Le Marchand L, Seifried A, Lum-Jones A, Donlon T, Wilkens LR. Association of the Cyclin D1 A870G Polymorphism With Advanced Colorectal Cancer. JAMA. 2003;290(21):2843-2848. doi:10.1001/jama.290.21.2843
Author Affiliations: Cancer Research Center of Hawaii, University of Hawaii, Honolulu.
Context Cyclin D1 (CCND1) is a key cell cycle regulatory protein, the overexpression
of which is often found in human tumors and is associated with cell proliferation
and poor prognosis. A common adenine-to-guanine substitution polymorphism
(A870G) in the CCND1 gene results in an altered messenger
RNA transcript and a longer-life protein, which are preferentially encoded
by the A allele.
Objective To test the overall and stage-specific associations of the CCND1 870A allele with colorectal cancer.
Design, Setting, and Participants A population-based case-control study conducted in the multiethnic population
of Hawaii between January 1, 1994, and August 31, 1998, which included 504
patients with incident colorectal cancer and 624 population-based participants
of Japanese, white, or Native Hawaiian origin. Participation rates were 58%
for cases and 52% for controls.
Main Outcome Measurement Ethnicity, gene-dosage effects, and stage (regional/distant) and subsite
(colon vs rectal) of cancer.
Results The odds ratio (OR) for the CCND1 870 GA and
AA genotypes compared with the GG genotype was 1.2 (95% confidence interval
[CI], 0.9-1.7) and 1.5 (95% CI, 1.0-2.1), respectively (P = .03 for gene-dosage effect). These risk estimates were significantly
greater for patients diagnosed at a regional or distant stage (GA vs GG: OR,
1.7; 95% CI, 1.1-2.5 and AA vs GG: OR, 1.9; 95% CI, 1.2-3.1; P = .008 for gene-dosage effect) compared with those estimates for
patients diagnosed at an earlier stage (P = .048).
In subset analyses, the association between the A allele and advanced colorectal
cancer was statistically significant in white and Hawaiian participants but
not in Japanese, and was stronger for rectal cancer.
Conclusion The CCND1 870A allele may be associated with
colorectal cancer, and particularly with forms of the disease that result
in severe morbidity and mortality.
Cyclin D1 (CCND1) is a key regulatory protein of the cell cycle, promoting
the transition through the restriction point in the G1 phase beyond which
the cell is committed to divide.1,2 Overexpression
of the CCND1 gene, which has been shown to occur
in 30% to 50% of breast and colorectal cancers, has been associated with increased
cell proliferation and poor prognosis for a number of human malignancies,
including colorectal cancer.3- 5 A
common adenine-to-guanine (A/G) substitution at nucleotide 870 in the conserved
splice donor region of exon 4 has been shown to modulate splicing of CCND1 messenger RNA.6 The
G allele preferentially splices transcript a, whereas
the A allele mainly splices transcript b, which encodes
a protein with a longer half-life.6 The A allele
has been associated with poor prognosis for several cancers6- 8 and
with increased risk of colorectal cancer in hereditary nonpolyposis colorectal
cancer families,9 as well as in 2 small hospital-based
case-control studies.10,11 Because
the A allele is common and may preferentially affect progression, it is important
to further clarify its association with colorectal cancer, a neoplasm that
is particularly difficult to cure once it has spread outside the intestines.
We investigated the association of the CCND1 870A
allele with colorectal cancer in a population-based case-control study originally
conducted to test gene-diet interactions in the multiethnic population of
The participants and data collection methods for this study have been
described in detail elsewhere.12- 14 Cases
were identified through the rapid reporting system of the Hawaii Tumor Registry,
a member of the Surveillance Epidemiology and End Results program of the National
Cancer Institute. Eligible cases consisted of white, Japanese, or Native Hawaiian
residents of the island of Oahu diagnosed with a first adenocarcinoma of the
colon or rectum between January 1, 1994, and August 31, 1998. The control
group was selected from participants in an ongoing population-based health
survey conducted by the Hawaii State Department of Health, which follows a
design modeled after that of the National Health Survey.15 For
the group aged 65 years or older, this source was supplemented with participants
from the Health Care Financing Administration. A control pool was created
for each case that agreed on sex, ethnicity, and age (within 2 years); the
matched control was randomly selected from the pool. Approval for the study
was given by the institutional review board of the University of Hawaii and
of each participating hospital. Written informed consent was obtained from
each participant. The participation rates were 58% for cases and 52% for controls.
Complete questionnaire information was obtained from 727 matched pairs of
cases and controls. A blood sample was obtained for 548 cases and 656 controls.
Forty-four cases and 32 controls had to be excluded because their DNA had
been depleted in genotyping for diet-related genes in the original study;
therefore, our study had a total of 504 cases and 624 controls.
The questionnaire was administered during an in-person interview and
included detailed information on demographics; a quantitative food-frequency
questionnaire; a lifetime history of tobacco, alcohol, and aspirin use; a
history of recreational sports activities since age 18 years; a personal history
of various relevant medical conditions; a family history of colorectal cancer
in parents and siblings; information on height and weight at different ages;
and for women, a history of reproductive events and hormone use. The Surveillance
Epidemiology and End Results summary staging-information was extracted from
the Hawaii Tumor Registry and is defined as follows: in situ tumors (10% of
cases) had remained intraepithelial, localized tumors (47%) were confined
to the colon or rectum, regional tumors (37%) had either extended through
the muscularis to adjacent tissue or metastasized to regional mesenteric lymph
nodes, and distant tumors (6%) had metastasized to distant sites.
DNA was extracted from blood lymphocytes using a standard method (QIAamp
DNA Blood Midi Kit, Qiagen, Valencia, Calif). Genotyping for CCND1 was performed with forward primer (CCND1F): AGTTCATTTCCAATCCGCCC
and reverse primer (CCND1R): TTTCCGTGGCACTAGGTGTC. Polymerase chain reaction
conditions consisted of an initial denaturation of 94 °C for 5 minutes,
followed up with 35 cycles of 94 °C for 30 seconds, 60 °C for 30 seconds,
and 72 °C for 30 seconds, with a final extension of 72 °C for 10 minutes.
The resulting 212 base pair (bp) polymerase chain reaction product was digested
with the restriction enzyme Moraxella species (MspI) and run on a 3% MetaPhor gel (Cambrex, Rockland,
Me), yielding 2 bands for the A allele (175 and 37 bp) and 3 bands for the
G allele (141, 37, and 34 bp) (Figure 1).
The statistical analysis used unconditional logistic regression to compute
odds ratios (ORs) and 95% confidence intervals (CIs).16 All
models were adjusted for the matching variables (age, sex, and ethnicity)
and for potential confounders (pack-years of cigarette smoking, lifetime recreational
physical activity [hours], body mass index [calculated as weight in kilograms
divided by the square of height in meters] 5 years ago, lifetime use of aspirin
[months], years of schooling, and intakes of nonstarch polysaccharides from
vegetables and calcium from foods and supplements). Because the ORs were similar
in men and women, results are presented for both sexes combined. Gene-dosage
effects were modeled by assigning a value of 1, 2, or 3 to the genotype variable
according to the number of A alleles (0, 1, and 2 A alleles, respectively).
The likelihood ratio test was used to statistically test interaction among
certain variables with respect to colorectal cancer. The test compares a main
effects no interaction model with a fully parameterized model containing all
possible interaction terms for the variables of interest. Polytomous logistic
models16 were performed, comparing cases by
stage and subsite of cancer to all eligible controls. Such models were used
to compare the risk of in situ/localized and regional/distant cancer, and
the risk of colon and rectal cancer by stage. The risk between groups was
compared statistically by a Wald test.16 Genotype
frequencies were tested for deviation from the Hardy Weinberg equilibrium
with the χ2 test. Statistical significance was defined as P<.05. All analyses were performed with SAS statistical
software version 8.2 (SAS Institute Inc, Cary, NC).
The characteristics of colorectal cancer cases and controls are shown
in Table 1.12- 14 Cases
were comparable with controls with regard to age, sex, and ethnicity but were
somewhat less educated, more likely to have a family history of colorectal
cancer, and less likely to have used aspirin regularly. Cases had smoked more
cigarettes and exercised less in their lifetimes. They were also heavier and
consumed more calories, less calcium, less folate, and less dietary fiber
(measured as nonstarch polysaccharides) from vegetable sources than controls.
Table 2 presents the distributions
of cases and controls with colorectal cancer by CCND1 genotype.
Based on the controls, the frequency for the putative high-risk A allele was
0.49, 0.43, and 0.57, in Japanese, white, and Hawaiian participants, respectively.
The genotype distributions were consistent with Hardy Weinberg equilibrium
in each ethnic group (Japanese, P = .31; white, P = .60; Hawaiian, P = .21). Overall,
the A allele was associated with a 30% increase in colorectal cancer risk
(OR, 1.3; 95% CI, 1.0-1.7) and the CCND1 870 AA genotype
was associated with a 50% increased risk of colorectal cancer (OR, 1.5; 95%
CI, 1.0-2.1), with a statistically significant gene-dosage effect (P = .03). This association was suggested similarly in each sex (data
not shown) and in white and Hawaiian participants, but not in Japanese. A
stronger effect for the A allele was suggested for rectal cancer vs colon
cancer. The ORs for the GG, GA, and AA genotypes were 1.0, 1.8 (95% CI, 1.1-2.9),
and 2.2 (95% CI, 1.9-3.9), respectively, for rectal cancer (P = .006 for genetic trend) and 1.0, 1.1 (95% CI, 0.8-1.6), and 1.3
(95% CI, 0.9-1.9), respectively, for colon cancer (P =
.17 for genetic trend).
Table 3 shows the same analysis
further stratified by stage of disease at diagnosis with the use of polytomous
logistic regression models. The association of the CCNDI 870 AA genotype with an increased risk of colorectal cancer was stronger
for advanced stage disease with ORs of 1.7 (95% CI, 1.1-2.5) and 1.9 (95%
CI, 1.2-3.1) for the GA and AA genotype, respectively, compared with the GG
genotype. This association showed a statistically significant gene-dosage
effect (P = .008). In contrast, no statistically
significant association was found with the A allele for early-stage colorectal
cancer among all participants combined. The OR for the presence of the A allele
among patients with advanced colorectal cancer was significantly different
from that in patients with early-stage disease (P =
.048). In the corresponding race-specific analyses, a nonsignificant association
was suggested for early-stage disease in white and Hawaiian participants,
and the association with advanced disease was observed or suggested in all
ethnic groups. The number of cases with distant stage at diagnosis was too
small to allow for separate analyses.
Table 4 compares the effect
of the A allele on colorectal cancer risk stratified by stage at diagnosis
and anatomical subsite with the use of polytomous logistic regression. Associations
were suggested for the A allele with late-stage colon cancer and early-stage
rectal cancer; however, none were statistically significant. The strongest
effect for presence vs absence of the A allele was found for late-stage rectal
cancer (OR, 2.8; 95% CI, 1.3-6.0), which was also significantly different
from the early-stage colon cancer (OR, 1.0; 95% CI, 0.7-1.5; P = .02).
Analyses for interaction showed no modifying effect of age or family
history on the association of the A allele with colorectal cancer.
In this population-based case-control study, we found that the CCND1 870 AA genotype was associated with a 50% increased
risk of colorectal cancer, with a statistically significant gene-dosage effect
(P = .03). The association with the A allele was
significantly stronger for advanced stage disease than for early stage disease.
The observed effect was consistent between sexes, across ethnic groups (particularly
for advanced disease), and stronger for rectal cancer.
Amplification and/or overexpression of the CCND1 gene
have been described in several forms of human cancer and associated with increased
cell proliferation and poor prognosis.3- 5 With
regard to colorectal cancer, overexpression of CCND1 is observed in 30% of
the tumors and expression of an antisense to CCND1 complementary
DNA has been shown to inhibit the proliferation of human colon cancer cells,
as well as their tumorigenicity in nude mice.17 The
A870G polymorphism in exon 4 of the CCND1 gene is
associated with a splice site variation coding for 2 messenger RNA transcripts.6 Transcript b, which skips
exon 5 and reads into intron 4, does not contain the exon 5 destruction box
sequence, resulting in a protein with a longer half-life. It has been shown
that, although both the A and G alleles encode the 2 transcripts, the A allele
preferentially encodes the altered transcript leading to a state of increased
CCND1 level, even in the heterozygous state.6,18
Five previous studies have reported on the association of the CCND1 A870G polymorphism and colorectal cancer. Kong et
al9 found that patients with 1 or 2 copies
of the CCND1 870A allele who also carry a mutation
in a DNA mismatch repair gene develop hereditary nonpolyposis colorectal cancer
an average of 11 years earlier than mismatch repair gene mutation carriers
with the GG genotype. In contrast, Bala and Peltomäki19 found
no correlation between the A allele and age of onset among 146 affected mismatch
repair mutation carriers; however, the presence of the variant transcript b in blood or healthy mucosa was associated with a significantly
lower age of onset compared with individuals with transcript a only (35 vs 46 years; P = .02). Porter et
al11 also showed that the CCND1 870A allele was overrepresented in 107 non–hereditary nonpolyposis
colorectal cancer familial cases of colorectal cancer compared with 171 patients
without cancer. In the same study, an overrepresentation of the A allele was
also observed in 128 "sporadic" colorectal cancer cases, which did not quite
reach statistical significance (P = .08). Kong et
al10 recently reported on a hospital-based
case-control study of 156 white patients with colorectal cancer younger than
60 years and 152 matched-dermatology control patients. Compared with the GG
genotype, Kong et al10 found that the AA genotype
was associated with an elevated OR of 2.6 (95% CI, 1.4-5.2), whereas the GA
genotype was unrelated to risk (OR, 1.1; 95% CI, 0.6-1.8). Finally, in a hospital-based
study of 100 patients and 101 blood donors, McKay et al5 reported
a lack of association between the CCND1 870A allele
and colorectal cancer; however, survival was significantly shorter in patients
with a high level of CCND1 expression in their tumors (>50% cells demonstrating
immunoreactivity). Overall, these past studies, which were hospital-based
and relatively small, were suggestive of a possible association of the CCND1 G870A allele with progression of colorectal tumors.
Our study expands these findings with a population-based design and
a larger sample size. We also report ethnicity, which has not been studied
before. The A allele is particularly common in Native Hawaiians, a group of
patients who often present at a late stage and experience a poorer stage-adjusted
cause-specific survival for various cancers, including colorectal cancer,
compared with white patients.20,21 Similarly,
the weaker effect for the A allele observed among Japanese participants in
this study would, if confirmed, be consistent with the early presentation
and better cause-specific survival of Japanese patients with colorectal cancer
in Hawaii.20,21 Colorectal cancer
is rarely curable when the disease has spread outside the large intestine.
Given the high frequency of the A allele (0.43-0.57 in the 3 ethnic groups)
and the stronger association for advanced disease, this polymorphism may be
responsible for a sizable portion of the morbidity and mortality from colorectal
cancer. If confirmed, this association may have implications for the colorectal
cancer screening and treatment of the A allele carriers.
Other methodological aspects of our study deserve consideration. Differences
in detection rate by genotype appears to be an unlikely explanation for our
results because the association was observed in 2 ethnic groups with markedly
different socioeconomic status and screening practices.21,22 This
is reflected in the proportions of in situ tumors in our population-based
case-series (Japanese, 10%; white, 9%; Hawaiian, 4%). Confounding by other
variables is also unlikely because the effects of known risk factors were
thoroughly investigated in the analysis. The association was observed in several
ethnic groups, arguing against residual confounding by ethnicity. Moreover,
the frequency of the CCND1 870A allele in our white
control participants (0.43) is very similar to that from previous reports,10,11 and the genotype frequencies were
in Hardy Weinberg equilibrium, arguing against selection bias. This is consistent
with the fact that characteristics of participants who gave blood were very
similar to those of all interviewed participants in this study.13
In conclusion, these data provide strong evidence that the CCND1 870A
allele may be associated with colorectal cancer, and particularly with forms
of the disease that result in severe morbidity and mortality.