, clinical
stages and Gleason scores. Moreover, patients with more comorbid conditions are more likely to
be under medical care and screen more often than those with few comorbid conditions.
18

Supplement
The initial stages of analysis were accomplished, along with the above analysis on tumor
characteristics and comorbidity. As mentioned before, the datasets included two sections, the
cancer registry dataset, and the survey questionnaires dataset. Besides these two datasets, there
was a dataset called the “merged” dataset generated by merging participants’ respond ID. For the
purpose of examining distributions of each variable, descriptive statistics reports were needed for
all three datasets.
We had two steps in the initial stages of analysis, data processing and cleaning steps. In the data
processing step, data processing reports would be generated based on the raw data in three
datasets. The raw data would be labelled and interpreted based on the NAACCR data
dictionary
12
and the survey questions, for cancer registry variables and survey questionnaire
variables separately. Then the descriptive statistics results, including distribution, percentage of
missing values, etc., would be outputted in the reports. We did not change any values in those
datasets in this stage. We continued to clean the data based on the data processing step. For the
cancer registry data, we picked 9 leading variables to be cleaned first, including marital status at
diagnosis, age at diagnosis, date of birth, grade, Surgery of Primary Site, Gleason's Primary
Pattern and Secondary Pattern Values, Gleason's Score on Needle Core Biopsy, PSA Lab Value,
and derived AJCC “T” stage. If these variables contain values of “unknown”, we converted
“unknown” values to missing values. For survey questionnaire data, there are values called
“Scantron error” in the data processing step. We converted these values to missing values as
well. Finally, we produced data cleaning reports for three datasets.
19

The above analysis and reports were produced by using R Markdown. R Markdown is a useful
tool for data science. It could execute codes and develop reports of high quality with texts,
results of codes, figures, and tables. It allows the analyst’s work results to be better presented to
the audience. In order to produce fancy figures and tables, R packages are used to achieve this
goal. The widely used packages which were used in our work were “ggplot2”, “knitr”,
“questionr”, “summarytools”, etc. For the cancer registry dataset, “summarytools” package was
mainly used. The output contained variable names, labels, values, frequencies, graphs of
distribution, and missing percentages. For the survey questionnaire dataset, “questionr” package
was mainly used. Meanwhile, the “kable” function in the “knitr” package was used to make the
format of results better. The output was a table containing numbers, percentages, valid
percentages, and cumulative percentages of each level for specified variables. See examples in
the Appendix Table 1.

20

Reference:
1. Siegel, RL, Miller, KD, Fuchs, H, Jemal, A. Cancer Statistics, 2021. CA Cancer J
Clin. 2021: 71: 7- 33. https://doi.org/10.3322/caac.21654
2. Chornokur, G., Dalton, K., Borysova, M. E., & Kumar, N. B. (2011). Disparities at
presentation, diagnosis, treatment, and survival in African American men, affected by
prostate cancer. The Prostate, 71(9), 985–997. https://doi.org/10.1002/pros.21314
3. Dess, R. T., Hartman, H. E., Mahal, B. A., Soni, P. D., Jackson, W. C., Cooperberg, M. R.,
Amling, C. L., Aronson, W. J., Kane, C. J., Terris, M. K., Zumsteg, Z. S., Butler, S.,
Osborne, J. R., Morgan, T. M., Mehra, R., Salami, S. S., Kishan, A. U., Wang, C., Schaeffer,
E. M., Roach, M., 3rd, … Spratt, D. E. (2019). Association of Black Race With Prostate
Cancer-Specific and Other-Cause Mortality. JAMA oncology, 5(7), 975–983.
https://doi.org/10.1001/jamaoncol.2019.0826
4. NCI CRCHD. National Cancer Institute. (n.d.). Retrieved September 16, 2021, from
https://www.cancer.gov/about-nci/organization/crchd. Ward E, Jemal A, Cokkinides V, et
al. Cancer disparities by race/ethnicity and socioeconomic status. CA Cancer J Clin.
2004;54(2):78-93. doi:10.3322/canjclin.54.2.78
5. Hoffman, R. M., Gilliland, F. D., Eley, J. W., Harlan, L. C., Stephenson, R. A., Stanford, J.
L., Albertson, P. C., Hamilton, A. S., Hunt, W. C., & Potosky, A. L. (2001). Racial and
ethnic differences in advanced-stage prostate cancer: the Prostate Cancer Outcomes Study.
Journal of the National Cancer Institute, 93(5), 388–395.
https://doi.org/10.1093/jnci/93.5.388
21

6. Barocas, D. A., & Penson, D. F. (2010). Racial variation in the pattern and quality of care for
prostate cancer in the USA: mind the gap. BJU international, 106(3), 322–328.
https://doi.org/10.1111/j.1464-410X.2010.09467.x
7. Rajan, P., Sooriakumaran, P., Nyberg, T., Akre, O., Carlsson, S., Egevad, L., Steineck, G., &
Wiklund, N. P. (2017). Effect of Comorbidity on Prostate Cancer-Specific Mortality: A
Prospective Observational Study. Journal of clinical oncology : official journal of the
American Society of Clinical Oncology, 35(31), 3566–3574.
https://doi.org/10.1200/JCO.2016.70.7794
8. Darst, B. F., Sheng, X., Eeles, R. A., Kote-Jarai, Z., Conti, D. V., & Haiman, C. A. (2021).
Combined Effect of a Polygenic Risk Score and Rare Genetic Variants on Prostate Cancer
Risk. European urology, 80(2), 134–138. https://doi.org/10.1016/j.eururo.2021.04.013
9. Daskivich, T. J., Fan, K. H., Koyama, T., Albertsen, P. C., Goodman, M., Hamilton, A. S.,
Hoffman, R. M., Stanford, J. L., Stroup, A. M., Litwin, M. S., & Penson, D. F. (2013). Effect
of age, tumor risk, and comorbidity on competing risks for survival in a U.S. population-
based cohort of men with prostate cancer. Annals of internal medicine, 158(10), 709–717.
https://doi.org/10.7326/0003-4819-158-10-201305210-00005
10. Cresci, S., Wu, J., Province, M. A., Spertus, J. A., Steffes, M., McGill, J. B., Alderman, E.
L., Brooks, M. M., Kelsey, S. F., Frye, R. L., Bach, R. G., & BARI 2D Study Group (2011).
Peroxisome proliferator-activated receptor pathway gene polymorphism associated with
extent of coronary artery disease in patients with type 2 diabetes in the bypass angioplasty
revascularization investigation 2 diabetes trial. Circulation, 124(13), 1426–1434.
https://doi.org/10.1161/CIRCULATIONAHA.111.029173
22

11. African American Prostate Cancer Study. RESPOND. (n.d.). Retrieved September 14, 2021,
from https://respondstudy.org/.
12. NAACCR data dictionary. (n.d.). North American Association of Central Cancer Registries
(NAACCR). Retrieved September 22, 2021, from
http://datadictionary.naaccr.org/default.aspx?c=10
13. D'Amico, A. V., Whittington, R., Malkowicz, S. B., Schultz, D., Blank, K., Broderick, G. A.,
Tomaszewski, J. E., Renshaw, A. A., Kaplan, I., Beard, C. J., & Wein, A. (1998).
Biochemical outcome after radical prostatectomy, external beam radiation therapy, or
interstitial radiation therapy for clinically localized prostate cancer. JAMA, 280(11), 969–
974. https://doi.org/10.1001/jama.280.11.969
14. Charlson, M. E., Pompei, P., Ales, K. L., & MacKenzie, C. R. (1987). A new method of
classifying prognostic comorbidity in longitudinal studies: development and validation.
Journal of chronic diseases, 40(5), 373–383. https://doi.org/10.1016/0021-9681(87)90171-8
15. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity
in adults: executive summary. Expert Panel on the Identification, Evaluation, and Treatment
of Overweight in Adults. (1998). The American journal of clinical nutrition, 68(4), 899–917.
https://doi.org/10.1093/ajcn/68.4.899
16. Laurila, M., Tammela, T. L., Auvinen, A., Isola, J., Visakorpi, T., Luukkaala, T., Määttänen,
L., Ruutu, M., Ala-Opas, M., Mildh, M., & Martikainen, P. (2009). Biological aggressiveness
of prostate cancer in the Finnish screening trial. International journal of cancer, 124(3), 547–
552. https://doi.org/10.1002/ijc.23873

23

17. Asmar, R., Beebe-Dimmer, J. L., Korgavkar, K., Keele, G. R., & Cooney, K. A. (2013).
Hypertension, obesity and prostate cancer biochemical recurrence after radical
prostatectomy. Prostate cancer and prostatic diseases, 16(1), 62–66.
https://doi.org/10.1038/pcan.2012.32

24

Appendix
a. Output of “summarytools” package

b. Output of “questionr” package

Appendix Figure 1. Examples of R Markdown outputs
25

Appendix Table 1. Comorbid condition distribution
Comorbid condition
Participants n (%)
(N = 3265)
Heart Attack
214 (6.6)
Heart Failure or CHF
196 (6.0)
Stroke
291 (8.9)
Hypertension
2282 (69.9)
Peripheral arterial disease
190 (5.8)
High Cholesterol
1648 (50.5)
Asthma, COPD
483 (14.8)
Stomach ulcers
277 (8.5)
Crohn’s Disease
103 (3.2)
Diabetes
1032 (31.6)
Kidney Problems
185 (5.7)
Cirrhosis, liver damage
64 (2.0)
Arthritis
386 (11.8)
Dementia
43 (1.3)
Depression
391 (12.0)
AIDS
35 (1.1)
Other Cancer
199 (6.1)

Abstract (if available)

Abstract Prostate cancer (PCa) has a high incidence rate and mortality among men in the United States, especially among African American (AA) men. Several studies have investigated the potential explanations for the disparity of PCa in the AA men population including socioeconomic status, the clinical stages at diagnosis, quality of care they received and genetic susceptibility. Additionally, there is evidence showing that AA men are more likely to have comorbidity conditions such as cardiovascular disease. Here, we assess the effect of comorbidity on PCa tumor characteristics, including prostate-specific antigen (PSA) levels, clinical stage, Gleason score, and tumor aggressiveness, among 3265 AA men from the RESPOND project. The tumor aggressiveness categories (low, intermediate, and high) are generated by the other three characteristics based on the D'Amico criteria. The comorbidity score is generated by using the modified Charlson comorbidity index. Descriptive statistics tables are used to describe the distribution of each PCa characteristic stratified by different comorbidity groups. The linear regression model is used to evaluate the association between comorbidity and PSA levels. For the effects of comorbidity on the clinical stage and Gleason score, we use multiple logistic regression models. We use a multinomial model to analyze three levels of aggressiveness, low, intermediate, and high. We also adjust for lifestyle factors based on the above models. There is a significantly negative association between PSA level and comorbidity after adjusting for lifestyle (p = 0.034). Other associations between comorbidity score and clinical stage, Gleason score and aggressiveness are all negative, albeit not significant. Individuals with more comorbid conditions were less likely to have high values of PCa characteristics compared to those with no comorbidities. This is an interim analysis, and the sample population may not be representative. Also, selection, screening and survival bias are possibly introduced in our study.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Body size and the risk of prostate cancer in the multiethnic cohort

PDF

Genes and environment in prostate cancer risk and prognosis

PDF

Using genetic ancestry to improve between-population transferability of a prostate cancer polygenic risk score

PDF

Prostate cancer disparities among Californian Latinos by country of origin: clinical characteristics, incidence, treatment received and survival

PDF

An analysis of disease-free survival and overall survival in inflammatory breast cancer

PDF

Prostate cancer: genetic susceptibility and lifestyle risk factors

PDF

Association between informed decision-making and mental health-related quality of life in long term prostate cancer survivors

PDF

Racial/ethnic differences in patient experiences with health care in association with the Gleason’s score level at prostate cancer diagnosis: findings from the SEER-CAHPS data

PDF

Disparities in colorectal cancer survival among Latinos in California

PDF

Genetic studies of cancer in populations of African ancestry and Latinos

PDF

Predictive factors of breast cancer survival: a population-based study

PDF

The impact of the COVID-19 pandemic on cancer care delivery

PDF

Air pollution and breast cancer survival in California teachers: using address histories and individual-level data

PDF

The effect of renal function on toxicity of E7389 (eribulin) among patients with bladder cancer

PDF

Identification and fine-mapping of genetic susceptibility loci for prostate cancer and statistical methodology for multiethnic fine-mapping

PDF

Dietary and supplementary folate intake and prostate cancer risk

PDF

Native American ancestry among Hispanic Whites is associated with higher risk of childhood obesity: a longitudinal analysis of Children’s Health Study data

PDF

Diet quality and pancreatic cancer incidence in the multiethnic cohort

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

A multivariate approach to explore nursing home characteristics contributing to high performance on CMS' publicly reported quality measures

Asset Metadata

Creator Hu, Weijia (author)

Core Title Association of comorbidity with prostate cancer tumor characteristics in African American men

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publication Date 11/15/2021

Defense Date 11/12/2021

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

Tag African American,comorbidity,OAI-PMH Harvest,prostate cancer,tumor characteristics

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Conti, David (committee chair), Haiman, Christopher Alan (committee member), Hamilton, Ann (committee member)

Creator Email weijiahu@usc.edu,weijiahuwh@163.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC17138267

Unique identifier UC17138267

Legacy Identifier etd-HuWeijia-10219

Document Type Thesis

Rights Hu, Weijia

Type texts

Source 20211115-wayne-usctheses-batch-896-nissen (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. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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