University of Southern California Dissertations and Theses

Fish consumption and risk of colorectal cancer

(USC Thesis Other)

Fish consumption and risk of colorectal cancer

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content !
! 1
!

Fish Consumption and Risk of Colorectal Cancer

by

Julia Sturgeon

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS AND EPIDEMIOLOGY)

August 2015
!
! 2

Table of Contents
List of Figures .............................................................................................................................................. 3
List of Tables ............................................................................................................................................... 4
Abstract ........................................................................................................................................................ 5
Chapter 1: Introduction ................................................................................................................................ 6
Chapter 2: Methods ................................................................................................................................... 14
Chapter 3: Statistical Analysis ................................................................................................................... 16
Chapter 4: Results ...................................................................................................................................... 18
!
Distribution of Fish Consumption ........................................................................................................................ 20
Pairwise Correlation between Fish Variables ..................................................................................................... 35
Unadjusted and Adjusted Odds Ratios for Fish Consumption (never consume fish versus consume fish) ......... 35
Conditional Logistic Regression: Analysis of Matched Pairs ............................................................................. 40
Assessing Dose-Response of Fish Consumption .................................................................................................. 40
Chapter 5: Discussion ................................................................................................................................ 45
References ................................................................................................................................................. 49

!
! 3
List of Figures
!
Figure 1. Distribution of Canned Tuna Consumption for all Participants ................................................ 21
Figure 2. Distribution of Mackerel/Salmon/Sardines Consumption for all Participants ........................... 22
Figure 3. Distribution of Cooked/Fried Fish Consumption for all Participants ........................................ 23
Figure 4. Distribution of Canned Tuna Consumption in Controls ............................................................ 24
Figure 5. Distribution of Mackerel/Salmon/Sardines Consumption in Controls ...................................... 25
Figure 6. Distribution of Cooked/Fried Fish Consumption in Controls .................................................... 26
Figure 7. Distribution of Canned Tuna Consumption by Age Quartile Groupings .................................. 27
Figure 8. Distribution of Mackerel/Salmon/Sardines Consumption by Age Quartile Groupings ............. 28
Figure 9. Distribution of Cooked/Fried Fish Consumption by Age Quartile Groupings .......................... 29
Figure 10. Distribution of Canned Tuna Consumption by Ethnicity Groupings ...................................... 30
Figure 11. Distribution of Mackerel/Salmon/Sardines Consumption by Ethnicity Groupings ................. 32
Figure 12. Distribution of Cooked/Fried Fish Consumption by Ethnicity Groupings .............................. 34
!
!
! 4
List of Tables
Table 1. Fish Species and Omega-3 Fatty Acids ........................................................................... 12
Table 2. Summary of Studies examining Fish Consumption and Colorectal Cancer Risk ........... 13
Table 3. Participant Characteristics ............................................................................................... 19
Table 4. Canned Tuna Consumption by Ethnicity ........................................................................ 31
Table 5. Mackerel/Salmon/Sardines Consumption by Ethnicity ................................................... 33
Table 6. Consumption of Cooked/Fried Fish by Ethnicity ............................................................ 34
Table 7. Correlation between Fish Variables (heatmap codes are designated with identity
correlations=1 are highlighted in red, with correlations > 0.3 shown in yellow, and
correlations <0.3 shown in green) ......................................................................................... 35
Table 8. 2x2 Contingency Table for Canned Tuna ....................................................................... 36
Table 9. 2x2 Contingency Table for Mackerel/Salmon/Sardines ................................................. 36
Table 10. 2x2 Contingency Table for Cooked/Fried Fish ............................................................. 37
Table 11. Crude and Adjusted Associations between Fish Consumption and Risk of Colorectal
Cancer .................................................................................................................................... 38
Table 12. Crude and Adjusted Associations between Fish Consumption and Risk of Colon
Cancer .................................................................................................................................... 38
Table 13. Crude and Adjusted Associations between Fish Consumption and Risk of Rectal
Cancer .................................................................................................................................... 38
Table 14. Crude Associations between Fish Consumption and Risk of Colorectal Cancer using
Conditional Logistic Regression ........................................................................................... 40
Table 15. 3x2 Contingency Table for Tertiles of Canned Tuna Consumption ............................. 41
Table 16. 3x2 Contingency Table for Tertiles of Mackerel/Salmon/Sardines Consumption ....... 41
Table 17. 3x2 Contingency Table for Tertiles of Cooked/Fried Fish Consumption ..................... 42
Table 18. Adjusted and Unadjusted Odds Ratios by Tertiles of Fish Consumption ..................... 43
Table 19. Adjusted and Unadjusted Odds Ratios by Tertiles of Mackerel, Salmon, and Sardines
Consumption .......................................................................................................................... 43
Table 20. Adjusted and Unadjusted Odds Ratios by Tertiles of Cooked/Fried Fish Consumption
............................................................................................................................................... 44
!
! 5

Abstract
!
The association between fish consumption and risk of colorectal cancer was investigated
in a population-based, incidence density case control study from northern Israel, the Molecular
Epidemiology of Colorectal Cancer study.

Subjects and Methods: Food frequency questionnaire data were used from 1,869 cases and 1,777
controls ascertained and recruited between March 31, 1998 and February 14, 2005.
Questionnaire data included three specific fish categories: canned tuna,
mackerel/salmon/sardines, and cooked/fried fish. Multivariate unconditional logistic regression
was used to assess the association between each fish consumption variable and risk of colorectal
cancer.

Results: After adjusting for age, sex, and ethnicity consumption of canned tuna was associated
with reduced risk of colorectal cancer. Participants who consumed canned tuna had an 18%
reduced risk of colorectal cancer compared to those who never consumed canned tuna (odds ratio,
0.82; 95% confidence interval, [0.71, 0.94]). Comparing the highest tertile of fish consumption
(1+ serving/week) to the lowest tertile of fish consumption (0 servings/month), canned tuna was
associated with a 24% reduction of colorectal cancer risk. Adjusting known risk factors and
potential confounders, consumption of mackerel/ salmon/sardines was associated with a 27%
reduced risk of colorectal cancer when comparing the highest and lowest tertiles of consumption
(1+ serving/week vs. 0 servings/month). Consumption of cooked/fried fish was not associated
with risk of colorectal cancer at any level of consumption.

Conclusions: Data from this planned interim analysis suggest that diets characterized by the
consumption of canned tuna as well as mackerel, salmon, and sardines are associated with
reduced risk of colorectal cancer. Cooked and fried fish consumption was not associated with
risk of colorectal cancer. This finding is consistent with animal studies that have examined the
association between omega-3 fatty acids and colon tumor growth, indicating that omega-3 fatty
acids inhibit colon carcinogenesis.
!
! 6
Chapter 1: Introduction

In 2012, there were 1.4 million newly diagnosed cases of colorectal cancer worldwide [1].
The International Agency for Research on Cancer reported an estimated incidence of 4,033 new
cases of colorectal cancer in Israel with an age-standardized rate of 35.9 cases per 100,000
people [2]. This age-standardized incidence rate ranked Israel 12
th
in the world for incidence of
colorectal cancer in 2012 [1]. Compared to other cancers, colorectal cancer has the third highest
age-standardized incidence rate in Israel, behind breast and prostate cancers for both sexes
combined [2]. Examining men and women independently, colorectal cancer has the second
highest age-standardized mortality rate [2].
A number of factors have been identified as contributing to increased colorectal cancer
risk. Genetic susceptibility, dietary habits, and lifestyle factors are three categories that have
been confirmed as links to colorectal cancer risk [3]. Previously published data from the MECC
study demonstrate the relationships between genetic risk factors ([4], [5], [6], [7], [8], [9]),
alcohol [10], carotenoids [11], physical activity [12], medications ([13], [14], [15], [16]) and risk
of colorectal cancer.
Genetic susceptibility is of particular interest in the Israeli population as known genetic
risk factors have been linked to the Ashkenazi Jewish population, or those in the Jewish
population who have European ancestry ([3], [17], [18]). In particular, germline mutations in
mismatch repair genes such as MLH1, MSH2, MSH6, PMS1, and PMS2 have been linked to
hereditary nonpolyposis colorectal cancer (HNPCC). More specifically, MSH2 has been
identified as a gene with many founder mutations in the Ashkenazi Jewish population [19].
While genetic susceptibility plays a role in risk of colorectal cancer, it does not cover the
spectrum of risk factors associated with this disease. Research in the past decades has indicated
contributions to colorectal cancer risk from both dietary habits and lifestyle factors.
Lifestyle and dietary habits have been identified as critical components in increased risk
of colorectal cancer. It has been shown that men who consume a higher calorie diet have a 74%
increased risk of colorectal cancer. Similarly, women who consume a higher calorie diet have a
70% increased risk for colorectal cancer (after adjusting for physical activity) [20]. This risk
factor is even greater for individuals with a first-degree relative who has had a colorectal cancer
!
! 7
diagnosis [20]. As a corollary to this finding, high Body Mass Index (BMI) has also been linked
to increasing colorectal cancer risk [21]. While a higher calorie diet has been linked to risk of
colorectal cancer, participants who had a higher proportion of their calories derived from fiber
did not experience as large of an increased risk, suggesting that fiber is protective in this
association [20]. In addition to dietary habits, lifestyle is known to play a role in colorectal
cancer risk. Sedentary lifestyles are a known risk factor for colorectal cancer [21]. Obesity and
sedentary lifestyles contribute to overall colorectal cancer risk on a broader scale, however,
specific dietary patterns have been shown to increase risk too.
Recently, red meat consumption and risk of colorectal cancer has been under
investigation. Using data from a 1990s cohort study, English et al. found a 2.3-fold increased
risk of rectal cancer for men in the highest quartile of fresh red meat consumption and a 2-fold
increased risk of rectal cancer for men in the highest quartile of processed meat [22]. Two main
biological hypotheses were at the forefront of the red meat investigation: 1) the link between
animal fat consumption and colorectal cancer risk and 2) the preparation of the meat and its
association with colorectal cancer risk [23]. The link between animal fat and colorectal cancer
risk has become more tenuous as saturated fat does not seem to have an impact on colorectal
cancer risk [22]. Both the molecular makeup of red meat and the changes that occur to red meat
when it is cooked at high temperatures have been theorized as the reason for the increased risk of
colorectal cancer [22]. Red meat has N-nitroso compounds that have been identified as
carcinogenic. In addition to this, when red meat is cooked at high temperatures, heterocyclic
amines are formed and have also been identified as carcinogenic [22].
The investigation into the link between red meat consumption and risk of colorectal
cancer has caused a cascade of other dietary factors to come under scrutiny, specifically: fish,
poultry, and eggs. No substantive links have been established between eggs and poultry and
colorectal cancer. Evidence for an association between fish consumption and risk of colorectal
cancer has been inconsistent [24].
The association between fish consumption and colorectal cancer risk remains
unsubstantiated. Studies have been conducted in both the West (the US and European countries)
as well as in the East (Japan) where fish consumption is highest. Fish species, however, vary in
micronutrient and macronutrient makeup. The studies that have examined the association
between fish and colorectal cancer risk have examined a range of fish types (such as lean versus
!
! 8
fatty) and fish preparation methods (such as frying, salting, preserving, and baking). Previous
studies have identified that fish species as well as preparation methods play an important role in
the impact on cancer risk. In particular, long-chain omega-3 polyunsaturated fatty acids (a
component of dark, fatty fish) have been hypothesized to inhibit colon carcinogenesis [25]. In
general, long-chain omega-3 polyunsaturated fatty acids have anti-inflammatory properties that
may be protective against colorectal cancer (since inflammation in the colon and rectum causes
an increased risk for cancer in this region) [25]. Preceding the investigation of long-chain
omega-3 polyunsaturated fatty acids and reduced risk of colorectal cancer in humans, results
from animal studies have suggested other plausible reasons (besides their anti-inflammatory
properties) that these fatty acids may have in protecting against colorectal cancer. Specifically,
the omega-3 fatty acids may inhibit tumor growth, suppress angiogenesis, or increase apoptosis
[25]. In a study conducted in rats, Chang et al. found a lower incidence of adenocarcinomas in
rats that were fed fish oil compared to rats that were fed corn oil. The biological findings from
this study suggest that increased apoptosis and differentiation (instead of a reduction in
proliferation) for those rats given fish oil is what is inhibiting the colon tumorigenesis [26].
These preliminary data from animal studies have spurred on the investigation into the role of
omega-3 fatty acids and risk of colorectal cancer in humans through fish and fish oil supplement
consumption.
In 2013, data from a large prospective cohort in Washington State was used to investigate
the role of long-chain omega-3 polyunsaturated fatty acid intake and colorectal cancer risk.
While the result was not statistically significant at a 5% Type I error rate (p-value=0.06),
investigators determined a 49% reduced risk of colorectal cancer for those participants who
reportedly consumed a fish oil supplement four or more times per week for three or more years
[25]. In addition to fish oil, dark fish consumption was also examined. There was no association
found between dark fish (in this case, salmon and fresh tuna) consumption and risk of colorectal
cancer [25].
Other studies have investigated the association between long-chain omega-3 fatty acid-
containing fish and risk of colorectal cancer. Using food frequency questionnaire information
from the Physician’s Health Study, Hall et al. investigated the long-chain omega-3 fatty acids
hypothesis by calculating the average amount of omega-3 fatty acids in different types of
reported fish and seafood. The amounts of omega-3 fatty acids were calculated for canned tuna,
!
! 9
dark fish (mackerel, salmon, sardines, blue fish, and swordfish), and seafood [27]. The authors
found a 40% decreased risk of colorectal cancer when comparing the highest and lowest fish
consumption categories [27]. In addition to this finding, a 27% reduction in colorectal cancer
risk was found when comparing the highest and lowest quartiles of omega-3 fatty acid
consumption [27]. While some studies examined dark fish consumption and fatty acids
separately, or just dark fish alone as a proxy for fatty acids, other studies have examined different
types of dietary fats independent of specific types of food (such as fish).
Using the Iowa Women’s Health Study cohort, Bostick et al. examined an array of
dietary items including red meat, fish, and different types of dietary fat [23]. Among the types of
fat looked at, omega-3 fatty acid consumption was investigated. Although none of the quintiles
of omega-3 fatty acid consumption were statistically significant, there was a significant trend in
reduced risk of colorectal cancer with increasing omega-3 fatty acids consumption [23].
While the omega-3 polyunsaturated fatty acids hypothesis holds biological plausibility
for a protective association with colorectal cancer, some investigators have found null results. In
Sweden, intake of polyunsaturated fatty acids was investigated in a cohort of women. These fats
included omega-6 polyunsaturated fatty acids, eicosapentaenoic acid (EPA), and
docosahexaenoic acid (DHA). EPA and DHA are two specific types of omega-3 fatty acids that
have been hypothesized as being protective against colorectal cancer risk [25]. The study found
no association between dietary fats and fatty acids intake and colorectal cancer risk (relative risk,
0.96; 95% confidence interval, [0.72, 1.28]) [28]. Nonetheless, studies continue to probe into the
association between different fish species and colorectal cancer risk. In the case of some studies,
fish species differentiation is not always possible, thus making any links between omega-3 fatty
acid fish consumption and colorectal cancer risk difficult to establish.
A pitfall to many of the studies that have examined the association between fish
consumption and colorectal cancer risk is that the dietary questionnaires are lacking in the types
of fish that are being consumed and, in many cases, lumping fish consumption into one category,
including fish that have undergone extensive processing or preparation. When the studies are
lacking in species-specific fish information, investigators often make inferences about the types
of fish being consumed based on regional preferences and availability. In a hospital-based case-
control study conducted in Krakow, Poland, investigators found a statistically significant inverse
!
! 10
association between fish consumption and risk of colorectal cancer. One limitation of the study
that was addressed by the authors was the lack of information about the specific types of fish
consumed by participants [29]. Examining the distribution of different species of fishery
products consumed in Poland, the authors made a tentative conclusion that participants were
mostly consuming pollock, herring, and mackerel [29]. These fish species are all known to have
varying amounts of omega-3 fatty acids, with mackerel and herring with the highest amount of
omega-3 fatty acids per serving. Based on nutritional information, herring has one of the highest
amounts of omega-3 fatty acids when compared to all other commonly consumed fish.
The question of fish consumption and risk of colorectal cancer has been examined in
Japan, a country with one of the highest fish consumption rates in the world [24]. In a
prospective cohort of 39,498 men and women (with 566 incident cases of colorectal cancer
during follow-up), fish consumption and its association with colorectal cancer was examined.
Although they are not reported independently of each other, the fish under investigation
consisted mostly of tuna, mackerel, and salmon, fish that are known to be high in omega-3 fatty
acids. Comparing the highest and lowest quartiles of fish consumption, no association was found
between fish consumption and risk of colorectal cancer for men or women [24].
Using data from five cohort studies and twelve case-control studies, Pham et al.
performed a systematic review examining the association between fish consumption and risk of
colorectal cancer in studies conducted in the Japanese population [30]. The authors concluded
that there was no evidence to support an association between fish consumption and colorectal,
colon, and rectal cancer risks. More specifically, in the meta-analysis, the pooled relative risk
for the cohort studies did not show an association with risk of colorectal cancer (pooled odds
ratio, 1.03; 95% confidence interval, [0.89, 1.18]). The pooled odds ratio for the case-control
studies showed a moderate reduction in risk of colorectal cancer for participants that consumed
fish (pooled odds ratio, 0.84; 95% confidence interval, [0.75, 0. 94]) [30]. In the discussion of
the results, the author mentions that fatty fish are most widely consumed in Japan. These fish,
however, may be contaminated by polychlorinated biphenyls, a compound that is known to be
associated with colorectal cancer risk. It is also important to note that the fish may have been
salted, a preparation method that may have negative health implications [30]. While this
!
! 11
systematic review and meta-analysis only examined studies conducted in the Japanese
population, a separate meta-analysis offers contrasting results.
In 2012, Wu et al. published a systematic review and meta-analysis that examined
twenty-two prospective cohort studies and nineteen case-control studies that were conducted
worldwide [31]. The authors found that fish consumption reduced colorectal cancer risk by 12%.
For rectal cancer alone, there was a 21% risk reduction. The authors note that one of the criteria
to be included in the meta-analysis was that the study had to examine fresh fish (not salted or
fried fish). Once again, the fish species and fish preparation method plays a large role in
examining the association between colorectal cancer risk and fish consumption [31].
The association between fish consumption and risk of colorectal cancer still remains
inconclusive. Our aim is to look at specific types of fish with known omega-3 fatty acid levels
and examine the relationship between fatty vs. non-fatty fish and colorectal cancer risk.

Two hypotheses are explored in this analysis:
1) Diets characterized by consumption of fish containing omega-3 fatty acids (specifically,
docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA)) are associated with
reduced risk of colorectal cancer.
2) Consumption of fish that are not high in omega-3 fatty acids are associated with less
reduction in risk of colorectal cancer than consumption of fish high in omega-3 fatty
acids.

!
! 12

Table 1. Fish Species and Omega-3 Fatty Acids
Fish Species
2
Serving size Omega-3 Fatty Acids
1

(mg)
EPA DHA
Herring 100 g 2218 mg 909 mg 1105 mg
Canned sardines (in
oil)
92 g (1 can)
100 g
1362 mg
1480 mg
435 mg
473 mg
468 mg
509 mg
Canned mackerel
(drained)
361 g (1 can)
100 g
4970 mg
1377 mg
1567 mg
434 mg
2873 mg
796 mg
Trout 100 g 1370 mg 259 mg 677 mg
Canned salmon
(sockeye, drained)
369 g (1 can)
100 g
4,882 mg
1323 mg
1816 mg
492 mg
2450 mg
664 mg
Canned tuna (white,
in water, drained)
172 g (1 can)
100 g
1636 mg
951 mg
401 mg
233 mg
1082 mg
629 mg
Carp 100 g 902 mg 305 mg 146 mg
Halibut 100 g 669 mg 91 mg 374 mg
Pollock 100 g 570 mg 91 mg 451 mg
Canned tuna (white,
in oil, drained)
178 g (1 can)
100 g
806 mg
453 mg
117 mg
66 mg
317 mg
178 mg
Grouper 100 g 265 mg 35 mg 213 mg
Tilapia 100 g 240 mg 5. 0 mg 130 mg
1
table is arranged in descending omega-3 fatty acids content (based on 100 g serving)
2
fish that are not classified as “canned” are cooked under dry heat
Nutritional Information is through the United States Department of Agriculture [32]

!
! 13

Table 2. Summary of Studies examining Fish Consumption and Colorectal Cancer Risk
Author, year Study
design
Sample Size Risk Groups Relative
Risk
Confidence
Intervals
Chiu, 2003 Case-
Control
931 cases, 1552
controls
Highest versus lowest
quartile of fish consumption
1.7 (men)
1.2
(women)
[1.2,2.4]
(men)
[0.8,1.7]
(women)
Fernandez,
1999
Case-
Control
828 cases, 7990
controls
2+ servings/week of fish
versus 0-1 servings/week of
fish
0.60 [0.50, 0.70]
Jedrychowski,
2008
Case-
Control
548 cases, 745
controls
Highest versus lowest
quartile of fish consumption
0.71 [0.5,0.98]
Kimura, 2007 Case-
Control
782 cases, 793
controls
Highest versus lowest
quintile of fish consumption
0.80 [0.57, 1.13]
Engeset, 2007 Cohort 63,914, 254 incident
cases of colorectal
cancer
Highest versus lowest tertile
of fish consumption
1.38 [0.98, 1.95]
English, 2004 Cohort 37,112, 452 incident
cases of colorectal
cancer
Highest versus lowest
quartile of fish consumption
0.90 [0.70,1.20]
Hall, 2008 Cohort 22,071, 500 incident
cases of colorectal
cancer
Highest versus lowest
category
0.60 [0.40,0.91]
Hsing, 1998 Cohort 17,633, 145 incident
cases of colorectal
cancer
Highest versus lowest
quartile of fish consumption
1.50 [0.90, 2.60]
Kantor, 2014 Cohort 68,109, 488 incident
cases of colorectal
cancer
Highest versus lowest
quartile of fish consumption
0.77 [0.55, 1.07]
Kato, 1997 Cohort 14,727, 221 incident
cases of colorectal
cancer
Highest versus lowest
quartile of fish consumption
0.49 [0.27,0.89]
Norat, 2005 Cohort 478,040, 1,329
incident cases of
colorectal cancer
Highest versus lowest
quintile of fish consumption
0.69 [0.54, 0.88]
Sugawara, 2009 Cohort 39,498, 566 incident
cases of colorectal
cancer
Highest versus lowest
quartile of fish consumption
1.07
(men)
0.96
(women)
[0.78, 1.46]
(men)
[0.61, 1.53]
(women)

!
! 14
Chapter 2: Methods

Study Design: Starting in Israel in 1998, the Molecular Epidemiology of Colorectal Cancer
(MECC) is an ongoing population-based case-control study [13]. Cases were recruited based on
a colorectal cancer diagnosis after May 31, 1998. Controls were matched to cases based on age,
sex, Jewish ethnicity, and clinic location [13]. All participants were interviewed in order to
gather information about their personal medical history and family cancer history. An extensive
food frequency questionnaire was administered to cases and controls to gain knowledge of their
dietary habits. Cases were asked to describe their dietary habits one year before their diagnosis
of colorectal cancer and controls were asked about their dietary habits in the past year. The food
frequency questionnaire is a validated tool that accurately represents dietary patterns in Israel.

Study Participants: Analysis included study participants who were recruited between May 31,
1998 and February 14, 2005. Participants were excluded from the analysis if they had missing
food frequency information. Those participants with extreme daily caloric intake were also
excluded from the analysis (extreme daily caloric intake was considered to be consumption of
less than 600 kilocalories per day or more than 4,000 kilocalories per day). The kilocalories for
each participant were calculated based on their food frequency questionnaire responses. After
implementing these parameters, the analysis included a total of 1,869 cases and 1,777 controls.

Exposure Variables: Dietary data were collected for 3,646 participants. Cases were asked to
recall their dietary patterns one year before their cancer diagnosis and controls were asked to
describe their dietary habits in the past year. Food frequency was assessed on a per serving basis
and categorized as 0 servings/month, 1-3 servings/month, 1 serving/week, 2-4 servings/week, 5-
6 servings/week, 1 serving/day, 2-3 servings/day, 4-5 servings/day, 6+ servings/day. The fish
consumption items included in the food frequency questionnaire are canned tuna, canned
mackerel/salmon/sardines, and cooked/fried fish (100-150 grams).
Total calorie consumption was computed from the food frequency questionnaire.
Vegetable consumption was also computed based on food frequency questionnaire responses and
then categorized into a dichotomous variable: five or more servings per day of vegetables and
!
! 15
fewer than five servings per day of vegetables. Aspirin use was assessed by whether the
participant had used a daily low-dose aspirin for two or more years. Statin use was recognized as
participants who had taken a daily statin for five or more years. Red meat consumption was
assessed using the hamburger/meatball/kabob food frequency questionnaire item. The
hamburger/meatball/kabob variable was analyzed on a per serving basis with the same serving
categories as the fish items (0 servings/month, 1-3 servings/month, 1 serving/week, 2-4
servings/week, 5-6 servings/week, 1 serving/day, 2-3 servings/day, 4-5 servings/day, 6+
servings/day). Physical activity was assessed using a dichotomous variable with participants
reporting either participating in sports or not. Since higher Body Mass Index (BMI) has been
linked to an increased risk of colorectal cancer, the participant’s BMI was calculated from the
questionnaire and used as a continuous covariate. For cases, BMI was calculated using the
participant’s height and weight recorded one year previous to cancer diagnosis. BMI was
calculated in the same way for controls with the exception that if the “weight year ago” variable
was missing in a control, their current weight was used instead. BMI for cases was only
calculated based on their weight one year prior to diagnosis. A participant was considered to
have a family history of colorectal cancer if they had a first-degree family member diagnosed
with either colon or rectal cancer. The ethnicity of participants was determined through the
participant’s description of their ethnic group, religious affiliation, and the nationality of their
parents and grandparents.

Outcome Assessment: The outcome under analysis in this study was colorectal cancer. Cases of
colorectal cancer were confirmed by one pathologist through a standardized pathological review.
Controls were excluded if they had a history of colorectal cancer [13].

!
! 16
Chapter 3: Statistical Analysis

All statistical analyses were performed using SAS software (Version 9.4; SAS Institute Inc, Cary,
NC). A two-tailed alternative and a 5% Type I error rate were assumed for each hypothesis test.
Exploratory data analysis included frequency counts and frequency distributions reported in
tables and histograms. These methods were used for the three fish consumption variables
assessed through the dietary interview and validated food frequency questionnaire (canned tuna,
mackerel/salmon/sardines, and cooked/fried fish) for all participants and controls alone. The
distribution of fish consumption for all participants combined was examined based on age
quartiles and ethnicity using stacked bar graphs. Frequency tables for the three fish variables
were examined by ethnicity to investigate the proportion of fish consumption per ethnicity group.
Matching and risk factor covariates were explored through frequency counts and averages for
participants, stratified by cases and controls (Table 3). A chi-square test was used to test whether
there were significant differences between the discrete covariate data when comparing the case
and control groups. For continuous covariates, a t test was performed to determine whether there
were significant differences between the group means. Pairwise correlation was performed on
the three fish variables and the corresponding pairwise correlation estimates were presented in a
heat map table.
Unadjusted main effects of canned tuna, mackerel/salmon/sardines, and cooked/fried fish
were assessed using 2x2 (case/control, never consume fish/consume fish) contingency tables to
determine the crude odds ratios and confidence intervals. Crude odds ratios and confidence
intervals were also computed for canned tuna, mackerel/salmon/sardines, and cooked fish using
unconditional logistic regression. Multivariate unconditional logistic regression was used to
calculate the adjusted odds ratios and confidence intervals for canned tuna,
mackerel/salmon/sardines, and cooked fish. Odds Ratios and confidence intervals were obtained
after adjusting for the variables on which the participants were initially matched on (age, sex,
and Jewish ethnicity). Previously established risk factors for colorectal cancer were assessed as
confounders by adding the risk factor covariates into the base model (with only the fish
consumption variable) one at a time and seeing whether the model point estimate for the fish
consumption variable changed by more than 10%. The risk factor covariates that were tested as
confounders included vegetable consumption, Body Mass Index, family history of colorectal
!
! 17
cancer, physical activity, aspirin use, NSAID use, and red meat consumption. Even though none
of the risk factor covariates changed the point estimate of the main effect by more than 5%, a
secondary adjusted analysis was performed using unconditional multivariate logistic regression
adjusting for family history of colorectal cancer, red meat consumption, and BMI as well as age,
sex, and ethnicity (the original matching factors). The secondary multivariate logistic regression
analysis was performed on a reduced data set due to missing values for BMI (approximately
3,260 total cases and controls). Adjusting for the same variables, multivariate unconditional
logistic regression was run after stratifying by tumor site (colon versus rectum). All controls
were used in the stratified analysis by tumor site. Analyses were repeated using conditional
logistic regression to accommodate the matched study design for the 3,496 participants with a
matching identifier.
The fish consumption variables were divided into tertiles of consumption (0 servings per
month, 1-3 servings per month, and 1 serving per week). Crude and adjusted odds ratios and
confidence intervals were computed for each of the fish consumption variables using the 0
servings/month category as the reference group. A test of trend was conducted on canned tuna,
mackerel/salmon/sardines, and cooked/fried fish consumption independently.
!
! 18
Chapter 4: Results

The characteristics of the participants were similar between cases and controls consistent
with the matched study design. Although cases were slightly younger than controls, the average
difference in age was within the matched criteria of within 1 year. Overall, Ashkenazi Jews were
in the majority for both cases and controls compared to other ethnicities. Controls had a higher
frequency of vegetable consumption and sport activity than cases. For cancer history, cases were
more likely to have a first-degree relative with a colorectal cancer diagnosis compared to
controls. BMI, a known modest risk factor for colorectal cancer, was not significantly different
in cases and controls. There was a significantly higher amount of daily aspirin and daily statin
use in controls compared to cases as well as red meat consumption. Finally, there were over
three times as many cases of colon cancer compared to rectal cancer among all cases of
colorectal cancer, consistent with population-based incidence rates.

!
! 19

Table 3. Participant Characteristics
Characteristic Cases Controls P Value
Total Participants—n (%) 1,869 (51) 1,777 (49)
Sex—n (%) 0.79
Male 956 (51) 901 (49)
Female 913 (51) 876 (49)
Age –years, mean +/- std. deviation 69.73 +/- 11.63 70.63 +/- 11.50 0.02
Ethnic Group –n (%) <.0001
Ashkenazi 1,321 (54) 1,142 (46)
Sephardi 324 (43) 431 (57)
Arab 187 (50) 185 (50)
Non-Jewish/non-Arab 37 (66) 19 (34)
Family History of Cancer—(yes) n (%) 126 (58) 91 (42) 0.03
Sports Participation—(yes) n (%) 577 (44) 721 (56) <.0001
Hamburger, meatball, and kabobs per week (yes) (%) 984 (53) 1,122 (63) <.0001
Vegetable Consumption <.0001
5+ servings/day 1,122 (48) 1,207 (52)
Less than 5 servings/day 747 (57) 570 (43)
Body Mass Index (BMI)—mean +/- std. deviation
1
27.3 +/- 4.7 27.0 +/- 4.6 0.08
Aspirin Use—daily low-dose aspirin for two or more years (yes) n (%) 276 (40) 421 (60) <.0001
NSAID Use—daily statin use for five or more years (yes) n (%) 109 (35) 205 (65) <.0001
Tumor Site
2

Colon 1444 (77) -----
Rectum 421 (23) -----
1
BMI was calculated on reduced sample n=3,278
2
1 missing

!
! 20

Distribution of Fish Consumption:
Both canned tuna consumption and mackerel/salmon/sardines consumption followed a right-
skewed distribution (Figures 1-2). Most participants reported only eating canned tuna and
mackerel/salmon/sardines 1-3 times per month or not at all. More participants reported eating
canned tuna 1-3 times per month than eating mackerel/salmon/sardines 1-3 times per month. For
cooked and fried fish, 1 serving per week had the highest frequency among all participants
followed by 1-3 servings per month (Figure 3). For controls alone, the distribution of canned
tuna, mackerel/salmon/sardines, and cooked/fried fish followed the same pattern as the
distribution among all participants (Figures 4-6).
Canned tuna, mackerel/salmon/sardines, and cooked and fried fish consumption were
examined separately and divided into age quartiles. Evaluating all participants combined, for
each level of canned tuna consumption, the frequency was distributed evenly between the four
age quartiles (Figures 7-9). This result was also seen in the mackerel/salmon/sardines
consumption and cooked/fried fish consumption. Examining the distribution of fish
consumption by ethnicity showed consumption patterns largely consistent with the distribution
of study participants by ethnicity.

!
! 21
Figure 1. Distribution of Canned Tuna Consumption for all Participants

1307
1195
781
314
22 19
0 month 1-3/month 1/week 2-4/week 1/day 5-6/week
Canned Tuna Consumption
0
500
1000
Frequency of Participant Responses
Distribution of Canned Tuna Consumption for all Participants
!
! 22
Figure 2. Distribution of Mackerel/Salmon/Sardines Consumption for all Participants

1916
1016
526
153
17 12
0 month 1-3/month 1/week 2-4/week 1/day 5-6/week
Mackerel/Salmon/Sardines Consumption
0
500
1000
1500
2000
Frequency of Participant Responses
Distribution of Mackerel/Salmon/Sardines Consumption for all Participants
!
! 23
Figure 3. Distribution of Cooked/Fried Fish Consumption for all Participants

562
1005
1325
351
21
13
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Cooked/Fried Fish Consumption
0
500
1000
Frequency of Participant Responses
Distribution of Cooked/Fried Fish Consumption for all Participants
!
! 24
Figure 4. Distribution of Canned Tuna Consumption in Controls

603
565
408
175
14
9
0 month 1-3/month 1/week 2-4/week 1/day 5-6/week
Canned Tuna Consumption
0
200
400
600
Frequency of Participant Responses
Distribution of Canned Tuna Consumption in Controls
!
! 25
Figure 5. Distribution of Mackerel/Salmon/Sardines Consumption in Controls

910
465
293
87
15
5
0 month 1-3/month 1/week 2-4/week 1/day 5-6/week
Mackerel/Salmon/Sardines Consumption
0
200
400
600
800
Frequency of Participant Responses
Distribution of Mackerel/Salmon/Sardines Consumption in Controls
!
! 26
Figure 6. Distribution of Cooked/Fried Fish Consumption in Controls

286
477
691
183
10
5
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Cooked/Fried Fish Consumption
0
200
400
600
Frequency of Participant Responses
Distribution of Cooked/Fried Fish Consumption in Controls
!
! 27
Figure 7. Distribution of Canned Tuna Consumption by Age Quartile Groupings

1120
1102
724
291
16 20
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Canned Tuna Consumption
0
200
400
600
800
1000
Frequency of Participant Responses
age > 78 72<= age <=78 63 < age < 72 age <= 63 Age Quartile
Distribution of Canned Tuna Consumption by Age Quartiles
!
! 28
Figure 8. Distribution of Mackerel/Salmon/Sardines Consumption by Age Quartile
Groupings

1684
937
482
143
11 16
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Mackerel/Salmon/Sardines Consumption
0
500
1000
1500
Frequency of Participant Responses
age > 78 72<= age <=78 63 < age < 72 age <= 63 Age Quartile
Distribution of Mackerel/Salmon/Sardines Consumption by Age Quartiles
!
! 29

Figure 9. Distribution of Cooked/Fried Fish Consumption by Age Quartile Groupings

562
1005
1325
351
21
13
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Cooked/Fried Fish Consumption
0
500
1000
Frequency of Participant Responses
age > 78 72<= age <=78 63 < age < 72 age <= 63 Age Quartile
Distribution of Cooked/Fried Fish Consumption by Age Quartiles
!
! 30

Figure 10. Distribution of Canned Tuna Consumption by Ethnicity Groupings

1120
1102
724
291
16 20
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Canned Tuna Consumption
0
200
400
600
800
1000
Frequency of Participant Responses
non-Jewish/non-Arab Arabs Sephardi Ashkenazi Ethnicity
Distribution of Canned Tuna Consumption by Ethnicity Groups
!
! 31
Table 4. Canned Tuna Consumption by Ethnicity
Table of Canned Tuna Consumption by Ethnicity
Ethnicity Canned Tuna
Frequency
Row Percent
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day Total
Ashkenazi 924
37.61
792
32.23
493
20.07
217
8.83
15
0.61
16
0.65
2457

Sephardi 207
27.42
259
34.30
202
26.75
79
10.46
4
0.53
4
0.53
755

Arabs 150
40.54
130
35.14
73
19.73
15
4.05
0
0.00
2
0.54
370

non-Jewish/non-Arab 26
46.43
14
25.00
13
23.21
3
5.36
0
0.00
0
0.00
56

Total 1307 1195 781 314 19 22 3638
Frequency Missing = 8

!
! 32
Figure 11. Distribution of Mackerel/Salmon/Sardines Consumption by Ethnicity
Groupings

1684
937
482
143
11 16
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Mackerel/Salmon/Sardines Consumption
0
500
1000
1500
Frequency of Participant Responses
non-Jewish/non-Arab Arabs Sephardi Ashkenazi Ethnicity
Distribution of Mackerel/Salmon/Sardines Consumption by Ethnicity Groups
!
! 33
Table 5. Mackerel/Salmon/Sardines Consumption by Ethnicity
Mackerel/Salmon/Sardines Consumption by Ethnicity
Ethnicity Mackerel/Salmon/Sardines
Frequency
Row Percent
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day Total
Ashkenazi 1275
51.87
681
27.71
365
14.85
114
4.64
7
0.28
16
0.65
2458

Sephardi 415
55.04
195
25.86
110
14.59
29
3.85
4
0.53
1
0.13
754

Arabs 200
53.76
117
31.45
46
12.37
8
2.15
1
0.27
0
0.00
372

non-Jewish/non-Arab 26
46.43
23
41.07
5
8.93
2
3.57
0
0.00
0
0.00
56

Total 1916 1016 526 153 12 17 3640
Frequency Missing = 6

!
! 34
Figure 12. Distribution of Cooked/Fried Fish Consumption by Ethnicity Groupings

Table 6. Consumption of Cooked/Fried Fish by Ethnicity
Table of Cooked/Fried Fish by Ethnicity
Ethnicity Cooked/Fried Fish
Frequency
Row Percent
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day Total
Ashkenazi 498
20.23
829
33.67
846
34.36
260
10.56
16
0.65
13
0.53
2462

Sephardi 87
11.52
131
17.35
431
57.09
98
12.98
7
0.93
1
0.13
755

Arabs 44
11.83
135
36.29
158
42.47
33
8.87
2
0.54
0
0.00
372

non-Jewish/non-Arab 7
12.50
17
30.36
24
42.86
6
10.71
1
1.79
1
1.79
56

Total 636 1112 1459 397 26 15 3645
Frequency Missing = 1
562
1005
1325
351
21
13
0/month 1-3/month 1/week 2-4/week 5-6/week 1/day
Cooked/Fried Fish Consumption
0
500
1000
Frequency of Participant Responses
non-Jewish/non-Arab Arabs Sephardi Ashkenazi Ethnicity
Distribution of Cooked/Fried Fish Consumption by Ethnicity Groups
!
! 35

Pairwise Correlation between Fish Variables:
Pairwise correlation between canned tuna, mackerel/salmon/sardines, and cooked and fried
fish were examined. All three pairwise comparisons (canned tuna vs. mackerel/salmon/sardines,
canned tuna vs. cooked/fried fish, and mackerel/salmon/sardines vs. cooked/fried fish) were
statistically significant (p<.0001). Canned tuna and mackerel/salmon/sardines had the largest
correlation estimate (R=0.38) followed by the correlation between mackerel/salmon/sardines and
cooked/fried fish (R=0.17) (Table 7). The weakest correlation was between canned tuna
consumption and cooked and fried fish consumption (R=0.11).
Table 7. Correlation between Fish Variables (heatmap codes are designated with identity
correlations=1 are highlighted in red, with correlations > 0.3 shown in yellow, and
correlations <0.3 shown in green)

Canned Tuna Mackerel/Salmon/Sardines Cooked/Fried Fish
Canned Tuna 1 0.38 0.11
Mackerel/Salmon/Sardines 0.38 1 0.17
Cooked/Fried Fish 0.11 0.17 1

Unadjusted and Adjusted Odds Ratios for Fish Consumption (never consume fish versus
consume fish):
The association between colorectal cancer risk and fish consumption was assessed using
three distinct fish consumption variables: canned tuna, mackerel/salmon/sardines, and
cooked/fried fish. First, these three variables were categorized from their original food
frequency levels to binary consumption outcomes: whether a participant ever consumes fish
versus participants who reported never consuming fish. Of the 3,646 participants, 3,638, 3,640,
and 3645 observations were used for the crude and multivariate model (adjusting for age, sex,
and ethnicity) for canned tuna, mackerel/salmon/sardines, and cooked/fried fish consumption,
respectively. Among the three categories of fish consumption, canned tuna was the only one to
show a protective association with colorectal cancer risk before adjusting for matching factors
and risk factors (crude odds ratio, 0.85; 95% confidence interval, [0.74, 0.97]). Neither
mackerel/salmon/sardines nor cooked/fried fish was significant at a 5% type I error rate. After
adjusting for age, ethnicity, and sex, the results were similar to those seen in the unadjusted
!
! 36
analyses. Adjusting for age, sex, and ethnicity, the odds ratio for canned tuna showed a
protective association with colorectal cancer risk (adjusted
1
odds ratio, 0.82; 95% confidence
interval, [0.71, 0. 94]) while mackerel/salmon/sardines and cooked/fried fish showed no
association with colorectal cancer risk. For colon cancer alone, canned tuna was associated with
a 16% reduction in risk of colorectal cancer (adjusted
1
odds ratio, 0.84; 95% confidence interval,
[0.72,0.97]). In the analysis of rectal cancer alone, canned tuna was associated with a 23%
reduction in risk of colorectal cancer (adjusted
1
odds ratio, 0.77; 95% confidence interval
[0.62,0.97]). After adjusting for known risk factors of colorectal cancer (as well as the matching
factors), none of the fish consumption variables showed a statistically significant association
with risk of colorectal cancer. Due to missing BMI values, the multivariate model adjusting for
both risk factors and matching factors was performed on a reduced dataset (3,265, 3,265, and
3,269 for canned tuna, mackerel/salmon/sardines, and cooked/fried fish, respectively).

Table 8. 2x2 Contingency Table for Canned Tuna
Cases Controls Total
Consume canned tuna 1,160 1,171 2,331
Never consume canned tuna 704 603 1,307
Total 1,864 1,774 3,638
Crude Odds Ratio [95% CI] 0.85 [0.74, 0.97]
*frequency missing: 8

Table 9. 2x2 Contingency Table for Mackerel/Salmon/Sardines
Cases Controls Total
Consume mackerel/salmon/sardines 859 865 1,724
Never consume
mackerel/salmon/sardines
1,006 910 1,916
Total 1,865 1,775 3,640
Crude Odds Ratio (95% CI) 0.90 [0.79, 1.02]
*frequency missing: 6

!
! 37

Table 10. 2x2 Contingency Table for Cooked/Fried Fish
Cases Controls Total
Consume cooked/fish fish 1,542 1,467 3,009
Never consume cooked/fried fish 326 310 636
Total 1,868 1,777 3,645
Crude Odds Ratio (95% CI) 1.00 [0.84, 1.19]
*frequency missing: 1

!
! 38
Table 11. Crude and Adjusted Associations between Fish Consumption and Risk of Colorectal Cancer

adj
1
Analyses are adjusted for age, sex, and ethnicity
adj
2
Analyses are adjusted for age, sex, ethnicity, BMI, family history of colorectal cancer, and red meat consumption

Table 12. Crude and Adjusted Associations between Fish Consumption and Risk of Colon Cancer
Fish Type OR
unadjusted
95% CI p-value n OR
adj
1
95% CI p-value n OR
adj
2
95% CI p-value n

Canned Tuna 0.86 [0.74,0.99] 0.04 3,218 0.84 [0.72,0.97] 0.02 3,218 0.90 [0.77,1.06] 0.21 2,904
Mackerel/Salmon
/Sardines
0.90 [0.79,1.04] 0.15 3,219 0.90 [0.78,1.04] 0.14 3,219 0.99 [0.85,1.15] 0.89 2,903
Cooked/Fried Fish 1.06 [0.88,1.27] 0.55 3,224 1.09 [0.90,1.31] 0.38 3,224 1.19 [0.97,1.46] 0.09 2,907

1
analyses are adjusted for age, sex, and ethnicity
2
analyses are adjusted for age, sex, ethnicity, family history of colorectal cancer, BMI, and red meat consumption

Fish Type
(colorectal)
OR
unadjusted
95% CI p-value n OR
adj
1
95% CI p-value n OR
adj
2
95% CI p-value n

Canned Tuna 0.85 [0.74,0.97] 0.02 3,638 0.82 [0.71,0. 94] 0.005 3,638 0.88 [0.76, 1.03] 0.10 3,265
Mackerel/Salmon
/Sardines
0.90 [0.79,1.02] 0.11 3,640 0.90 [0.79,1.02] 0.10 3,640 0.98 [0.86, 1.13] 0.83 3,265
Cooked/Fried
Fish
1.00 [0.84,1.19]

1.00 3,645 1.02 [0.86, 1.21] 0.82 3,645 1.12 [0.93, 1.35] 0.24 3,269

!
! 39
Table 13. Crude and Adjusted Associations between Fish Consumption and Risk of Rectal Cancer
Fish Type
(rectal)
OR
unadjusted
95% CI p-value n OR
adj
1
95% CI p-value n OR
adj
2
95% CI p-
value
n

Canned Tuna 0.83 [0.66, 1.03]
0.08
2,198 0.77 [0.62,0.97] 0.02 2,198 0.83 [0.65, 1.05] 0.12 2,010
Mackerel/Salmon
/Sardines
0.89 [0.72, 1.10] 0.29 2,200 0.90 [0.73,1.12] 0.35 2,200 0.97 [0.77, 1.22] 0.79 2,011
Cooked/Fried
Fish
0.85 [0.65,1.10] 0.22 2,202 0.85 [0.65,1.12] 0.25 2,202 0.95 [0.70, 1.28] 0.73 2,013

1
analyses are adjusted for age, sex, and ethnicity
2
analyses are adjusted for age, sex, ethnicity, family history of colorectal cancer, BMI, and red meat consumption

!
! 40
!
Conditional Logistic Regression: Analysis of Matched Pairs
Of the 3,646 participants in this study, 1,748 matched pairs were identified and indicated by a
pair identification variable, leading to a total of 3,496 participants in the sample (a 150 participant
reduction compared to analyses performed above). Using conditional logistic regression with proc
logistic, the associations between canned tuna, mackerel/salmon/sardines, and cooked/fried fish and
colorectal cancer were all examined (Table 14). After performing unadjusted conditional logistic
regression, there was a 16% decrease in risk of colorectal cancer for those who reported eating canned
tuna at least one or more times per month versus those who never consumed canned tuna (crude odds
ratio, 0.84; 95% confidence interval, [0.73, 0.97]). Mackerel/salmon/sardines consumption and
cooked/fried fish consumption did not have a significant association with colorectal cancer risk. These
results are quantitatively similar to the results from unconditional logistic regression when the matching
variables were adjusted for (Table 11). Compared to the unconditional logistic regression analysis, the
conditional logistic regression had modestly attenuated p-values likely due to the smaller number of
samples.

Table 14. Crude Associations between Fish Consumption and Risk of Colorectal Cancer using
Conditional Logistic Regression
Fish Type OR

95% CI p-value n

Canned Tuna 0.84 [0.73, 9.97] 0.02 3,488
Mackerel/Salmon/Sardines 0.89 [0.78, 1.02] 0.09 3,490
Cooked/Fried Fish 1.00 [0.83. 1.19] 0.96 3,495

Assessing Dose-Response of Fish Consumption:
To examine dose trends in the data, the fish consumption variables (canned tuna,
mackerel/salmon/sardines, and cooked/fried fish) were categorized by tertiles of consumption. The
tertiles were categorized as: 0 servings/month, 1-3 servings per month, and 1+ servings per week with 0
servings/month as the reference category. Contingency tables for the three fish variables are presented
below followed by the crude and adjusted odds ratios for canned tuna, mackerel/salmon/sardines, and
cooked/fried fish (Tables 15-18). Comparing canned tuna consumption 1-3 times per month versus 0
servings/month, canned tuna consumption showed no association with risk of colorectal cancer (for both
!
! 41
crude and adjusted analyses). Comparing the highest tertile of canned tuna consumption to the lowest
tertile, canned tuna was associated with reduced risk of colorectal cancer (adjusted
2
odds ratio, 0.76;
95% confidence interval [0.74, 0.90]). With increasing consumption of canned tuna, there was a
statistically significant protective trend in canned tuna consumption and colorectal cancer risk (adjusted
2

p-trend=.0009).
Similar results were seen when examining mackerel/salmon/sardines by tertiles of fish
consumption. There was no strong relationship between mackerel/salmon/sardines consumption and
risk of colorectal cancer for those participants who consumed this type of fish 1-3 times per month
compared to those participants who never consume mackerel/salmon/sardines. However, comparing the
highest tertile of mackerel/salmon/sardines consumption to the lowest tertile of
mackerel/salmon/sardines consumption, mackerel/salmon/sardines consumption showed a strong
protective association with colorectal cancer risk (adjusted
2
odds ratio, 0.73; 95% confidence interval
[0.60, 0.87]; p-trend < 0.0001). Participants who ate mackerel/salmon/sardines one or more times per
week had a 27% reduced risk of colorectal cancer compared to those who never ate
mackerel/salmon/sardines. Our data did not indicate a significant relationship between cooked and fried
fish consumption and risk of colorectal cancer.

Table 15. 3x2 Contingency Table for Tertiles of Canned Tuna Consumption
Cases Controls Total
1+/week 530 606 1,136
1-3 /week 630 565 1,195
0/month 704 603 1,307
Total 1,864 1,774 3,638
!
Table 16. 3x2 Contingency Table for Tertiles of Mackerel/Salmon/Sardines Consumption
Cases Controls Total
1+/week 308 400 708
1-3 /week 551 465 1,016
0/month 1,006 910 1,916
Total 1,865 1,775 3,640

!
! 42
Table 17. 3x2 Contingency Table for Tertiles of Cooked/Fried Fish Consumption
Cases Controls Total
1+/week 948 949 1,897
1-3 /week 594 518 1,112
0/month 326 310 636
Total 1,868 1,777 3,645

!
! 43

Table 18. Adjusted and Unadjusted Odds Ratios by Tertiles of Fish Consumption

!
adj
1
Analyses are adjusted for age, sex, and ethnicity
adj
2
Analyses are adjusted for age, sex, ethnicity, BMI, family history of colorectal cancer, and red meat consumption
!
Table 19. Adjusted and Unadjusted Odds Ratios by Tertiles of Mackerel, Salmon, and Sardines Consumption
Mackerel,
Salmon, &
Sardines
OR
unadjusted
95% CI OR
adj
1
95% CI OR
adj
2
95% CI
(n=3,640) (n=3,640) (n=3,265)
0/month 1.00 1.00 1.00
1-3/month 1.07 [0.92, 1.25] 1.07 [0.92, 1.25] 1.13 [0.96, 1.32]
1+/week 0.70 [0.59, 0.83] 0.69 [0.58, 0.82] 0.73 [0.60, 0.87]

p-trend 0.0008 <0.0001 <0.0001

adj
1
Analyses are adjusted for age, sex, and ethnicity
adj
2
Analyses are adjusted for age, sex, ethnicity, BMI, family history of colorectal cancer, and red meat consumption

Canned Tuna

OR
unadjusted
95% CI OR
adj
1
95% CI OR
adj
2
95% CI
(n=3,638) (n=3,638) (n=3,265)
0/month 1.00 1.00 1.00
1-3/month 0.96 [0.82, 1.12] 0.93 [0.79, 1.09] 1.02 [0.86, 1.21]
1+/week 0.75 [0.64, 0.88] 0.72 [0.61, 0.84] 0.76 [0.74. 0.90]

p-trend 0.0005 0.0002 0.0009

43
!
! 44

Table 20. Adjusted and Unadjusted Odds Ratios by Tertiles of Cooked/Fried Fish Consumption
Cooked &
Fried Fish
OR
unadjusted
95% CI OR
adj
1
95% CI OR
adj
2
95% CI
(n=3,645) (n=3,645) (n=3,269)
0/month 1.00 1.00 1.00
1-3/month 1.09 [0.90, 1.33] 1.10 [0.91, 1.34] 1.23 [1.00, 1.52]
1+/week 0.95 [0.79, 1.14] 0.97 [0.81, 1.17] 1.05 [0.87, 1.28]

p-trend 0.29 0.2577 0.08

adj
1
Analyses are adjusted for age, sex, and ethnicity
adj
2
Analyses are adjusted for age, sex, ethnicity, BMI, family history of colorectal cancer, and red meat consumption

!
! 45
Chapter 5: Discussion

Based on the data from this study, canned tuna and mackerel, salmon, and sardines are associated
with reduced risk of colorectal cancer. When consumed one or more times per week, a diet
characterized by consumption of canned tuna was associated with a 18% reduced risk of colorectal
compared to those who never consume canned tuna. A similar protective association was seen when
looking at canned tuna consumption and colon cancer risk as well as canned tuna consumption and
rectal cancer risk. For mackerel, salmon, and sardines consumption, when consumed one or more times
per week, consumption of these types of fish was associated with a 27% reduced risk of colorectal
compared to those who never consume mackerel, salmon, and sardines. There is no association with
cooked and fried fish consumption and risk of colorectal cancer.
Conflicting results have been published on the subject of fish consumption and risk of colorectal
cancer. Some research has suggested that the presence of omega-3 fatty acids in certain types of fish
provides protection against colorectal cancer by inhibiting colon carcinogenesis [26]. In particular, the
long-chain polyunsaturated fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)
have been theorized to have protective properties that reduce incidence of tumors in the colon [25].
This topic has been explored in animal studies. Comparing the incidence of colon adenocarcinomas
between rats that were fed a carcinogen followed by fish oil and rats that were fed a carcinogen and then
corn oil, Chang et al. found the fish oil group to have a significantly lower incidence of colon
adenocarcinomas [26]. This result has been inconsistent in human studies. In a study conducted in
Sweden, neither EPA nor DHA were found to have any association with colorectal cancer risk [28].
Conversely, significant protective associations have been found between omega-3 fatty acids
(specifically looking at EPA and DHA) and risk of colorectal cancer [27].
In this study, canned tuna was one of the specific types of fish that was examined for its association
with colorectal cancer. Traditionally preserved in either water or oil, canned tuna is a known source of
omega-3 fatty acids (Table 1). The preservation method, however, may have an impact on how much of
the omega-3 fatty acids are retained in the fish. When canned tuna is preserved in water, it retains more
of the omega-3 fatty acids from the fish due to hydrophobic nature of the oils in the fish. When canned
tuna is preserved in oil and then drained before consumption, it loses more of the omega-3 fatty acids
because of the mixing of the oils. Using nutritional information from the United States Department of
!
! 46
Agriculture (USDA), canned tuna preserved in water has two times the total amount of omega-3 fatty
acids compared to canned tuna preserved in oil (Table 1) [32]. With close to 1,000 mg of omega-3 fatty
acids per 100 gram serving, canned tuna preserved in water is a very good source of omega-3 fatty acids.
In addition to this, canned tuna (in water) is rich in both EPA and DHA, the two types of long-chain
omega-3 polyunsaturated fatty acids that have been hypothesized to be protective against colon
carcinogenesis.
Mackerel, salmon, and sardines are also fish species that are high in omega-3 fatty acids. For 100
grams of canned mackerel, there is approximately 1,400 mg of omega-3 fatty acids. Canned salmon has
a similar amount of omega-3 fatty acids compared to mackerel (for 100 grams of canned salmon, there
are 1,323 mg of omega-3 fatty acids). Among all types of fish species, sardines have one of the highest
omega-3 fatty acids content with over 1,400 mg of omega-3 fatty acids per 100 grams. Of the total
amount of omega-3 fatty acids in sardines, EPA and DHA comprise almost 1,000 mg of the omega-3
fatty acids [32]. Canned tuna, mackerel, salmon and sardines are all fish species with known amounts of
omega-3 fatty acids per serving. Cooked and fried fish, however, were the most frequently consumed
among participants in this study. Unfortunately, the generality of this type of categorization does not
allow for information on exactly what fish species are being cooked and fried. Israel has some
traditional fish dishes that are both ingrained in their culture and religious practices. Examining the fish
species that are commonly used in these dishes, as well as the availability of certain types of fish in this
region, we can infer what fish species are most commonly consumed in fried and cooked fish.
Among the fish species and preparation methods examined in this study, cooked and fried fish were
the most commonly consumed among all of the participants. Most participants reported consuming
cooked or fried fish at least once per week. Israel is a large producer of tilapia, or St. Peter’s fish, which
is mostly prepared by grilling and then garnishing with lemon [33]. In addition to tilapia, trout and gilt-
head sea bream are also commonly cooked fish in Israel [33].
The majority of the participants in this study were Ashkenazi Jews, an ethnicity known to have fish
as part of their cultural and holiday dishes. Specifically, Gefilte fish is a dish traditionally made from
carp, or another type of white fish, that are formed into balls and cooked in fish broth [33]. Pickled
herring is also a very commonly consumed fish among Ashkenazi Jews [33].
For Sephardi Jews, a dish known as hraime is made by braising grouper or halibut and cooking it in a
sauce with spices [34]. This dish is most often served during Jewish holidays such as Rosh Hashanah,
!
! 47
Passover, and the Sabbath [34]. This fish preparation method is not limited to just holidays. It is also
served in restaurants as well as cooked in homes in Israel.
Reviewing the omega-3 fatty acids content in each of these types of fish, it is clear that most of the
fish used for cooking and frying in Israel are lean, white fish. In particular, tilapia has a comparatively
small amount of omega-3 fatty acids compared to tuna, mackerel, salmon, sardines, and herring (for a
100 gram serving, tilapia has about 240 mg of omega-3 fatty acids) (Table 1). It is also uncertain as to
whether any sort of exposure to heat, through cooking, reduces the amount of omega-3 fatty acids in
these fish and the differences in the type of heat exposure. This data suggests a protective trend with
dark, fatty fish consumption and risk of colorectal cancer.
Compared to other studies that have examined the association between fish consumption and risk of
colorectal cancer, this study had almost complete information about fish species and preparation
methods. While canned tuna, mackerel, salmon, and sardines all had species and preparation-specific
information (with known omega-3 fatty acids content), the cooked and fried fish category was not fish-
species specific. Thus, any definitive conclusions on the amount of omega-3 fatty acids that comprise
this category cannot be made. Inferences about the fish species that make up the cooked and fried fish
category can be made based on general availability in Israel and traditional Israeli cuisine. Herring, a
fish that tops the chart in omega-3 fatty acid content was not asked about in this food frequency
questionnaire. It is, however, a known staple of diet in Israel, particularly smoked herring among
Ashkenazi Jews. This would be a fascinating fish to consider in future dietary studies because of the
protective association of fish rich in omega-3 fatty acids and risk of colorectal cancer mixed with the
potentially carcinogenic properties of smoked foods. While canned tuna is both a species-specific and
preparation method specific category, it did not differentiate between whether the tuna was preserved in
oil or water. As previously mentioned, tuna preserved in water retains almost two times the amount of
omega-3 fatty acids compared to tuna preserved in oil. Due to this difference, this would be another
category to stratify in future studies in order to at the impact of omega-3 fatty acids consumption on
colorectal cancer risk.
Other types of seafood, such as shrimp, crab, and lobster, have also been examined in the broader
investigation of this subject. For our study, shrimp and crab consumption was one of the items included
on the food frequency questionnaire. Almost 97% of the participants reported never eating shrimp or
crab therefore this would not be an informative population to make conclusions on the subject of
shellfish consumption and risk of colorectal cancer.
!
! 48
One limitation faced in this study, and in all case-control studies, is the possibility of recall bias.
That is, a participant with a cancer diagnosis may underreport or over-report on an aspect of their diet or
lifestyle that they believe may have contributed to their disease status. Fish typically have the reputation
for being nutrient rich and an overall good source of energy. This may have lead to the under-reporting
of fish consumption among cases. The food frequency questionnaire, however, had many food related
items thus it is highly unlikely that cases could be aware of a hypothesis regarding fish consumption and
their risk of colorectal cancer.
Finally, our study considers canned fish and cooked and fried fish separately. For the canned fish
(tuna, mackerel, salmon, and sardines) the grams of fish per serving can range from the size of the can.
The USDA reports canned tuna to come in cans that are about 170 grams. Sardines are often served in
smaller cans, around 90 grams per tin. Mackerel and salmon are often served in larger cans (around 300
grams per can). The cooked and fried fish component of the food frequency questionnaire looked at
100-150 gram servings of fish. With the differences in grams of fish consumed in each of the fish
categories, the amount of omega-3 fatty acids may be affected by how people view a serving of the
canned fish versus the cooked and fried fish.
In our data, mackerel, salmon, and sardines only showed the same strength of association with
colorectal cancer risk as canned tuna consumption when assessed by tertiles. This is not entirely in line
with the omega-3 fatty acids hypothesis since mackerel, salmon, and sardines are all higher in omega-3
fatty acids compared to canned tuna. Future studies could investigate these three fish species
independently to determine if the protective association is derived from one species relative to another
and if a certain threshold of consumption must be reached in order to elicit a protective association.
Our data indicate that there is a strong protective association between omega-3 fatty-acid rich fish
consumption and risk of colorectal cancer. Specifically, fish that are rich in docosahexaenoic acid
(DHA) and eicosapentaenoic acid (EPA), two types of omega-3 fatty acids, protect against colorectal
cancer. Animal studies and observational studies have produced similar results. More detailed fish
consumption frequency questionnaires (with specific fish species and preparation methods) will reveal
further information about this association.
!
! 49
References

1. World Cancer Research Fund International. Colorectal cancer statistics. 2013 [cited 2015 9
Feb]; Available from: http://www.wcrf.org/int/cancer-facts-figures/data-specific-
cancers/colorectal-cancer-statistics.
2. Globocan. Estimated Cancer Incidence, Mortality and Prevalence Worldwide in 2012. 2012
[cited 2015 9 Feb]; Available from:
http://globocan.iarc.fr/Pages/fact_sheets_population.aspx?country=860.
3. Barchana, M., I. Liphshitz, and P. Rozen, Trends in colorectal cancer incidence and mortality in
the Israeli Jewish ethnic populations. Fam Cancer, 2004. 3(3-4): p. 207-14.
4. Gruber, S.B., et al., BLM heterozygosity and the risk of colorectal cancer. Science, 2002.
297(5589): p. 2013.
5. Foulkes, W.D., et al., The founder mutation MSH2*1906G-->C is an important cause of
hereditary nonpolyposis colorectal cancer in the Ashkenazi Jewish population. Am J Hum Genet,
2002. 71(6): p. 1395-412.
6. Tenesa, A., et al., Genome-wide association scan identifies a colorectal cancer susceptibility
locus on 11q23 and replicates risk loci at 8q24 and 18q21. Nat Genet, 2008. 40(5): p. 631-7.
7. Raskin, L., et al., Characterization of two Ashkenazi Jewish founder mutations in MSH6 gene
causing Lynch syndrome. Clin Genet, 2011. 79(6): p. 512-22.
8. Peters, U., et al., Meta-analysis of new genome-wide association studies of colorectal cancer risk.
Hum Genet, 2012. 131(2): p. 217-34.
9. Schmit, S.L., et al., A novel colorectal cancer risk locus at 4q32.2 identified from an
international genome-wide association study. Carcinogenesis, 2014. 35(11): p. 2512-9.
10. Crous-Bou, M., et al., Polymorphisms in alcohol metabolism genes ADH1B and ALDH2, alcohol
consumption and colorectal cancer. PLoS One, 2013. 8(11): p. e80158.
11. Chaiter, Y., et al., Smoking attenuates the negative association between carotenoids consumption
and colorectal cancer risk. Cancer Causes Control, 2009. 20(8): p. 1327-38.
12. Khoury-Shakour, S., et al., Recreational physical activity modifies the association between a
common GH1 polymorphism and colorectal cancer risk. Cancer Epidemiol Biomarkers Prev,
2008. 17(12): p. 3314-8.
13. Poynter, J.N., et al., Statins and the risk of colorectal cancer. N Engl J Med, 2005. 352(21): p.
2184-92.
14. Rennert, G., et al., Use of bisphosphonates and reduced risk of colorectal cancer. J Clin Oncol,
2011. 29(9): p. 1146-50.
15. Rennert, G., et al., A case-control study of levothyroxine and the risk of colorectal cancer. J Natl
Cancer Inst, 2010. 102(8): p. 568-72.
16. Rennert, G., et al., Use of hormone replacement therapy and the risk of colorectal cancer. J Clin
Oncol, 2009. 27(27): p. 4542-7.
17. Laken, S.J., et al., Familial colorectal cancer in Ashkenazim due to a hypermutable tract in APC.
Nat Genet, 1997. 17(1): p. 79-83.
18. Niell, B.L., et al., Genetic anthropology of the colorectal cancer-susceptibility allele APC
I1307K: evidence of genetic drift within the Ashkenazim. Am J Hum Genet, 2003. 73(6): p.
1250-60.
!
! 50
19. Goldberg, Y., et al., An Ashkenazi founder mutation in the MSH6 gene leading to HNPCC. Fam
Cancer, 2010. 9(2): p. 141-50.
20. Slattery, M.L., et al., Dietary energy sources and colon cancer risk. Am J Epidemiol, 1997.
145(3): p. 199-210.
21. Le Marchand, L., et al., Associations of sedentary lifestyle, obesity, smoking, alcohol use, and
diabetes with the risk of colorectal cancer. Cancer Res, 1997. 57(21): p. 4787-94.
22. English, D.R., et al., Red meat, chicken, and fish consumption and risk of colorectal cancer.
Cancer Epidemiol Biomarkers Prev, 2004. 13(9): p. 1509-14.
23. Bostick, R.M., et al., Sugar, meat, and fat intake, and non-dietary risk factors for colon cancer
incidence in Iowa women (United States). Cancer Causes Control, 1994. 5(1): p. 38-52.
24. Sugawara, Y., et al., Fish consumption and the risk of colorectal cancer: the Ohsaki Cohort
Study. Br J Cancer, 2009. 101(5): p. 849-54.
25. Kantor, E.D., et al., Long-chain omega-3 polyunsaturated fatty acid intake and risk of colorectal
cancer. Nutr Cancer, 2014. 66(4): p. 716-27.
26. Chang, W.L., R.S. Chapkin, and J.R. Lupton, Fish oil blocks azoxymethane-induced rat colon
tumorigenesis by increasing cell differentiation and apoptosis rather than decreasing cell
proliferation. J Nutr, 1998. 128(3): p. 491-7.
27. Hall, M.N., et al., A 22-year prospective study of fish, n-3 fatty acid intake, and colorectal cancer
risk in men. Cancer Epidemiol Biomarkers Prev, 2008. 17(5): p. 1136-43.
28. Terry, P., et al., No association between fat and fatty acids intake and risk of colorectal cancer.
Cancer Epidemiol Biomarkers Prev, 2001. 10(8): p. 913-4.
29. Jedrychowski, W., et al., Protective effect of fish consumption on colorectal cancer risk.
Hospital-based case-control study in Eastern Europe. Ann Nutr Metab, 2008. 53(3-4): p. 295-
302.
30. Pham, N.M., et al., Fish consumption and colorectal cancer risk: an evaluation based on a
systematic review of epidemiologic evidence among the Japanese population. Jpn J Clin Oncol,
2013. 43(9): p. 935-41.
31. Wu, S., et al., Fish consumption and colorectal cancer risk in humans: a systematic review and
meta-analysis. Am J Med, 2012. 125(6): p. 551-9 e5.
32. United States Department of Agriculture. 3/8/2015 [cited 2015 03/07/2015]; Available from:
http://www.usda.gov/wps/portal/usda/usdahome.
33. Ganor, A. and R. Maiberg, Taste of Israel : a Mediterranean feast. 1990, New York: Rizzoli.
240 p.
34. Gur, J., The book of New Israeli food : a culinary journey. 1st American ed. 2007, New York:
Schocken Books. 303 p.

Abstract (if available)

Abstract The association between fish consumption and risk of colorectal cancer was investigated in a population-based, incidence density case control study from northern Israel, the Molecular Epidemiology of Colorectal Cancer study. Food frequency questionnaire data were used from 1,869 cases and 1,777 controls ascertained and recruited between March 31, 1998 and February 14, 2005. Questionnaire data included three specific fish categories: canned tuna, mackerel/salmon/sardines, and cooked/fried fish. Multivariate unconditional logistic regression was used to assess the association between each fish consumption variable and risk of colorectal cancer. After adjusting for age, sex, and ethnicity consumption of canned tuna was associated with reduced risk of colorectal cancer. Participants who consumed canned tuna had an 18% reduced risk of colorectal cancer compared to those who never consumed canned tuna (odds ratio, 0.82

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Red and processed meat consumption and colorectal cancer risk: meta-analysis of case-control studies

PDF

Disparities in colorectal cancer survival among Latinos in California

PDF

Genetic variation in the base excision repair pathway, environmental risk factors and colorectal adenoma risk

PDF

Meat intake, polymorphisms in the NER and MMR pathways and colorectal cancer risk

PDF

Racial/ethnic differences in colorectal cancer patient experiences, health care utilization and their association with mortality: findings from the SEER-CAHPS data

PDF

Development and validation of survey instrument designed for cervical cancer screening in Malawi, and other low resource settings

PDF

Genetic and environmental risk factors for childhood cancer

PDF

Carcinogen metabolism genes, meat intake, and colorectal cancer risk

PDF

Dietary carcinogens and genetic variation in their metabolism: epidemiological studies on the risk of selected cancers

PDF

Diet quality and pancreatic cancer incidence in the multiethnic cohort

PDF

Genes and environment in prostate cancer risk and prognosis

PDF

Investigating racial and ethnic disparities in patient experiences with care and health services use following colorectal cancer diagnosis among older adults with comorbid chronic conditions

PDF

The interplay between tobacco exposure and polygenic risk score for growth on birthweight and childhood acute lymphoblastic leukemia

PDF

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

PDF

Origins of the gender disparity in bladder cancer risk: a SEER analysis

PDF

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

PDF

Associations between isoflavone soy protein (ISP) supplementation and breast cancer in postmenopausal women in the Women’s Isoflavone Soy Health (WISH) clinical trial

PDF

Dietary and supplementary folate intake and prostate cancer risk

PDF

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

PDF

Pathogenic variants in cancer predisposition genes and risk of non-breast multiple primary cancers in breast cancer patients

Asset Metadata

Creator Sturgeon, Julia D. (author)

Core Title Fish consumption and risk of colorectal cancer

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publication Date 07/21/2015

Defense Date 07/21/2015

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

Tag colorectal cancer,fish,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Gruber, Stephen B. (committee chair), Gauderman, William James (committee member), Stern, Mariana C. (committee member)

Creator Email jsturgeo@usc.edu,julia.sturgeon91@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-601425

Unique identifier UC11300149

Identifier etd-SturgeonJu-3663.pdf (filename),usctheses-c3-601425 (legacy record id)

Legacy Identifier etd-SturgeonJu-3663.pdf

Dmrecord 601425

Document Type Thesis

Format application/pdf (imt)

Rights Sturgeon, Julia D.

Type texts

Source 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 a...

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