I

.. . J
‘ .. 6ka .3.
.. h. J ‘ . 1:

.3. M... a.
, «WW...
it?
: who. 1

uh:

‘fcéi'

72%

4.3

; .
""q““'
.A

‘ V

’1‘

. ‘ _
n . an”
. . . .91
.. _ 5

Fri!

3
I
iii

v
‘v
“2

@éffi .

t

It I
5mg.

I
y

r
1:...

5
J
~
‘ :
I I:
.

an.
n:

‘91::

?

gnu
.

in
s

a .
... ..
g.

W

. .
YunnuJfi

: ‘
1%.: . .

_ . {an

..

‘ . . IQ.

. 1%.”

«in. Y‘ ‘

. , . V . at... ‘m

V 2.:

.9. 29%.

. g

«a?

r... y
1.3.911"...
3. . 2
. 1.; I: .
2.. 58:.
, .. 2.3.7..
1.

7.2:.
.4323. . ‘
A 4mm, gs».

“an“ mmfifla. .. .
, . s. a a ‘
. .. iwxayrxwfl w u.
fins: .. rfivm
n3 :3 ...z
in... ; ,, 1....
9..“ . r ...

«r.
‘2 . i:

H ‘

I
, Laue...
5i; . .

2.: iv...

figééfigfi

r .;

:_ t z.

 

 

LIeRARY

Michigan State
University

 

 

 

This is to certify that the

thesis entitled

APC11307K AND THE RISK OF
COLORECTAL CANCER'IN ISRAEL

presented by

Joseph Bonner

has been accepted towards fulfillment
of the requirements for

_M_a_€;_t._§er'_degree in Wgy

\ ’CWOi' ' M ' '

Majoi) professor

Date December 12, 2000

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINE return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 c/chCJDmDmptss—p.“

APC Il307K AND THE RISK OF COLO—RECT AL CANCER IN ISRAEL

By

Joseph Donald Bonner

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

2000

ABSTRACT

APC I 1307K AND THE RISK OF COLO-RECT AL CANCER IN ISRAEL

Joseph Donald Bonner

Colo-rectal cancer (CRC) is the leading cause of cancer mortality in Israel Most
CRC patients have no known risk factors other than advanced age. In 1997, AFC
Il307K, a single nucleotide polymorphism, was discovered and identified as conferring
nearly a two-fold excess risk of CRC in carriers and six-fold excess risk in carriers with a
family history of CRC. In 1998, the Molecular Epidemiology of Colorectal Cancer
(MECC) Study was launched in Israel and the United States to investigate the genetic and
environmental risk factors for CRC. This thesis project analyzes data from the ongoing
MECC study. The preliminary MECC data show an odds ratio of 1.875 (95 %CI 0.795-
4.422) that is consistent with published relative risk estimates. This small and likely
underpowered analysis found no evidence for interactions or confounding between

APCIl307K and other known CRC risk factors.

ACKNOWLEDGEMENTS

To my bride Tracie, for her patience, consideration and gentle reminding, I am indebted.
Without which I would never have finished.

To my boss and friend Steve Gruber, I extend my gratitude and appreciation.

To the faculty and staff of the MSU Department of Epidemiology, I am very grateful for
their dedication and excellence.

iii

TABLE OF CONTENTS

1. LIST OF TABLES ...................................................................................................... vi
2. LIST OF FIGURES ................................................................................................... viii
3. LIST OF SYMBOLS AND ABBREVLATIONS ........................................................ ix
4. INTRODUCTION ........................................................................................................ l
4.1. Epidemiology of CRC ............................................................................................... 1
4.1.1. Geographic Differences in Incidence - The western lifestyle ................................ 1
4.1.2. CRC Risk changes in Migrants .............................................................................. 2
4.1.3. Ethnic Risk Differences for CRC ........................................................................... 3
4.1.4. Family history as a CRC risk factor ....................................................................... 4
4.1.5. Hereditary NonPolyposis Colorectal Cancer ......................................................... 5
4.1.6. Familial Adenomatous Polyposis (FAP) ................................................................ 6
4.2. APC structure and function ....................................................................................... 6
4.2.1. APCIl307K ............................................................................................................ 7
4.3. Other CRC Risk Factors ........................................................................................... 8
4.3.1. Aspirin and Non-Steroidal Anti-inflammatory Drugs ........................................... 9
4.3.2. Body Mass Index (BMD ....................................................................................... 10
4.3.3. Inflammatory Bowel Disease ............................................................................... 10
4.3.4. Meat consumption ................................................................................................ 10
4.3.5. Sedentary Lifestyle ............................................................................................... 12
4.3.6. Vegetable consumption ........................................................................................ 12
4.4. Measurement of diet in epidemiologic studies ....................................................... 13
4.4.1. The 24-hour food recall instrument ..................................................................... 13
4.4.2. The food diary instrument .................................................................................... 14
4.4.3. The food frequency instrument. ........................................................................... 14
4.4.4. Nutrient Databases ................................................................................. l6
5. HYPOTHESIS AND OBJECTIVES OF THIS STUDY ........................................... 17
6. METHODS ................................................................................................................ 18
6.1. The MECC Study .................................................................................................... 18
6.2. Laboratory Methods ................................................................................................ 19
6.3. The MECC food frequency instrument. .................................................................. 19
6.4. Statistical Methods .................................................................................................. 20

iv

TABLE OF CONTENTS cont’d

7. RESULTS ................................................................... 22

7.1. Univariate, Unmatched frequencies and Unadjusted Analyses ............................... 22
7.1.1. Demographic variables ......................................................................................... 22
7.1.2. APCIl307K .......................................................................................................... 23
7.1.3. Aspirin .................................................................................................................. 23
7.1.4. Body Mass Index .................................................................................................. 24
7.1.5. Ethnicity ............................................................................................................... 24
7.1.6. Inflammatory Bowel Disease ............................................................................... 24
7.1.7. Meat consumption ................................................................................................ 25
7.1.8. NSAIDS ............................................................................................................... 25
7.1.9. Relatives with CRC .............................................................................................. 26
7.1.10. Sports Participation ............................................................................................ 26
7.1.11. Vegetable Consumption ..................................................................................... 27
7.2. Unadjusted Analyses ............................................................................................... 27
7.2.1. Aspirin .................................................................................................................. 27
7.2.2. Body Mass Index .................................................................................................. 27
7.2.3. Vegetable Consumption ....................................................................................... 28
7.2.4. Other variables of interest. ................................................................................... 28
7.3. Adjusted Analyses ................................................................................................... 29
8. CONCLUSION .......................................................................................................... 3O
9. DISCUSSION AND LIMITATIONS OF THIS STUDY .......................................... 31
10. Further Study Needs ................................................................................................. 34
11. REFERENCES ........................................................................................................ 35

1. LIST OF TABLES

Table 1 Geographic difference in age-adjusted CRC Incidence ................................ 39
Table 2 Ethnic differences in CRC incidence in the US ......................................... 40
Table 3 Ethnic differences in Age-adjusted CRC Incidence in Israel .......................... 41
Table 4 Amsterdam Criteria ......................................................................... 42
Table 5 Studies of CRC and APCIl307K ......................................................... 43
Table 6 FFQ Items on the MECC Questionnaire ................................................. 44
Table 7 Logic to harmonize FFQ Frequency codes into weekly servings ..................... 46
Table 8 Gender among pairs ......................................................................... 47
Table 9 Age univariate statistics .................................................................... 48
Table 10 Education between groups ............................................................... 49
Table 11 Religion between groups .................................................................. '50
Table 12 Ethnicity between groups ................................................................. 51
Table 13 Frequency of cases and controls missing APCIl307K ............................... 52
Table 14 Unmatched analysis of APC11307K .................................................... 53
Table 15 Matched analysis of APC11307K ........................................................ 54
Table 16 Unmatched analysis of APC11307K among Ashkenazim ........................... 55
Table 17 Matched analysis of APC11307K among Ashkenazim ............................... 56
Table 18 Frequency of birth country among APC11307K Carriers ........................... 57
Table 19 Unmatched analysis of study variables — Aspirin .................................... 58
Table 20 Unmatched analysis of study variables - Body Mass Index ......................... 59
Table 21 Unmatched analysis of study variables Ethnicity ...................................... 60

vi

LIST OF TABLES cont’d

Table 22 Unmatched analysis of study variables — Inflammatory Bowel Disease ........... 61
Table 23 Unmatched analysis of study variables — Meat Consumption ....................... 62
Table 24 Unmatched analysis of study variables — N SAIDS ................................... 63
Table 25 Relatives queried on the MECC Questionnaire ....................................... 64

Table 26 Unmatched analysis of study variables — Relatives with cancer/Family History.. 65

Table 27 Family history of CRC among APCIl307K Carriers ................................. 66
Table 28 Unmatched analysis of study variables — Sports Participation ...................... 67
Table 29 Unmatched analysis of study variables — Vegetable Consumption ................. 68
Table 30 Matched analysis of significant study variables - Aspirin ........................... 69
Table 31 Matched analysis of significant study variables - Body Mass Index ............... 70
Table 32 Matched analysis of significant study variables - Vegetable Consumption ....... 71
Table 33 Matched analysis of non-significant study variables ................................. 72
Table 34 Multivariate analysis of APCIl307K and significant study variables .............. 74
Table 35 Final adjusted model ...................................................................... 76

vii

2. LIST OF FIGURES

Figure 1 APC Mutation frequency by codon ...................................................... 77

Figure 2 Map of Israel ................................................................................ 78

viii

3. LIST OF SYMBOLS AND ABBREVIATIONS

APC = Adenomatous Polyposis Coli protein and gene

APC11307K = a substitution of isoleucine for lysine at position 1307 of the APC gene
CRC = Colo-rectal Cancer

PHREG = SASTM Proportional Hazards Regression procedure

FAP = Familial Adenomatous Polyposis

HNPCC = Hereditary NonPolyposis Colorectal Cancer

BMI = Body Mass Index

MECC = Molecular Epidemiology of Colorectal Cancer

KHC = Klalit Health Services

COX2 = cyclooxygenase-2

ix

4. INTRODUCTION

In Israel, cancer of the colon and rectum is the leading cause of cancer mortality.
In the United States it is the second cause of cancer mortality. In Israel, there are an
estimated 2,000 new CRC cases annually (Israel Statistical Abstract, 1996). Across the
globe, CRC is third-most incident behind lung and breast cancers. Nearly 9% of new
cancer cases are CRC. (Landis, 1998)

In the United States there are estimated 150,000 new cases of CRC and 130,000
deaths from CRC annually. ( Landis, 1998) This background section will summarize the

global patterns in CRC epidemiology, CRC risk factors, and CRC molecular pathways.

4.]. Epidemiology of CRC

4.1.1. Geographic Differences in Incidence - The western lifestyle

Internationally, the age adjusted incidence rates of colon cancer vary from 34.1
per 100,000 Japanese men in Hawaii to 2.9 per 100,000 women in Bombay India. (Cancer
Incidence on Five Continents, 1992) Within this range between Hawaii and India, the
countries with the highest incidence rates are often considered “westemized” (T able 1).
Increased CRC incidence follows a “western lifestyle” around the globe. This “western
lifestyle” has the hallmarks of low physical activity, low vegetable/low fiber and high
meat/high fat consumption. Among the 24 global geographic regions with colon cancer

incidence data, Israel ranks 16‘“.

4.1.2. CRC Risk changes in Migrants

Migrant studies show that immigrants often adopt the health risks of a destination
population. The risk changes among migrants are evident in the 1985 work of King et a].
(King, 1985) They demonstrated that the age-adjusted mortality rate of large bowel
cancers among first-generation Chinese-Americans in the US from Guangzhou China
were 5.6 and 2.7 times the age-adjusted mortality rates of men and women remaining in
Guangzhou China. (King, 1985; Schottenfeld,l996) A 1991 survey of CRC mortality
rates showed that following 20 years of residency in the US, Japanese migrants to the
United States have a CRC mortality rate similar to US whites; while Japanese in Japan
have mortality and incidence rates significantly lower than in the US (W ynder, 1991;
Schottenfeld, 1996). Puerto Rican-born residents of New York City exhibit 2.5 times the
mortality and incidence rates of CRC than in Puerto Rico. At the same time this group
also exhibits 0.5 - 0.67 the CRC mortality and incidence rates of whites in New York
City. (Schottenfeld 1996; Warshauer, 1986). The picture of risk change in migrants is
not always as clear as the Chinese, Japanese and Puerto Ricans in the United States. In
Australia, the CRC incidence rate among Caucasians is 24.9 per 100,000 men and 25.6
per 100,000 women. (McMichael, 1980) Immigrants to Australia from Poland,
Yugoslavia, Italy and Greece, after 16 years of residence, show differing patterns of risk
changes and CRC mortality rate. The risk among Polish immigrants increased to the risk
of other Australian whites. The risks among Greek, Italian and Yugoslavian immigrants

were 0.7 times that of Australian whites. (McMichael, 1980, Schottenfeld 1996).

While the evidence of CRC risk changes in immigrants is not consistent, in
general, migrants from low-risk areas to high-risk areas acquire the risk of destination
areas. (Schottenfeld, 1996) These patterns of risk change indicate that CRC is a complex

disease involving environmental and lifestyle exposures.

4.1.3. Ethnic Risk Differences for CRC

CRC incidence rates vary within populations sharing a common geographic area.
Between 1986 and 1988, age-adjusted CRC incidence rates among the United States
white population was of 27.7 and 19.8 per 100,000 males and females respectively. In
the same time period, the age-adjusted CRC incidence rate among Blacks was 30.4 and
23.3 per 100,000 men and women respectively (Schottenfeld, 1996, MMWR, 1992).
Among the Hispanic population in the US, CRC incidence rates show marked sub-
population heterogeneity. Overall, Hispanics had an age adjusted CRC incidence rate of
13.6 and 8.9 per 100,000 men and women. Within the Puerto Ricans Hispanic sub-
population, the age-adjusted mortality rate is 18.9 and 13.8 per 100,000 men and women.
Within the Mexican sub-population, the age adjusted incidence rate is less than one-half
that of Puerto Ricans; one-third that of Whites and nearly one—quarter that of Blacks.
(Schottenfeld, 1996; Warshauer, 1986) (Table 2)

In Israel, the colonic and rectal cancer incidence rates vary by six-fold among
ethnic groups. The population of Israel is 81% Jewish and 19% non-Jewish (Arab,
Christian and other non-Jewish religions). (Statistical Abstract of Israel, 1996). The

highest incidence rate is 21.1 per 100,000 Jews born in Europe or America and the lowest

incident rate is 3.4 per 100,000 non-Jews. (Cancer Incidence on Five Continents 1992;
Schottenfeld,l996). (Table 3).

This striking degree of ethnic variation suggests that ethnic specific differences
may account for some of the variability in rates of CRC. Well-documented evidence of
founder mutations in tumor—suppressor genes is consistent with the hypothesis that
genetic susceptibility contributes to ethnic variation for some cancers (e. g. Breast Cancer
and BRCA1 and BRCA2) (Strewing, 1997). However other ethnic-specific lifestyle

factors are equally plausible as explanations for these observed differences.

4.1.4. Family history as a CRC risk factor

Individuals who have a first degree relative with CRC appear to be at two to three
times background risk for CRC. (Potter, 1999; Schottenfeld, 1996, St. John 1993). About
15-20% of all CRC cases occur in families where another relative is affected. (Cannon-
Albright, 1988; Gryfe, 1999). There are two highly penetrant autosomal dominant
familial CRC syndromes: hereditary non-polyposis colorectal cancer (HNPCC) and
familial adenomatous polyposis (FAP), which account for about 7% and 1% respectively
of all CRC burden. (Potter, 1999; Schottenfeld, 1996). It has been well established for
some time that a family history of CRC is a risk factor. With this common knowledge,
individuals with a family history might seek screening more rigorously than individuals
with a negative family history. Also physicians probably recommend more intensive

screening in individuals with a positive family history. Intensive screening efforts in the

population with a positive family history, probably augments the effect of family history

as a CRC risk factor.

4.1.5. Hereditary NonPolyposis Colorectal Cancer

HNPCC is characterized by early onset CRC in individuals of families where
cancers of other body systems are also present. CRC tumors in HNPCC patients are
histopathologically distinct from sporadic CRC. These tumors typically demonstrate poor
differentiation, a mucinous appearance, lymphocytic infiltration, histologic heterogeneity,
and Signet ring features. Tumors in HNPCC patients also have a distinct molecular
fingerprint. The somatic mutations within HNPCC tumors demonstrate a loss of DNA
replication error detection and correction capacity, termed microsatellite instability.
(Lynch, 1985). Microsatellite instability is the phenotypic expression in cells arising as a
consequence of mutations in five genes responsible for mismatch repair: MSH2, MHLl,
MSH6, PMS], PMSZ. (Kolodner, 1995). HNPCC families also have individuals with
cancers in other body systems: endometrium, stomach, upper urinary tract, small bowel,
ovary, and bile ducts (Lynch, 1985). HNPCC is classically recognized by "Amsterdam
Criteria" of the International Collaborative Group on HPNCC (Vassen, 1991). (Table 4).
A recent report suggests that the microsatellite instability negative, HNPCC—suspected
cases might be caused by mutations in the T GF-Beta type II receptor gene (Lu, 1998). It
is thought that the TGF—Beta type 11 receptors work with APC, B-catenin and E-cadherin

to facilitate inter-cellular communication and cellular adhesion. (Lu, 1998)

4.1.6. Familial Adenomatous Polyposis (FAP)

Familial adenomatous polyposis (FAP) is characterized by hundreds or thousands
of adenomas that blanket the large intestine. This form of polyposis usually becomes
evident in patients before age 35 and is usually identified in adolescence. In the absence
of a colectomy, most patients with FAP will develop CRC by the age of 50. FAP
patients can also present with pigmented ocular fundus lesions, super-numerary teeth,
ostemoas, odontomas, dermoid tumors, epiderrnoid cysts, and follicular thyroid cancer.
FAP is found in 1 out of every 8000 persons with CRC. (Foulkes, 1995) In 1987, the
locus for FAP was mapped to chromosome 5q21 (Bodmer, 1987;Leppert, 1987). In
1991, the APC (adenomatous polyposis coli) gene was cloned. Studies of this protein
have demonstrated that inactivating gerrnline mutations of APC are responsible for FAP.
(Kinzler, 1987;Fodde,1991;Potter, 1997). Attenuated FAP is clinically similar to FAP but
is characterized by fewer polyps. Attenuated FAP is associated with an inactivating
mutation within the first 158 codons of the APC gene. The multiple clinical signs of FAP
and attenuated FAP appear to be associated with mutations at different loci within the

APC gene. (Giardello, 1997).

4.2. APC structure and function

The APC protein contains 2843 amino acids encoded across 15 exons.
(Cunningham, 1996) The APC protein is expressed in colonic epithelium (Cunningham,
1996). Germline mutations in various positions of the gene are associated with different

FAP manifestations and different types of CRC. A recent study identified the region

between codon 1300 and 1500 as a mutational hot spot for the germline mutations of FAP
and the somatic mutations identified in sporadic CRC. (Bodmer, 1999). Figure 1

APC has regions that correspond to B-catenin binding sites. B-catenin interacts
functionally with E-cadherin. Together this triad (APC, B-catenin, E-cadherin and
possibly TGF-Beta type receptors) acts as a gatekeeper for cell division and cell death.
(Kinzler, 1997) Mutations in APC result in an inability to bind and degrade B-catenin,
causing intracellular B-catenin levels rise. High B-catenin levels turn on transcription
through the TCF-LEF system and a cascade of downstream proteins, including myc, are
expressed. These downstream proteins, in turn are responsible for the unregulated
cellular growth of polyps. Colonic epithelial cells with APC mutations fail to
differentiate and fail to migrate from the base of the crypt thereby clustering as polyps. It
is hypothesized that as polyps grow, APC mutated cells are exposed to the mutagenic
environment of the lumen where they are likely to become a cancerous lesion.

(Rubinfeld, 1987; Su, 1993; Bodmer, 1999; Potter, 1999).

4.2.1. APCIl307K

In 1997, the work of Laken et a1. identified a germline T to A tranversion of codon
1307 of the APC gene that was transmitted in an autosomal dominant manner in a family
with an increased susceptibility to colonic neoplasia. (Laken, 1997) This polymorphism is
a functional change from isoleucine to lysine in the APC protein. APC11307K appears to

make cells highly susceptible to somatic mutations within the APC gene. (Laken, 1997)

The initial clinical case-control study of the polymorphism showed a relative risk
for CRC of 1.8 among Ashkenazim CRC cases and Ashkenazim and non-Ashkenazim
controls. This initial study did not find the polymorphism in the 243 cancer-free, non-
Ashkenazim that they examined. (Laken, 1997) The published studies of APCIl307K
and CRC show odds ratio estimates of 1.78, 1.887 and 3.4 (Laken 1997;Woodagel998;
Drucker,2000). Of the studies that publish ethnicity values of subjects they show a carrier
rate among Ashkenazim of 4.5%, 10%, 10%, 13% and 13.5%. (Petrukhin, 1997;Gryfe,
1999;Woodage, 1998; Drucker,2000). (Table 5) Since the initial publication in 1997, the
polymorphism has also been discovered in Sephardim. (Drucker, 2000) One study of
APC11307K and familial CRC and breast cancer in 393 Norwegian patients found the
single Jew in the study to be the single carrier. This individual was of Ashkenazi decent,
affected with colon cancer and reported no first or second-degree relatives with either

colon or breast cancer. (Lothe, 1998).

4.3. Other CRC Risk Factors

All CRC is the result of either somatic or germline genetic changes. The known
inherited contributions to CRC genetics represent a small fraction of the CRC burden.
(Kinzler, 1987) Other known risk factors for CRC include: advanced age, inflammatory
bowel disease, a diet low in fiber and high in fat, and a sedentary lifestyle. There are
other risk factors that are significant yet less consistent. The other risk factors include:
excessive alcohol, smoking, changes in body weight, reproductive history, exogenous

estrogens, and specific dietary components. (Potter, 1999). Aspirin use and non-steroidal

anti-inflammatory drug use have been identified as a protective factor against CRC.
(Potter 1999) However, most patients with CRC exhibit no known or suspected risk
factors other than age.

Most CRC patients are diagnosed when they are above 60 years old (Cancer Facts
and Figures, 2000) . Very few individuals die of a competing cause of death with latent
CRC. A recent autopsy series predicts the fraction of death of all causes with latent CRC
to be 2.5% (6 / 231) (Kawaharada, 1998). The population in this series was all

Japanese where the incidence of CRC is low but increasing. (Schottenfeld, 1996)

4.3.1. Aspirin and Non-Steroidal Anti-inflammatory Drugs

Individuals who use non-steroidal anti-inflammatory drugs (NSAIDS) appear to
have less risk of developing CRC. (Potter, 1999) In rats, aspirin and other NSAIDS
inhibit the production of COX2 and decrease the incidence rate of gastrointestinal
neoplasia. COX2 promotes carcinogenic processes and specific COX2 inhibitors reduce
both the number of and size of polyps in both animal and human studies.(Potter, 1999)
The evidence generally shows a strong protective effect of aspirin but not all of the
evidence is consistent. As of mid-1998, seven case control studies of aspirin and 4
cohort studies of aspirin have shown a significant effect on reducing CRC mortality rates.
(Potter, 1999). However, an intervention trial of aspirin analyzed as a randomized trial
that showed a strong protective effect, when analyzed as a follow-up study showed no
association between aspirin and CRC mortality rates. (Paganinni-Hill, 1989;Gann, 1999;

Sturmer, 1998).

4.3.2. Body Mass Index (BMI)

Body Mass Index (BMI) measures obesity. Epidemiologic studies have
demonstrated weak and inconsistent associations between elevated BMI and elevated risk
for CRC. Data from the Health Professionals Follow-up study was examined to address
the issue. (Giovanucci, 1995) In 1995 Giovanucci et al. reported on a large (2000 cases
and 2400 controls) multi-site study of CRC in the US, men with the largest body mass,
lowest physical activity and highest energy intake demonstrated an odds ratio of 7.3
(95 %CI 3.4-5.2) for CRC when compared with the opposite extremes of these variables.
In the Giovanucci study there was little association between these individual variables
(exercise, BMI and energy intake) and CRC, however the highest level of each variable,

versus the lowest of each variable, demonstrated a strong association. (Giovanucci, 1995)

4.3.3. Inflammatory Bowel Disease

Inflammatory Bowel Disease is a clinical condition consisting of ulcerative colitis
and/or Crohn’s disease. Individuals with either one of these conditions are at 10—20 times
elevated risk for CRC. In these conditions colonic epithelial cells are in a continual state
of inflammation. It is thought that these cells are at greater risk for genetic changes to
occur and be replicated. In the mutagenic environment of intestinal lumen, these cells are

likely to acquire a cancerous mutation. (Potter, 1999)

4.3.4. Meat consumption

A diet high in meat consumption appears to increase the risk of CRC.

Specifically, eating red meats prepared at high temperatures or otherwise processed

10

shows elevated risks of CRC (Potter, 1997). Red meats prepared at lower temperatures
show somewhat elevated risks, however between studies, the evidence is inconsistent
(Potter, 1997). White meats shows inconsistent relative risks, some studies show modest
risk elevations; others show risk reductions (Potter, 1997). According the World Cancer
Research Fund, there were seven cohort studies of meat consumption published by 1997.
(Potter, 1997) A study of Seventh-day Adventists, a population of mostly vegetarians,
found no association with meat consumption and CRC. (Phillips, 1985) In the Nurses
Health Study, frequent vs. rare red-meat consumption showed a relative risk of 2.5 (1 .2-
5.0). (W illett, 1990) When consumption of all types of meat are examined collectively
there is little or no elevated risk of CRC; however when individual types of meat and
different types of preparation are examined the evidence points toward an association
with CRC. Commercially prepared meat products like sausage show elevated risks
similar to high-temperature meat preparation. (Potter, 1997)

The biological mechanisms by which meats promote cancerous growth are
unclear. There have been two mechanisms postulated, 1) while meat is prepared at higher
temperatures, carcinogens are formed (Potter, 1997; Potter, 1999) and 2) the consumption
of meats causes the body to produce bile which has a promoting effect of CRC on colonic
cells (Potter, 1997; Potter, 1999)

According to the World Cancer Research Fund, by 1997, of the 86 studies on the
relationship between meat consumption and CRC, 47 show relative risks above unity, 31
show risks consistent with unity and eight report a protective effect. This group makes

the following statement regarding meats and cancer of the colon and rectum:

11

“It is unclear whether the specific mechanisms of increased risk associated
with meat intake involve animal fat, processing and cooking methods or
other factors. The evidence shows that red meat probably increases risks
and processed meat possibly increases risk of colorectal cancer.” World
Cancer Research Fund (Potter, 1997) Page 246.

“Cooking meat at high temperatures possibly increases the risk of colorectal
cancer.” World Cancer Research Fund (Potter, 1997) Page 251.

4.3.5. Sedentary Lifestyle

In many studies men reporting high levels of physical activity are at a lower risk
of colon cancer. There is no evidence for an association between rectal cancer and
physical activity. The evidence in women is inconsistent but tends to lean on the side of
no association. Upon examination of physical activity at various periodsof one’s lifetime,
vigorous activity throughout one’s lifetime is associated with lower risks than physical
activity in young adulthood or older adulthood. It is thought that rigorous physical activity
stimulates or assists peristalsis and thus decreasing transit time. As transit time is
decreased, mutagens within the lumen are expelled faster and susceptible cells have less

exposure. (Giovanucci,l995; Potter, 1997; Potter, 1999)

4.3.6. Vegetable consumption

According to the World Cancer Research Fund, by 1997 there were four published
cohort studies and 21 published case-control studies examining the association between
vegetable consumption and risk of CRC. Of these studies, 17 show some reduced risk of
CRC with increased vegetable consumption. (Potter, 1997) When specific vegetable
items are examined, most studies show the protective effect of broccoli to be greater than

all vegetables combined. (Potter, 1997) Most studies also show raw vegetables to have a

12

greater protective effect than cooked vegetables. (Potter, 1997) A postulated biologic
mechanism for the protective effect of vegetables involves roughage in the lumen forcing
cells to be removed rather than waiting for cells to slough. Thus cells have less time
exposed to the mutagenic environment and are less likely to become a cancerous lesion.
(Potter, 1997) It is also postulated that vegetables contain a vast array of chemicals that
the body needs to bolster natural biological protection against cancer (Potter, 1997).

The World Cancer Research Fund makes the following statement regarding vegetable
consumption and risks of CRC:

“Evidence that a diet rich in vegetables protect against cancers of the colon and
rectum is convincing.” World Cancer Research Fund (Potter, 1997) Page 239.

4.4. Measurement of diet in epidemiologic studies

In the study of human chronic disease etiology, measurement of diet is of
paramount importance, yet is highly difficult. In modern epidemiology, there are three
basic instruments used to measure diet, namely: 24-hour food recall, food diary and food

frequency. Each instrument has both positive and negative aspects.

4.4.1. The 24-hour food recall instrument

The 24-hour food recall instrument asks subjects to recall all food items and
quantities consumed in the past 24 hours on a randomly chosen day, and can be self or
interviewer-administered. When interviewer administered, research subjects are queried
for all items consumed over the previous 24-hour period. This instrument can only gain
information about the most current dietary patterns. Some obscure, yet meaningful, items

can be missed. For example a research subject mi ght eat a food item of particular interest

13

regularly but just happened to not eat it in the reference period. When using the 24-hour
recall instrument, users estimate the portion sizes and preparation methods of foods eaten.
This instrument is often used when the period of time of interest is short and when the
interview time from subjects is short and the degree of comrrritrnent of research subjects
is limited. This instrument might work better in some cultures than in others. (W illett,
1990) This instrument has an underlying assumption that food consumption patterns are

similar in day to day comparisons.

4.4.2. The food diary instrument

The food dietary instrument asks subjects to document all food consumed in
accurate portion sizes. Research subjects typically have a printed form with them at each
meal and document every item that they consume during a given time period. Research
subjects must have a high degree of commitment to the project since the labor required by
subjects using a diary type of instrument to be cumbersome. In the measurement of diet in
chronic disease etiology, the dietary diary is considered the gold standard. This
instrument is also subject to the same criticisms as the 24-hour recall in that items

consumed with frequency of less than 1 per day are seldom found. (Willett, 1990)

4.4.3. The food frequency instrument.

The food frequency instrument queries subjects on their typical consumption
patterns of a given list of items over a given period of time. It is commonly employed in
chronic disease research. It can be self— or interviewer- administered. It takes

comparatively little time compared with the dietary diary instrument but more time than a

14

24-hour recall. Food frequencies are recorded as number of times per month, per week or
per day a specific food in a given portion item is consumed. This instrument is frequently
used because it is easy to administer and can cover a broader period of time than other
instruments. However this instrument is subject to recall bias in that users are often
unable to accurately recall consumption frequency of some items. (W illett, 1990) In
studies of chronic diseases, like cancer, the period of exposure is often lengthy.
Constructing or administering a food frequency instrument to query on the biologically
relevant period of time is very difficult and often of questionable validity. (Willett, 1990)
When .a food frequency instrument is used it is often subjected to a measure of its
reliability. Subsets of subjects are asked to complete a diary instrument as well as the
food frequency instrument. Results of macronutrients consumed as measured by each
instrument are correlated. If the macronutrient values consumed as measured by a food
frequency instrument correlates with those measured by diary method then the food
frequency method is then considered reliable. Often a food frequency instrument will be
subjected to validity tests as well as reliability tests. When validity is being tested, a
measure of a micronutrient is measured from a subject’s blood. The results are then
correlated with the micronutrient values calculated from the food frequency instrument.
When the instrument and the blood levels correlate, then the food frequency instrument is

considered valid. (Willett, 1990)

15

4.4.4. Nutrient Databases

When any dietary instrument is used in epidemiologic research, the
macronutrients (calories, carbohydrates, fat and protein) and micronutrients (vitamins and
minerals) can be referenced with every item consumed. Typically researchers use a
commercially available database of common food items and their individual components
to estimate subject’s consumption of macro and micronutrients. However, when the
research population is from a different or broad geographic area, nutrient databases
prepared for one group or area might not be comparable to another group or area.
Frequently when unique'populations are being researched, the general nutrient databases
are not appropriate and specific nutrient databases must be developed and used. (Willett,

1990)

16

5. HYPOTHESIS AND OBJECTIVES OF THIS STUDY

The objectives of this thesis project are: 1) an examination the relative risk
estimate of APC11307K and CRC in the MECC study. 2) an evaluation of the relative
risk estimate in the presence of other risk factors for CRC. Each individual risk factor
(aspirin, body mass index (BMI) ethnicity, inflammatory bowel disease (IBD), meat
consumption, NSAIDS, relatives with cancer, sports participation and vegetable
consumption) will be tested individually for association with CRC. The variables
presenting significant odds ratio estimates will be tested in an adjusted model with
APC11307K.

This thesis project will test these null hypotheses 1) there is no association of
APC11307K and CRC and 2) an association of APC11307K and CRC is independent of

the relative risks of other risk factors and CRC.

17

6. NIETHODS

6.1. The MECC Study

The Molecular Epidemiology of Colorectal Cancer (MECC) study is an ongoing
population-based case control study aimed at measuring the cancer risks associated with
APC11307K.

The sampling frame for the MECC study is Northern Israel (Figure 2), which
includes the Northern and Haifa governmental districts. All men, women and children
who are newly diagnosed with CRC from five major hospitals in northern Israel are
eligible cases. The five major hospitals are Carmel, Rambarn, Nahariya, Ben Zion and
Afula. More than 80% of all CRC among patients who live in northern Israel are
diagnosed at these five hospitals. Controls are gleaned from the membership list of Klalit
Health Services (KHS), the largest health maintenance organization in Israel covering
65% of the population of the country.

Newly diagnosed patients with CRC are invited to participate. After obtaining
informed consent they are invited to donate blood samples and answer interview
questions. Following a case interview, a list of 10 controls is generated for each case from
the database of KHS subscribers. Controls are individually matched to cases based on
exact year of birth, gender and clinic code. The clinic code offers a match for geographic
region of residence of the case. Each individual from the list is contacted until one
agrees to participate. Controls are invited to participate, asked to give informed

consented, asked to donate blood and interviewed at a scheduled appointment.

18

When cases are not members of KHS, they are still invited to participate. The
KHS clinic near the residence of the non-KHC cases is used to match for a control.
Geographic regions and even neighborhoods within Israel are ethnically homogenous.
Individuals live near other individuals of their religion and ethnicity. Matching on clinic
code and thus geographic area potentially matches on ethnicity.

The MECC questionnaire contains 1071 questions. It includes questions on
demographics, in-depth health history, medication health history, nutrition history, four-
generation pedigree with specific cancer questions and a 171-item food frequency

questionnaire. Trained interviewers administer the questionnaire instrument in person.

6.2. Laboratory Methods

In Israel, DNA is extracted from subject’s blood and shipped to Ann Arbor,
Michigan. Molecular assays for APC11307K are preformed in the laboratory of Dr.
Stephen Gruber at the University of Michigan Medical School following the protocol of

Laken et a1. 1997. (Laken, 1997)

6.3. The MECC food frequency instrument.

The MECC food frequency instrument follows the prototype of Willet's 189—item
food frequency questionnaire. It queries subjects on 171 food items common in the
Israeli diet. Each item is presented within groups of like items and with common serving
sizes. Table 6 lists the items within the meat and vegetable sections. This instrument is a

close adaptation of Willet’s instrument but its reliability and validity are currently being

19

studied. Currently a data source of micro- and macronutrients corresponding to this
instrument is under development and unavailable for this analysis.

The MECC food frequency instrument was delivered in two forms. The first form
allowed users to select a frequency as one of nine coded classifications; a second form
offered seven coded classifications. The form with seven frequency categories was
administered to 91/374 (24%) of subjects. The instrument with nine frequency categories
was administered to 283/374 (76%) of subjects. For this analysis the scores for the
individual FFQ items are harmonized into servings per week following the logic in Table

7.

6.4. Statistical Methods

All analyses are presented in matched and, unmatched contingency tables. The
unmatched tables are used to illustrate the frequency distribution of specific factors
among cases and controls. The matched tables are used to display the matching within
pairs and calculate the matched odds ratio. Contingency tables, matched and unmatched
are both produced by SAS Proc FREQ. Given the matched design of the study,
conditional logistic regression was required for multivariate analysis. Conditional logistic
regression modeling was performed using SAS PHREG. This procedure is typically
used to calculate proportional hazards in a survival analysis; however, it can be used to
calculate the conditional logistic regression models required for a matched analysis. In
survival analysis, a time variable and a censoring value are used. When the PHREG

procedure is used for conditional logistic regression, a one-unit change in the time

20

variable is used. Cases have the value of 1 and controls have a value of 2 for the dummy
time variable. The model building process starts with the dummy time and censoring
variables and APC11307K and following a forward selection strategy introduces the
statistically significant variables found in the bi-variate analyses.

This analysis includes three continuous variables (Body mass index , weekly
servings of meat and weekly servings of vegetables). Any value equal to or greater than
the mean of these variables plus or minus three standard deviations of the value in the

controls is considered an outlier.

21

7. RESULTS

7 .1. Univariate, Unmatched frequencies and Unadjusted Analyses

7.1 . 1. Demographic variables

The MECC project as of March 1, 2000 had 610 CRC cases and 260 Controls.
This analysis is on the 219 case and control pairs for which data were available. Given
the matched design of the study, cases and controls are not different in age and gender.
In this analysis there 112/219 (51%) male matched pairs and 95/219 (43.4%) female
matched pairs. Gender as a variable in these data was missing for 12/219 (5.5%) of the

matched pairs. (Table 8)

The age of the matched pairs ranged from 22-98. The mean values 71.8 and the
standard deviation were 10.5. The inner quartile range was 67-78. (Table 9)
Cases and controls were also not different between each group in country of birth,
religion or ethnicity. However, a statistically significant difference between each group
was evident in years of education. Controls were more educated than cases. p=0.0346.
(Table 10)

In this study population of 438 subjects 369 (96.6%) were Jews, 6 (1.6%) were
Christian Arab, 1 was a Moslem Arab and 2 were Christian non-Arabs. (Table 11)
Among the Jews, 296/369 (80%) were of the Ashkenazi ethnic group, 71/369 (19%) were

Sephardi and 2/369 (1%) were of mixed ethnicity. (Table 12)

22

7.1.2. APCIl307K

Of the 438 subjects in the 219 matched pairs, APCIl307K results were available
for 183/219 (84%) of cases and 195/219 (89%) of controls. (Table 13) The APC11307K
polymorphism was found in 28/378 (7.4%) of all subjects 18/ 183 (9.8%) of all cases and
10/195 (5.1%) of all controls. (Table 14) Among the Ashkenazim, APC11307K was
found in 13/ 127 (10%) of cases and 7/ 126 (5.5%) of controls. (Table 15). The relative
risk of CRC and APC11307K in this matched analysis was 1.875 (95%CI 0.795 - 4.422).
(Table 15) This value, although not statistically significant, is consistent with published
relative risk estimates as outlined in Table 5. In the unmatched analysis the odds ratio of
APC11307K and CRC among all ethnicities was 2.0 (0.9-4.5). (Table 15) Among
Ashkenazim the unmatched odds ratio was 1.9386 (0.7467 — 5.331). (Table 16) Each of
these odds ratio values is consistent with the published odds ratio estimates. There were
two discordant pairs among Ashkenazim matched pairs for a matched analysis. (Table
17). Thus a matched relative risk estimate among the Ashkenazim is not available.

Of the reported countries of birth reported, Lithuania had the highest carrier rate
of 1/3 (33.3%), followed by Germany with 5/ 17 (29.4%). The remaining countries

reported in the study had carrier rates of less than 10%. (Table 18).

7.1.3. Aspirin

The MECC questionnaire includes the following four questions “Have you taken
aspirin regularly within the past year?” and “Have you ever taken aspirin at least once a

week during your life?”. If a subject answers yes to either of these aspirin questions, in

23

this analysis they were considered positive for aspirin. If individual answers no to both
questions they were coded as negative. The variables were missing for 49/219 (22.4%) of
cases and 43/219 (19.6%) of controls and in 33/219 (15%) of cases. Among cases

77/176 (44%) and among controls 62/186 (33%) reported positive for aspirin. (Table 19).

7.1.4. Body Mass Index

Body Mass Index was calculated as weight*height'2. The mean BMI among cases
was 26.4 kg"‘cm’2 (+\-3.7). The mean BMI among controls was 25.8 kg"‘cm‘2 (+\—4.3).
The cutoff point for outliers was above 38.7. Within both groups the mean BMI was
26.1 (+/- 4.0). One case reporting BMI of 39.5, and three controls reporting BMI of 43.4,

39.2 and 40.1 were removed as outliers. (Table 20)

7.1.5. Ethnicity

In this analysis, the ethnicity was analyzed as Ashkenazi and non-Ashkenazi.
Among cases 153/219 (70%) were Ashkenazi and 62/219 (28%) where non-Ashkenazi
cases; among controls 143/219 (63%) were Ashkenazi and 76/219(35%) non-Ashkenazi
controls. Ethnicity was not associated with CRC risk in this analysis (OR=1.39

95%CI=0.89-2.18). Table 21

7.1.6. Inflammatory Bowel Disease

The MECC questionnaire asks: “Have you ever been diagnosed with Ulcerative
Colitis?’ and “Have you ever been diagnosed with Crohn’s Disease?”. The allowed

responses are: Yes, No, Don’t Know. In this analysis a yes to either is a positive for

24

inflammatory bowel disease. No to both is negative. Among cases 1/ 188 (<1%)

reported positive and 1/ 186 (<1%) controls reported positive. Table 22

7.1.7. Meat consumption

Table 6 lists the meat and vegetable items on the MECC food frequency
instrument. For this analysis, the sum of weekly servings was calculated for both meat
items and vegetable items. For each item the coded frequency response was converted to
a weekly serving amount following the logic in Table 7. The mean weekly sum of meat
consumption was 10.6 (+/-7.8)weekly servings among cases and 10.7 (+\- 7.1) weekly
servings among controls. (Table 23). The inner quartile range of meat consumption for
cases was 6 to 13.5. Among controls the average reported value was 10.7 (+/- 7.2).
Among both cases and controls the mean reported meat servings per week was 10.6 (+/-
7.5). Any reported value above 32.3 was considered an outlier. Three cases reporting 35,
37 and 42.5 servings per week were removed. Two controls reporting 38.5 and 55

servings per week were removed as outliers.

7.1.8. NSAIDS

As with aspirin use, the MECC questionnaire asks subjects “Have you taken non-
steroidal anti-inflammatory drugs regularly within the past year?”; “Have you ever taken
non-steroidal anti-inflammatory drugs at least once a week during your life?”. Yes to
either is coded as positive for NSAIDS. No to both is coded as negative for NSAIDS. In

this analysis, 57/219 (26%) of cases and 40/219 (18%) of controls were missing NSAIDS

25

responses. Among cases 16/162(9%) reported positive for NSAID use and among

controls, 14/ 179 (8%) reported positive for NSAID use. (Table 24)

7.1.9. Relatives with CRC

The MECC instrument asks users about their relatives. It queries subjects on
individuals in four generations of their pedigree. Subjects are asked to report any cancer
in any relative as they answer general questions about the relative. Table 25 outlines the
relatives about whom cancer history information is queried from subjects. In this
analysis, if a subject reports cancer of the colon or rectum in a Mother, Father, Son,
Daughter, Sister or Brother then the subject was determined to have a positive family
history of CRC among first-degree relatives. Family history data was available
for19l/219 (87%) of cases and 190/219 (87%) or controls. In this analysis 23/191
(12.4%) cases and 14/ 190 (7.4%) of controls had a positive family history of CRC. (Table
26). No APC11307K carrier reported a positive family history of CRC among first-degree

relatives. (Table 27).

7.1 . 10. Sports Participation

The MECC questionnaire queries subjects as to whether they participate in a sport
activity. For this analysis, an answer of YES to this question is used as a proxy of
physical activity beyond activities of daily life. In this analysis 91/ 187 (49%) of controls

reported sports participation and 80/187 (43%) of cases. Table 28.

26

7.1. 1 1. Vegetable Consumption

Vegetable items were calculated as outlined above in the meat section. The mean
of weekly vegetable servings was 44 (+l-26.6) among cases and 51 (+/- 25.9) among
controls. Among cases and controls the mean reported value was 48.0 (+/- 26.5). Among
the controls the mean value reported was 51.4 (+/- 25.9). Any value above 129.1 was
removed as an outlier. Three cases who reporting vegetable servings per week of 142.5,
151 and 162 were removed as outliers. Two controls reporting 143.5 and 148.5 servings

per week were also removed as outliers. (Table 29)

7.2. Unadjusted Analyses

7.2.1. Aspirin

In the matched analysis, taking aspirin once a week ever in ones lifetime was
significantly associated with an elevated risk of CRC OR=1.696 (95%CI=1.013-2.839).
(Table 30). This significant odds ratio shows aspirin as a risk factor, which is not

consistent with the literature, which generally reports this exposure as a protective effect.

7.2.2. Body Mass Index

In this analysis, body mass was bifurcated at the median value of controls. The
relative risk for CRC and BMI was 1.750 (1.100-2.784). This significant relative risk

estimate is stronger that appears in most literature as mentioned above. (Table 31)

27

7.2.3. Vegetable Consumption

In this analysis, a weekly vegetable serving score for all subjects was bifurcated at
the median value reported among controls. The relative risk estimates of CRC and
vegetable consumption were 0.460 (0281-0754). This significant protective effect of

vegetables is consistent with the literature as mentioned above. (Table 32)

7.2.4. Other variables of interest.

No other variable measured in this analysis demonstrated a statistically significant
relative risk estimate in a matched bi-variate analysis. Education, which was significant

in the unmatched analysis, retained no significance in matched analysis. (Table 33)

28

, 7.3. Adjusted Analyses

Aspirin use, body mass index and vegetable consumption demonstrated
statistically significant unadjusted odds ratios. Using a forward selection strategy, an
initial model was evaluated using these three variables and APCIl307K and all possible
interaction terms. The interaction terms were not statistically significant. A second
model was built using APC11307K, Aspirin, body mass index and vegetables, in which
body mass index was not statistically significant. A third model with APC11307K,
Aspirin and Vegetables was built, in which aspirin was not significant. The final model
was one of APCIl307K and vegetable consumption on a continuous scale. The matched
odds ratio estimate of APC11307K was 1.875 (0.795-4.422). (Table 34) In the final
model the odds ratio estimate of APCIl307K was 1.320. At first glance, using the rule of
a +/- 20% change to identify possible confounding, it would appear that the association
between APC11307K and CRC is confounded by vegetable consumption. To examine
this potential confounding further a model was built of APC11307K among the cases and
controls where vegetable consumption was available for both the case and the control. In
this interim, model the APC11307K odds ratio estimate changed to 1.199. In the final
model the ACPIl307K OR term was 1.320. The odds ratio difference for ACPIl307K
between these two models is 10% and does not suggest substantial confounding. (Table

34 and Table 35).

29

8. CONCLUSION

The first null hypothesis of this study was that there was no association between
ACPIl307K and CRC. In these data this null hypothesis cannot be rejected as the
confidence interval of 0.795 to 4.422 around the odds ratio point estimate of 1.875
includes unity. The second null hypothesis of this study is that the association between
APC11307K and CRC is independent of other known risk factors for CRC. Since the
first null hypothesis cannot be rejected, the second null hypothesis can only be examined
and not thoroughly tested. In these data the non-significant point estimate odds ratios
indicate that the association could be independent of other factors however this null
hypothesis cannot be rejected.

The final sample size of the MECC study will be 2100 matched pairs. This
analysis on the first 219 or 10% of these pairs is likely to be quite different from the
analyses to be performed on the final complete dataset. This analysis is underpowered to
find the effects in the direction and magnitude of the final sample size. As stated earlier,
MECC is a population-based 1:1 matched case-control study. It is beyond the scope of
this project to estimate power of the given sample size to find effects anticipated from the

literature review, given this complex study design.

30

9. DISCUSSION AND LIMITATIONS OF THIS STUDY

This study examined measures of association between many known risk factors
for CRC. In this analysis, only vegetable consumption and body mass index were found
to have the anticipated magnitude and direction of odds ratio.

Ethnicity did not show the anticipated risk factor association, however the study
design effectively matches for ethnicity as thus predeterrnines the inability to measure an
effect. Inflammatory bowel disease, meat consumption, relatives with cancer and sports
participation did not show the anticipated effects however the study is quite definitely
underpowered to measure these effects.

The matching scheme for the MECC study includes a clinic code that might serve
as a geographic indicator. Also stated previously within given areas of Israel individuals
of a given ethnic group generally live close to others of their ethnic group. It is infrequent
to find heterogeneity within a given geographic area or neighborhood. This with absence
of ethnic heterogeneity in a geographic area, matching on geographic area probably
functionally matches on ethnicity. In the literature, being Ashkenazi is associated with
increased risk of CRC. The absence of an association in this study is due in part to lack
of power, and in part to the functional match on ethnicity. The case population for this
project, as stated earlier will include 80% of the CRC cases diagnosed in the region of
Northern Israel. The control population for this project is limited to the 65% of the
population who are members of the participating health maintenance organization. If the

20% of cases missed or the 35% of controls ineligible for the study have a different ethnic

31

frequency distribution, then the study design might also preclude finding an effect of
ethnicity based on eligibility criteria.

The population carrier rate in this study was 7.9% (Table 14). Among the
Ashkenazim in this study the carrier rate was 8.5%. Other studies among Ashkenazim in
the United States (Petrukhin, 1997), 6.7% among Ashkenazim and Sephardim in Israel,
(Drucker, 2000), 7.0% among Ashkenazim and non-Jews the United States (Laken,
1997), 7.2% among Ashkenazim in the United States (Woodage, 1998) and 10% among
Ashkenazim in Toronto Canada (Gryfe, 1999). The carrier rate results of this analysis
are consistent with other published studies.

Among the Ashkenazim the polymorphism was found in 13/ 127 (10.2%) of
Ashkenazim cases and 7/ 126 (5.6%) of Ashkenazim controls. (Table 15) The studies,
which publish ethnicity and APCIl307K results, show 4.5% (Petrukhin, 1997), 10%
(Gryfe,1999;Laken,l997), 13% (W oodage,1998) and 13.5 (Drucker,2000) in Ashkenazim
cases and 5.7% (Drucker,2000), 6%(Laken,1997), 7%(Woodage,1998) among
Ashkenazim controls (Table 5). The results of this analysis are clearly consistent with
these published estimates.

The lack of positive family history among carriers is not consistent with the work
of Laken et a1. They found a 6—fold elevated risk among carriers with a family history.
Their conclusion was based on a set of 25 individuals with family history data (Laken,
1997). However, this thesis analysis is underpowered to conclude the lack of

association.

32

It is unclear as to why the early analysis of data in this project, estimates aspirin
exposure as a risk factor and not a projective effect. One possible explanation might be
that aspirin consumption is associated with gastrointestinal side effects. It is feasible that
the side effects cause cases to seek medical attention, where a CRC is found; but yet the
aspirin has little to do with the genesis of the CRC. Another possible explanation is that
individuals experiencing early symptoms of CRC often self-medicate with aspirin-like
medications. But if the self-medication does not alleviate the symptoms they seek
medical attention. A third possible explanation is that the questions regarding aspirin are

somehow not measuring the anticipated exposure patterns.

The validity MECC food frequency instrument has yet to be tested. However, to
make the results of this thesis and the MECC project more scientifically rigorous the

instruments validity and reliability should be tested and the results published.

It has long been supposed that specific types of vegetables, namely broccoli,
confer greater protective effect than vegetables in general. This analysis did not test for
an effect of broccoli separate from all vegetables. One could surmise that the relationship
between APCIl307K and vegetables might change with broccoli tested separately from
other vegetables. This lack of vegetable-specific analysis is a source of unmeasured
confounding in this analysis.

In cancer epidemiology, histopathologic confirmation of cancer cases is of
paramount importance, generally when cancer cases are collected from a many

institutions, a pathologist will review slides from the many institutions, and confirm or

33

refute the presence of the diagnosed cancer in the case. The MECC study followed this
standard of cancer epidemiology, however this analysis did not incorporate the pathology
data when including cases for analysis. The resultant set of subjects might include some
misclassified cases. The net result of this rnisclassification is difficult to anticipate and

further analysis will fully incorporate the pathology data being collected.

10. Further Study Needs

Future studies of APC11307K and CRC should include both a measurement of
vegetable consumption and history of polyps and pathology data in a larger dataset in

order to fully evaluate the genetic and environmental contributions to CRC.

34

11. REFERENCES

Aaltonen LA, Salovaara R, Kristo P. Canzian F, Hemrninki A, Peltomaki P. et.al.
Incidence of hereditary nonpolyposis colorectal cancer and the feasibility of
molecular screen for the disease. N Engl J Med 1998;338:1481-7.

Benito E, Cabeza E, Moreno V, Obrador A, Bosh FX. Diet and colorectal adenomas: a
case-control study in Majorca - I dietary factors. Int] J. Cancer 1990;55:166-213.

Bingham SA, Nelson M. Assessment of food composition and nutrient intake in :BM
Margetts , M. Nelson (eds) Design and concepts in Nutritional Epidemiology
(pp153-191) Oxford: Oxford Medical Publications.

Bingham SA, Williams DDR, Cole TJ, James WPT Dietary fibre and retional large
bowel cancer mortality in Britian. Br J Cancer 1979:40;456-63.

Bodmer WF, Bailery CH, Bodmer J, Bussey HJR, Ellis A, Gorman P et.al. Localization
of the gene for familial adenomatous polyposis on chromosome 5. Nature 1987;
328:614-16

Bodmer W. Familial adenomatous polyposis (FAP) and its gene, APC. Cytogenet Cell
Genet 1999 ; 86:99-104.

Cancer Incidence on 5 Continents 1992, Volume VI

Cannon-Albright LA, Skolnick MH, Bishop DT. Le RG, Burt RW Common inheritance
of susceptibility to colonic adenomatous polyps and associated colorectal cancers.
N Engl J Med 1988;391 : 533-537

Central Burreau of Statistics. Statistical Abstract of Israel 1996.

Cunningham C, Dunlop MG. Molecular genetic basis of colorectal cancer susceptibility
[Reviews] The British Journal of Surgery 1996; 83(3):321-329.

Drucker L, Shpillberg O, Neumann A, Shapira J, Stackievcz R, Beyth Y, Yarkoni S.
Adenomatous Polyposis Coli Il307K Mutation in Jewish patients with different
ethnicities: prevalence and phenotype. Cancer 2000: 88#4; 755-760

Ekbom A, Helmick CG, Zack M, Holmberg L, and Adarni HO Gastroenterology
1992;103:954-960.

Evans DG, Walsh S, Jeacock J, Robinson C, Hadfield L. Davies DR et.al. Incidence of
hereditary non-polyposis colorectal cancer. Br J Surg 1997; 84: 1281-5.

Fodde R, Srnits R, Hofland N, Kielman M, Khan PM. Mechanisms of APC-driven
tumorigenesis: lessons from mouse models. Cytogenet Cell Genet 1991 86:105-
1 1 1.

Food, Nutrition and the prevention of cancer: a global perspective. World Cancer
Research Fund in association with the American Institute for Cancer Research.
1997 John D Potter (Editor) Washington DC.

Foulkes WD. A tale of four syndromes: familial adenomatous polyposis, Gardner
Syndrome and attenuated APC and turcot syndrome. QJM 1995 88: 12 p853-863.

Gann PH, Manson JE, Glynn RJ, Burning JE, Hennekens CH. Low-dose aspirin and
incidence of colorectal tumors in a randomized trial. J Natl Cancer Inst

35

 

1999;85:1220-4.

Giardello FM, Peterson GM, Piantadosi S, Gruber SB, Traboulsi EI, Offerhaus GJ et.al.
APC gene mutations and extraintestinal phenotype of familial Adenomatous
Polyposis. Gut 1997; 40:521-525

Giovanucci E, Ascherio A, Rimm EB, Colditz GA, Stamper MJ, Willett WC. Physical
Activity, Obesity and Risk for Colon cancer and adenoma in men. Annals of
internal medicine 1995:122;327-334.

Giovanucci E. Stampfer MG, Colditz GA, Rimm EB, Willett WC. Relationship of diet
to colorectal adenoma in men. JNCI 1992;84;91-98.

Gryfe R, N ando DN, Geeta L, Gallinger S, Redston M. Inherited colorectal polyposis
and cancer risk of the APCIl307K polymorphism Am J Hum Genet 1999
64:37 8-384.

Gryfe R, Nicola ND, Lal G, Gallinger S, Redston M. Inherited colorectal polyposis and
the risk of the APCIl307 polymorphism. Am J Hum Genet 1999: 64;378-384.

Harpaz N, Talbot IC Semin Diagn. Pathol. 1996 :13;339-357.

Hoff G, Moen IE, Mowinckel P, Rosef O. Nordbo E, Sauar J, Vatn MG, Torgrimsen T.
Drinking water and the prevalence of colorectal adenomas: an epidemiologic
study in Telemark Norway. Eur J Cancer Prev 1992;423-428.

Kato I, Torninaga 8, Ikari A. A case-control study of male colorectal cancer in Aichi
Prefecture, Japan: with special reference to occupational activity level, drinking
habits and family history. Jpn J Cancer Res 1990 : 81;115-121.

Kawaharada M. Satoh Y. Nippon Ronen Igakkai Zasshi - Japanese Journal of
Geriatrics. 35(12):891-7,1998 Dec.

King H, et a] 1985, Patterns of site-specific displacement in cancer mortality among
migrants, American Journal of Public Health 1985 75:237-242.

Kinzler KW, Vogelstein B, Gatekeepers and caretakers, Nature 1997; 386:762-3

Kinzler KW, Vogelstein B. Landscaping the Cancer Terrain. Science 1998: 280;1036-
1037.

Kinzler KW, Vogelstein B: Lessons from hereditary colorectal cancer. Cell 1987:159-
170.

Kolodner RD Mismatch repair: mechanisms and relationships to cancer susceptibility.
Trends Biochem Sci 1995;20:397-401.

Kune GA, Kune S, Read A, MacGowan K, Penfold C, Watson LF. Colorectal polyps,
diet, alcohol and family history of colorectal cancer: a case-control study. Nutr
Cancer 1991;16:25-30.

Laken SG, Peterson GM, Gruber SB, Oddous C, Ostrer H, Giardiello FM, Hamilton
SR, Hampel H, Markowitz A, Klimstra D, Jhanwar S et a1. Familial colorectal
cancer in Ashkenazim due to a hyperrnutable trace in APC. Nature Genetics
1997; 17:79-83.

Landis SH, Murray T, Bolden S, Wingo PA. Cancer Statistics 1998. Ca: A Cancer
Journal for Clinicians 1998; 4826-29.

Leppert M, Dobbs M, Scambler P. O‘Connell P, Nakamura Y, Stauffer D. et.al. The

36

gene for familial polyposis coli maps to the long arm of chromosome 5. Science
1987; 238: 1411-13

Lothe RA, Hektoen M, J ohnsen H, Meling GI, lkdahl TI, Rognumm TO, Linblom,
Borresen-Dale AL. The APC gene Il307K variant is rare in Norwegian patients
with familial and sporadic colorectal or breast cancer. Cancer Research 1998.
58(14):2923-4.

Lu S-L, Kawabata T, Irnarnura T, Yoshirnitsu A, Nomizu T, Miyazono K, et.al.
HNPCC associated with germline mutation in the TGF-Beta type II receptor gene.
Nature Genetics 1998: 19:17-18

Lynch HT, de la Chapelle A. Genetic susceptibility to non-polyposis colorectal cancer.
Genetic susceptibility to non-polyposis colorectal cancer [Review] Journal of
Medical Genetics. 36(Potter, 1999):801-19,1999 Nov.

Lynch HT, Lynch JF, Cristofaro G. Genetic epidemiology of colon cancer. In Lynch HT
Hirayama T. editors Genetic Epidemiology of cancer. Boca Raton (FL) CRC
Press; 1989 p. 251-77.

Lynch PM, Lynch HT, Colon cancer genetics, New York, NY: Van Nostrand Reinhold;
1985.

MacQuart-Moulin G, Riboli E, ComJe J et a1. Case-control study in Marseilles. Int J
Cancer 1987; 40:179-188.

McKeownEyssen G. Epidmeiology of colorectal cancer revisited: are serum
triglycerides and/or plasma glucose associated with risk? Cancer Epidemiol
Biomark Prev 1994: 3; 687-695.

McMichael AJ, McCall MG, Hartshome JM eta]. 1980 Patterns of gastrointestinal
change in European mi grants to Australia: The role of dietary change. Int J
Cancer (Suppl) 25:431-437.

Morbidity and mortality Weekly Report (MMWR), 1992.

Nugent FW, Haggitt RC, Gilpin PA Gastroenterology 1991;100:1241-1248.

Paganini-Hill A, Chao A, Ross RK, Henderson BE. Aspirin use and chronic diseases: a
cohort study of the elderly. BMJ 1989;299:1247-50.

Petrukhin L, Dangel J, Vanderveer L, Costalas J, Bellacosa A, Grana G, Daly M,
Godwin AK. The Il307K APC mutation does not predispose to colorectal cancer
in jewish Ashkenazi breast and breast-ovarian kindred. Cancer Research
1997:57;5480-5484.

Phillips RL, Snowdon DA. Dietary habits and risk of cancer among Seventh-day
Adventists. JNCI 1985;74:307-317.

Potter JD. Colorectal Cancer: Molecules and Populations. JNCI 1999 91:11 pp 919-932

Rose DB, Boyar AP, Wynder EL. International comparisons of mortality rates for
cancer of the breast, ovary, prostate and colon and per capita food consumption
Cancer 1986:58;2363-71.

Rubinfeld B, Souza B, Albert 1, Muller O, Chamberlain HS, Masiarz FR, Mumenmitsu
S, Polaskis P. Association of APC gene prouct with beta-catenin. Science 1987;
262: 1734-1737.

37

Sandler RS, Lyles CM, Peipins LA, McAuliffe CA, Woosley JT, Kupper LL. Diet and
risk of colorectal adenomas: macronutrients, cholesterol and fiber JNCI
1993;85:884-891.

Schottenfeld D, Winawer S. Cancers of the Large Intestine. In Cancer Epidemiology
and Prevention.1996 Schottenfeld and Fraumeni (eds) Oxford University Press,
Oxford.

St John DJB, McDermott FT, Hopper JL, et.a1. Cancer risk in relatives of patients with
common colorectal cancer. Ann Intern Med 1993 118:785-790.

Strewing JP, Hartage P, Wacholder S, Baker SM, Berlin M, McAdarns M, et.al. The
risk of cancer associated with specific mutations of BRCA1 and BRCA2 among
Ashkenazi Jews [see comments]. New Engl J Med. 1997; 336:1401-1408.

Stunner T. Glynn R], Lee IM, Manson JE, Burning JE, Hennekins CH, Aspirin use and
colorectal cancer: post-trial followup data from the physicians health study.
Annals of Internal Medicine 1998;128:713-20.

Su L-K, Vogelstein B, Kinzler KW. Association of the APC tumor suppressor protein
with catenins. Science 1993 ; 262:1734-7

Vassen HF, Mecklin JP, Khan PM, Lynch HT. The International Collaborative Group
on Hereditary Non-Polyposis Colorectal Cancer (ICG-I-INPCC). Dis Colon
Rectum 1991;34:424-5.

Vassen HF, Mecklin JP, Khan PM, Lynch HT. The International Collaborative Group
on HNPCC. Anticancer Res 1994; 14:1661-1664.

Warshauer ME, Sliverman DT, Shottenfeld D. et.al. 1986. Stomach and colorectal
cancer incidence and mortality in Puerto Rican-bom residents in New York City.
J Natl Cancer Institute 1986 76:591-595

Willet WC, Nutritional Epidemiology. Oxford University Press, New York 1990.

Willett, Stampfer MG, Colditz GA et a1. Relation of meat fat and fiber intake to the risk
of colon cancer in a prospective study among women. N Engl J Med
1990;323:1664-1672.

Woodage T, King SM, Wacholder S, Hartage P, Streuwing JP, McAdarns M, Laken SJ,
Tucker MA, Brody LC. The APC11307K allele and cancer risk in a community-
based study of Ashkenazi Jews. Nature Genetics 1998: 20;62-65.

Wynder EL, Fujita Y, Harris ER et.al. 1991 Comparative epidemiology of cancer
between the United States and Japan. Cancer 1991 67:746-763.

Yin J, Harpaz N, Souza RF, Zou T, Kong D, Wang S, Leytin A, Medalie NS, Smolinski
KN, Abraham JM, Fleisher AS, Meltzer SJ. Low prevalence of APC1107K
sequence in Jewish and non-Jewish patients with inflammatory bowel disease.
Oncogene 1999 18:3902-3904.

Zeigler RC, Devesa SS, Fraumeni JF Jr. Epidemiologic patterns of colorectal cancer. In
DeVita VT, Jr. Hellman S. Rosenberg SD (eds): Important Advance in Oncology.
Philadelphia, J .B. Lippincott, pp. 209-232.

38

Table 1 Geographic difference in Age-Adjusted colonic cancer incidence rates

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Per 100,000 Region/Population Per 100,000
Males Females
34.] Hawaii, Japanese 22.0
34.1 Connecticut, Whites 26.1
33.] California, Alameda, Blacks 28.5
32.0 Detroit, Blacks 29.0
24.9 Australia, Queensland 25.6
24.] Canada 22.5
22.6 Hawaii, Chinese 20.4
20.8 Italy, Varese 16.5
19.4 Norway, Urban 18.9
18.9 Denmark 18.9
16.8 Sweden 15.8
16.6 UK England and Wales 14.7
15.9 Norway, Rural 16.5
15.8 Hong Kong 12.5
14.7 France, Calvados (urban) 10.5
13.0 Israel, Jews born in Israel 14.2
10.3 Puerto Rico 9.7
9.6 Japan, Miygai 9.4
8.5 Shanghai 7.6
7.8 Hungary, Szaboica 7.1
5.3 Costa Rica 5 .3
5.2 Columbia, Cali 6.3
4.7 Israel, Non-Jews 4.5
3.4 India, Bombay 2.9

Adjusted to 1970 world population.

Source: Cancer Incidence on Five Continents,

Vol VI, 1992

 

39

 

Table 2 Ethnic differences in CRC in the US. Per 100,000

 

 

 

 

 

 

 

 

 

 

 

Race Males Females
Total 27.7 19.8
Whites 27.9 19.4
Blacks 30.4 23.3
Hispanics 13.6 8.9
Puerto Ricans 18.9 13.8
Mexicans 8.3 6.7
Native Americans/Alaskan Americans 12.5 11.5

 

Source Morbidity and Mortality Weekly Report, 1992

40

 

Table 3 Ethnic differences in Age-adjusted CRC Incidence in Israel. Per 100,000

 

 

 

 

 

 

 

Males Females Overall
Population Colon Rectum Colon Rectum Colon
All Jews 19.7 16.3 16.9 13.6 18.3
Jews born in Europe or 22.5 19.0 19.7 16.1 21.]
America
Jews born in Africa or 13.2 11.0 11.1 9.2 12.2
Asia
Jews born in Israel 18.] 12.8 20.5 11.0 19.3
Non-Jews 4.6 3.6 4.6 2.9 3.4

 

 

 

 

 

 

 

 

Source: Cancer Incidence in Five Continents, Vol V], 1992

41

Table 4 Amsterdam Criteria

 

 

1. Three cases of familial CRC, in which two of the affected individuals are first-degree
relatives of the third.

2. Colorectal cancer occurring across two generations

3. One colorectal cancer occurring before 50.

No evidence of FAP

 

42

 

Table 5 Studies of CRC and 11307K

 

 

 

 

 

 

 

Reference Ethnicity APC11307 APC11307K Odds Comment
K Positive Positive rate Ratio
rate in in controls (95% Cl)
cases (%) (%)
Laken 1997 Ashkenazim 22/211 47/766 (6%) 1.78 Initial
Carriers = 69 (10%) (1.08- Publication
N: 977 2.96)
Prevalence =
7%
Population 0/243
Woodage, Ashkenazim 7/55 (13%) 355/4948 1.887
1998 Carriers = (7%) (084-42)
362
N = 5003
+Rate =
7.2%
Petrukhin, Ashkenazim, 12/264 - n0 PubliShed Reported
1998 Breast/Ovaria (4.5%) controls no elevated
n Kindred “8'5““:
patients,
Reported
0/2 positive
CRC
Carriers = 12 patients.
N : 264 Would have
Prevalence=4 “9°th
. 5% fewer than
1 CRC case
in person
years
reported.
Gryfe, 1999 48/476 - no published
(10%) controls
Drucker, Jewish, Among Among Among - Very high
2000 Ashkenazi Ashkenazim Ashkenazim Ashkenazim. positive rate
and Sephardi. 5/1 1 1 17/298 among
Hospital (13.5%) (5.7%) 3.4 “5“-
Commls Among Among (1.2-7.2) ' “‘3‘.”
Sepgargm 881319171121; Population 2:32: rate
' = 2‘7 se hardim.
$22338 3“ (75%) (4.8%) (2.6-2.9) ,2me
Prevalence = mm
6.7% Yemen

 

 

 

 

 

 

43

 

 

Table 6 - FFQ Items on the MECC Questionnaire.

 

 

 

 

Categiry Item
Vegetables Artichoke (1 medium)
Beet leaves cooked (100 g)

 

Beet root cooked red beet (1)

 

Cooked Broccoli (1/2 cup)

 

Cooked Cabbage (1/2 cup)

 

Cooked Carrots (1/2 cup)

 

Cooked Okra (1/2 cup)

 

Cooked asparagus (1/2 cup)

 

Cooked cauliflower (1/2 cup)

 

Cooked celery (1/2)

 

Cooked pumpkin (1/2 cup)

 

Cooked/baked sweet potato (1/2 )

 

Coriander

 

Corn 1 ear or 1/2 cup canned

 

Cucumber (1)

 

Egg plant (1/2 cup)

 

Fresh Broccoli (1/2 cup)

 

Fresh Cabbage (1/2 cup)

 

Fresh Celery (33 g/stalk)

 

Fresh carrot (1)

 

Fresh cauliflower (1/2 cup)

 

Garlic (fresh or powdered) l CLOVE

 

Green Beans (1/2 cup)

 

Lettuce (30 grams)

 

Mushrooms (1/2 cup)

 

Onion (1/4)

 

Other vegetable

 

Parsley (10g 1 stalk)

 

Peas (1 cup)

 

ngper (1)

 

Pickled cucumbers/eggflant/peppers

 

Radish (1)

 

Spinach (1/2 cup cooked)

 

Sprouts (1/2 cup)

 

Tomato (1)

 

 

Zuchini (1/2 cup)

 

44

 

Table 6 cont’d

 

Meats and Eggs

Beef-steak,roast (100-150g)

 

 

Canned Tuna

 

Chicken/Turkey w/Skin (100-150g)

 

Cooked Fish (100-150g)

 

Deli Meat (Sliced Salami)

 

Egg (1)

 

Hamburg/meatballs/Kabob ( 100- 1 50g)

 

Hot Dogs (3)

 

Internal Organs

 

Lamb

 

Liver (50-100 g)

 

Mackerel, Salmon, Sardines (canned)

 

Other meat or fish

 

Pork

 

Schnitzel (1)

 

Shrimp or Crab

 

Skinless Chicken/Turkey (100-150g)

 

Smoked Turkey/Sliced Pastrarni

 

 

Soy shnitzel

 

45

 

Table 7 — Logic to harmonize FFQ frequency codes into weekly servings

 

 

 

 

 

 

 

 

 

 

 

 

FFQ Reported Weekly Harmonized
Frequency Frequency
0 per month 0 / week
1-3 per month 0.5 / week
1 per week 1 / week
2-4 per week 3 / week

1 per day 7 / week
2-3 per day 17.5 / week
4-5 per day 31.5 / week
4+ per day 28 / week
6+ per day 42 / week

 

46

 

Table 8 — Gender among pairs

 

 

 

 

 

Male Pairs 112 (51%)
Female Pairs 95 (42%)
Missing 12 (7%)

 

 

47

Table 9 Age Univariate statistics of matched pairs

 

 

 

 

 

 

 

 

 

Missing 1

Min 22

25" Percentile 67

Median 73

75th Percentile 78

Max 98

Mean (SD) 71.8 (10.5)

 

48

 

Table 10 Education between groups

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Education Cases (%) Controls (%) p-value
0.0346

Never Studied 5 ( 2.2) 6 ( 2.7)

Primary Grade 1-6 29 (13.2) 29 (17.4)

Partial High School 46 (21) 38 (17.4)

Grades 7-1]

Completed High 52 (23.7) 38 (17.4)

School

Post Secondary 19 (8.6) 12 (5.5)

Vocational

Post Secondary 34 (15.5) 58 (26.5)

Academic

Missing 24 (10.9) 29 (13.2)

 

49

 

Table 11 Religion between groups

 

 

 

 

 

 

 

 

 

 

Religion Cases (%) Controls (%)
Jew 187 (85.4) 185 (84.5)
Christian Arab 3 (1.4) 3 (1.4)
Moslem Arab 0 1 (<1)
Christian - Non Arab 2 (<1) 0

Missing 27 (12.3) 30 (13.7)

 

 

50

Table 12 Ethnicity between groups

 

 

 

 

 

 

 

 

 

Ethnicity Cases (%) Controls (%)
Ashkenazi 153 (83.61) 143 (76.88)
Sephardi 29 (15.85) 42 (22.58)
Mixed 1 ( 0.55) 1 ( 0.54)
Arab 1 ( 0.55) 0

 

51

 

Table 13 Frequency of Cases and controls missing APC11307K results

 

 

 

 

Missing APC11307K Results
Cases Controls
36 24

 

 

 

 

52

Table 14 Unmatched analyses of APCIl307K

 

 

 

 

 

 

 

 

Unmatched Analysis - All subjects

APC11307K Cases Controls
Carrier Status

Positive 18 10
Negative 166 l 85
Odds Ratio 2.006 (1.20 - 4.468)

 

53

 

Table 15 Matched analysis of APC11307K

 

 

 

 

 

 

 

 

Matched Analysis - All Matched Subjects
APCIl307k APCIl307k
Controls Controls
Positive Negative

APCIl307k Cases 0 15

Positive

APCIl307k Cases 8 141

Negative

Odds Ratio 1.875 (95% CI 0.795 - 4.422)

 

 

54

Table 16 Unmatched analysis of APC11307K among Ashkenazim

 

 

 

 

 

 

 

 

Unmatched Analysis - Among Ashenazim
APCIl307K Cases Controls
Carrier Status

Positive 13 7
Negative 114 119
Odds Ratio 1.938 (95% CI 0.9845 - 5.0331)

 

 

55

Table 17 Matched analysis of APCIl307K among Ashkenazim

 

Matched Analysis - Both Case and Control Askenazi

 

 

 

 

 

 

APCIl307k APCIl307k
Controls Controls
Positive Negative
APCIl307k Cases 0 0
Positive
APCIl307k Cases 2 26
Negative

 

Both case and control = Ashkenazi

56

 

Table 18 - Frequency of birth country among APCIl307K Carriers

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Country of Number in Number of Carrier Cases Controls

Birth Study (%) Carriers Rate (% “P0519" (% 0' Ma“
(%of cases) controls)
positives)

Lithuania 3 (0.7) 1 (5) 33.3% 1 (7) 0

Germany 17 (3.9) 5 (24) 29.4% 1 (7) 4 (57)

Romania 63 (14.4) 6 (29) 9.5% 5 (36) 1 (l4)

Moldova 13 (3.0) 1 (5) 7.7% 1 (7) 0

Poland 70 (16.0) 5 (24) 7.1% 4 (29) l (14)

Ukraine 39 (8.9) 2 (10) 5.1% 2 (14) 0

Israel 55 (12.6) 1 (5) 1.8% 0 1 (14)

Other 120 (27.4)

Missing 58 (13.2) 7

 

57

 

Table 19 Unmatched analyses of study variables - Aspirin

 

 

 

 

 

 

 

 

 

Aspirin Cases Controls Statistic
OR=1.56
(95%CI = 1 .016-2.382
Yes 77 62
No 99 124
Missing 43 33

 

 

58

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 20 Unmatched analyses of study variables Body Mass Index
Body Mass Index Cases Controls Statistic
Missing 50 42
Min 17 14.8
25lh Percentile 24.1 22.9
Median 25.9 25.1
75lh Percentile 28.4 28.]
Max 39.6 43.4
Mean (SD) 26.4 (3.7) 25.8 (4.3)
38.7 38.7
Number of Outliers 1 3
Binary at control OR = 1.5
Median (95%Cl = 0.98 2.31)
Above 25.1 103 90
Below 25.] 66 87

 

 

 

 

 

 

59

 

Table 21 Unmatched analyses of study variables - Ethnicity

 

 

 

 

 

Ethnicity Cases I Controls Statistic
OR = 1.3
(95%CI=0.87- l .96)
Ashkenazi 153 143
Non Ashkenazi 62 76
Missing 4 0

 

 

 

 

 

 

60

Table 22 Unmatched analyses of study variables Inflammatory Bowel Disease

 

 

 

 

 

 

 

 

 

Inflammatory Cases Controls Statistic
Bowel Disease
OR=O.98
(95%CI=0.06 15.9)
Yes 1 1
No 187 185
Missing 31 33

 

61

 

Table 23 Unmatched analyses of study variables Meat Consumption

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Meat Cases Controls Statistic
(Servingg per week)
Missing 44 29
Min 0 0
25‘" Percentile 6.0 6
Median 9.5 10
75" Percentile 13.5 13
Max 72.5 55
Mean (SD) 9.6 (7.9) 10.7 (7.2)
32.3 32.3
Number of Outliers 4 3
Binary at control OR=0.71
Median g (95%CI = 0.47-1.07)
Above 10 71 93
Below 10 104 97

 

62

 

Table 24 Unmatched analyses of study variables - NSAIDS

 

 

 

 

 

 

 

 

 

NSAIDS Cases Controls Statistic
OR=1.29
(95%CI = 0.61 2.73)
Yes 16 14
No 146 165
Missing 57 40

 

63

 

Table 25 - Relatives queried on the MECC questionnaire

 

* First Degree Relatives

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Relationship Number Number
reported reported per
per case ** control **

* Father 0.98 0.99

* Mother 0.98 0.99

* Sons 1.3 1.3

* Daughters 1.0 1.2

* Brothers 1.5 1.3

*Sisters 1.4 1.4

Paternal Grandfather 0.4 0.4

Paternal Grandmother 0.5 0.4

Maternal Grandfather 0.5 0.4

Maternal Grandmother 0.5 0.5

Paternal Uncles 1.1 1.0

Paternal Aunts 0.9 0.9

Maternal Uncles 1.1 1.0

Maternal Aunts 0.9 1.0

Cousins with cancer 0.2 0.3

 

** Numbers less than 1 indicate a subject’s lack of information about a given person.

64

 

Table 26 Unmatched analyses of study variables Relatives with cancer/Positive Family

 

 

 

 

 

History
Cases Controls Statistic
Relatives with OR = 1.7
Cancer (95%CI=O.9 - 3.4)
Yes 23 14
No 168 176
Missing 28 29

 

 

 

 

 

 

65

Table 27 Family history of CRC in first degree relatives among APCIl307K carriers.

 

 

 

 

 

 

 

First No First
Degree Degree
relative relative with
with CRC CRC
APCIl307K Positive 0 20
APCIl307K Negative 29 278

 

66

Table 28 Unmatched analyses of study variables Sports Participation

 

 

 

 

 

 

 

 

 

Cases Controls Statistic
Sports OR = 0.8
(95%CI=0.5 1.2)
Yes 80 91
N0 107 96
Missing 32 32

 

67

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 29 Unmatched analyses of study variables Vegetables
Cases Controls Statistic
Vegetables
(sum per week)
Missing 42 29
Min 1.5 0.5
25‘" Percentile 28.5 32.5
Median 40 47.5
75‘“ Percentile 55 64.5
Max 162 148.5
Mean (SD) 41 (26.6) 51.4 L259)
129.] 129.]
Number of Outliers 3 2
Binary at control OR=0.5
Median (95%CI=O.33-0.78)
Above 47.5 60 95
Below 47.5 117 95

 

68

 

Table 30 Matched analyses of study variables Aspirin

 

 

 

 

 

 

 

 

Aspirin Controls Controls Odds Ratio
Positive Negative (95%CI)
1.69
(1.013-2.84)
Cases Positive 69 39
Cases Negative 23 27

 

69

 

Table 31 Matched analyses of study variables - Body Mass Index

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Body Mass Index Controls Controls Odds Ratio
Positive Negative (95%CI)
Continuous 1.063 (0.995-1.136)
Positive/Negative 1.68 (1.05-2.68)
(Above and belowControl Median of
25.0 kg *cm'2 )
Cases Positive 42 47
Cases Negative 28 26
Second quartile of Controls 1.20 (0.36-3.9)
(22.8 kg"'cm'2 - 25.0 kg*cm’2)
Vs
First Quartile of Controls
(< 22.8 kg"'cm'2 )
Cases Positive 10 6
Cases Negative 5 5
Third quartile of Controls N/A
(25.0 kg*cm‘2 — 28.0 kg*cm'2)
Vs
First Quartile of Controls
(< 22.8 1rg"'cm’2 )
Cases Positive 0 0
Cases Negative 0 5
Fourth quartile of Controls 2.40 (1.15 - 5.02)
(>280 kg‘cm’z)
Vs
First Quartile of Controls
(< 25.0 ltg"‘cm’2 )
Cases Positive 44 24
Cases Negative 10 5

 

 

 

 

 

 

 

70

Table 32 Matched analyses of study variables — Vegetables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Vegetable Consumption Controls Controls Odds Ratio
Positive Negative (95 % CI)
Continuous 0.984 (0.974-0.994)
Positive/Negative 0.472 (0.293 - 0.759)
(Above and belowControl Median of
47.5 servingsfiveek )
Cases Positive 56 53
Cases Negative 25 27
Second quartile of Controls 1.5 (0.534 - 4.214)
(32.5 servings/week - 47.5 servings/week)
Vs
First Quartile of Controls
(<32.5 Servings /week)
Cases Positive 10 9
Cases Negative 6 9
Third quartile of Controls 0.4 (0.125 - 1.275)
(47.5 servings/week - 64.5 servings/week)
Vs
First Quartile of Controls
(<32.5 Servings lweek)
Cases Positive 4
Cases Neggtive 10 9
Fourth quartile of Controls 0.353 (0.139 - 0.895)
(>645 servings/week)
Vs
First Quartile of Controls
(<32.5 Servings /week)
Cases Positive 16 6
Cases Negative l7 9

 

 

 

 

 

 

71

Table 33 Matched analysis of other study variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ethnicity Controls Controls Odds Ratio
Ashkenazi vs. Not- Positive Negative (95%CI)
Ashkenazi
1.39 (0.89 — 2.18)
Cases Positive 107 46
Casesfigative 33 29
Inflammatory Bowel Controls Controls Odds Ratio
Disease Positive Negative (95%CI)
Ever vs. Never
1.0 (0.63 — 15.9)
Cases Positive 0 1
Cases Negative 1 169
Meat Servings Per Week Odds Ratio
(95 % CI)
Continuous 0.979 (0.939-1.020)
NSAIDS Controls Controls Odds Ratio
Positive Negative (95% CI)
1.56 (0.67-3.59)
Cases Positive 119 14
Cases Negative 9 1
Relatives with CRC Controls Controls Odds Ratio
Positive Negative (95% CI)
1.818 (0.871 - 3.79)
Cases Positive 1 20
Cases Negative 1] 142

 

 

 

 

72

 

 

 

 

 

Table 33 cont’d

 

Sports Participation Controls Controls Odds Ratio
Positive Negative (95 % CI)

 

0.805 (0509-1273)

 

 

 

 

 

 

Cases Positive 39 33
Cases Negative 4] 55

 

 

73

Table 34 Multivariate analyses of APC11307K and significant study variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Model Variables Parameter Hazard 95%CI
Estimate Ratio
1. APCIl307K alone 11307K 0.43 1.875 0.795
4.422
-2LogL=227.352
2. Vegetables alone VEG -0.01620 0.984 0.974
Servings/Week 0994
Continuous
-2LogL = 217.648
3. Vegetables Binary VEGBin <47.5 vs -0.75122 0.434 0.266-
>=47.5 0.708
-2LogL = 205.478
4. BMI Alone BMIQ4 0.87547 2.400 1.148
weight‘height" 5,019
<22.9 vs >28.0
-2LogL==l 11.186
5. Aspirin Alone ASP 0.52805 1.696 1.013-
2.839
-2LogL=220.029
4. Fully Saturated Il307K -15.65 0 0-.
-2LogL=89.24l VEGBin 0.408 1.504 0.]-
22.663
BMIQ4
ASP 3.800 44.719 1.49-
1338.32
VEGBin*BMIQ4 -2.657 0.1265 0.002-
2.1 19
VEGBin‘Il307K -33.781 0 0-.
BMIQ4*11307K 32.1170 1.78*10"’ 0-.
VEGBin*BMIQ4*ASP 24.34 3.75*lO"° 0-.
ASP] 1 307K -3.455 0.021 0-.
ASP*VEGBin -20.57 0.032 0-.
ASP‘BMIQ4 p=0. 021 -3.84 0.021 0.001-
0.774

 

 

 

 

74

 

Table 34 cont (1

 

 

 

 

 

Model Variables Parameter Hazard 95%CI
Estimate Ratio

5. Higher Order Terms 11307K -0.19384 1.214 0.192-
Removed 7.657

-2LogL= 72.288 Veg continuous -0.027 0.973 0.949-
' 0.997

BMIQ4 0.5817 1.789 0.685-
4.668

ASP 1.154 3.171 0.995-

10.104

 

 

 

 

 

 

 

 

 

 

 

 

75

 

Table 35 — Final Adjusted Analysis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Model Variable Matched Parameter Hazard 95%
Pairs Estimate Ratio Confidence
Interval
1. APC11307K 164
alone
-2LogL=227.352 Il307K 0.43 1.875 0.795 — 4.422
2. Vegetables alone 148
-2LogL = 217.648 VEG -0.01620 0.984 0.974 — 0.994
Servings/Week
Continuous
3. Il307K in
subjects with
vegetable scores
11 307K 0.605 1.199 0366-3930
6. Final Model 139
-2LogL=155.265 Il307K 0.27797 1.320 0.3804587
Veg- -0.02232 0.978 0.965-O.991
confinuous

 

 

 

 

 

 

76

 

Frequency

Figure l Mutation Frequency in APC gene. From Bodmer et. Al

APC Mutation Frequency

 

3° C "'.='

   

 

 

 

 

 

 

 

 

 

ll.. n .

 

Q Q 0 Q 0 Q Q Q Q Q

 

IE FAP I Sporadic Codon Number

 

 

77

Figure 2 - Map of Israel.

 

 

 

 

78

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

It](It))jljljtljjjtlll