RESEARCH ARTICLE
A systematic analysis of the association studies between
CASP8 D302H polymorphisms and breast cancer risk
KAN HE1,2, WEI LI1,2, YINLIANG ZHANG1, YI HONG1,2, GUOYING WU1,2 and
DAHAI LIU*1,2
1 School of Life Sciences, Anhui University, Hefei City, Anhui 230601, P. R. China
2 Center for Stem Cell and Translational Medicine, Anhui University, Hefei City,
Anhui 230601, P. R. China
Contributed equally: KAN HE , WEI LI , YINLIANG ZHANG.
*For correspondence. E-mail: [email protected].
Abstract
Caspase 8 (CASP8) is a regulator of apoptosis, whose genetic variation has been reported to be
associated with the risk of various cancers. Specifically, the single nucleotide polymorphism (SNP)
rs1045485, which generates the substitution D302H in CASP8, is likely to be associated with breast
cancer. Several previous studies have reported the association of CASP8 D302H polymorphism with
breast cancer; however, the results are inconsistent. To validate the association between CASP8 D302H
polymorphism and breast cancer risk, we performed an updated meta-analysis of 18 studies including
27,807 cases and 32,332 controls. We tested the overall association between this SNP and breast cancer
susceptibility and stratified subgroups based on countries where cases are from. We confirmed a
significant correlation between CASP8 D302H polymorphism and the reduced breast cancer
susceptibility in population from U.K., Germany and Poland, but no significant association was
observed in other countries, such as Finland or U.S.. Our findings indicate the relationship of SNP
CASP8 D302H and breast cancer would not be universal but only be sensitive in some particular
European countries. The genetic difference for diverse countries may be useful in individual and
precision medicine or health.
Keywords: Breast cancer; CASP8; rs1045485 polymorphism; meta-analysis
Introduction
CASP8 encodes a member of the cysteine-aspartic acid protease (caspase) family, which is a cysteine
peptidase that can activate various cellular proteases or proteins, leading to apoptosis through the
FAS/FASLG-mediated apoptosis pathway. CASP8 is located on chromosome 2q33-34, harboring 10
exons that span ∼30 kb, in which there were at least 168 single nucleotide polymorphisms (SNPs)
reported for CASP8, mostly rare or non-coding. Several studies have evaluated the associations
between some of these CASP8 SNPs and risk of various cancers (Couch et al., 2009; Lubahn et al.,
2010; Pittman et al., 2008; Ramus et al., 2008). Two functional SNPs, rs3834129 named as
six-nucleotide insertion/deletion (−652 6N ins/del) in the promoter region and rs1045485 named as
D302H in the coding region of CASP8, are both thought to be important in cancer etiology. The SNP
rs3834129 is considered to be associated with susceptibility to multiple cancers while the effect of SNP
rs1045485 is mainly associated with the risk of breast cancer (Sun et al., 2007).
Previous studies on the association between rs1045485 polymorphism and breast cancer indicate an
inconsistent result. Several studies observed a significant association (Cox et al., 2007; Long et al.,
2013; MacPherson et al., 2004a), while some other studies showed no association or even no
polymorphism (Guan et al., 2014; Michailidou et al., 2015). Recent meta-analysis study pooled the
result from four case-control studies, including 18791 breast cancer cases and 20318 controls of
Caucasians (Sergentanis & Economopoulos, 2010). Another meta-analysis study including cases of
various cancer diseases showed that rs1045485 was found to be only associated with breast cancer risk
(Ji et al., 2014). Thus, we conducted an updated meta-analysis by adding the latest data, avoiding
sample overlapping and stratifying subgroups with the aim of gaining a more reliable evaluation of the
association between rs1045485 polymorphism and breast cancer susceptibility. We here performed a
large-scale meta-analysis including 60139 individuals (27807 cases and 32332 controls) from European
countries, U.S. and Australia, aimed to find out the exact relationship between SNP rs1045485 and
breast cancer risk across different countries.
Materials and Methods
Eligibility of relevant studies
All original articles published in English that examined the association of the rs1045485 polymorphism
with breast cancer (published before June 2015) were considered for our meta-analysis. The PubMed
and Web of Science were searched to identify appropriate studies. The following combinations of terms
were used in our database searches: (“breast cancer”) and (“rs1045485” or “CASP8 D302H”).
Furthermore, the searches were supplemented by references cited in other papers. The flowchart of our
analysis was shown in Figure 1.
To include relevant studies in this meta-analysis, the following criteria were used: (1) studies assessed
linkage of rs1045485 polymorphism with breast cancer risk; (2) female breast cancer patients / breast
cancer cases should be diagnosed explicitly; (3) controls should be unrelated cancer-free individuals /
case-control design; (4) reported in English. When multiple reports had overlapping sample
populations, only the study with largest sample size was retained.
The studies were excluded if they were: (1) data were re-used on the same polymorphism; (2) control
genotype distributions not in Hardy-Weinberg equilibrium (HWE); (3) Incomplete reporting of
genotype frequencies.
Data extraction
For each eligible study, the following information was extracted: the first author, year of publication,
ethnicity of participants, source of controls, number of genotyped cases/controls, method for quality
control of genotyping result. The data were primarily extracted from tables and supplemented by
significant information presented in texts and/or figures.
Statistical analysis
HWE was assessed in controls for each study. The chi-square goodness of fit is used to test deviation
from HWE. Studies were considered to deviate from HWE at P<0.05(Guo & Thompson, 1992). The
inconsistency index, I2, was calculated to evaluate the variation among studies owing to heterogeneity
(0%–25% was considered to have no heterogeneity; 25%–50% was considered to have moderate
heterogeneity; 50%–75% was considered to have large heterogeneity; 75%–100% was considered to
have extreme heterogeneity)(Higgins et al., 2003). The data were combined using logistic regression
with the fixed-effects pooling model if there was no or moderate heterogeneity (I2 < 50%).
Alternatively, the random effects model was used (I2 > 50%). Sensitivity analysis was performed by
excluding one study at a time to determine the corresponding magnitude of the weight of each study to
the summary results. The most biologically fit genetic model was selected according to the
comprehensive effect of the gene using logistic regression. The association between rs1045485
polymorphism and breast cancer risk was evaluated using the odds ratio (OR) and the 95% confidence
interval (CI). Funnel plots, used to observe the publication bias, were complemented with Egger’s
regression and Begg’s rank correlation test (P > 0.10). The statistical analyses were performed using
STATA version 11.2 (Stata Corporation, USA).
Results and Discussion
Study characteristics
Six articles including twenty-two studies were identified to meet the inclusion criteria, and the details
were shown in Table 1. We thoroughly reviewed these articles to detect overlapping samples. Shephard
et al used a staged study design from three data sets, the United Kingdom, Germany and Utah
(Shephard et al., 2009). Study of German data set by Shephard et al and Frank et al consisted of the
same breast cancer cases and controls of German patients; we included only the study by Shephard et al.
Cox et al included data from 14 studies. Study of U.K. data set by Shephard et al, study of Sheffield by
Cox et al and MacPherson et al were found to share common sample sources of North European origin
and resident in the Sheffield area (Cox et al., 2007; MacPherson et al., 2004a). Therefore, only the
study by Shephard et al, which had the largest sample size and latest data, was used in our
meta-analysis (Shephard et al., 2009). Studies by Sigurdson et al and Cox et al were found to share
common sample sources of U.S. radiologic technologists, and the study by Sigurdson et al was used in
our meta-analysis because of the slightly larger sample size (Sigurdson et al., 2007). Finally, eighteen
studies including 27807 breast cancer cases and 32332 controls are used in our meta-analysis. We
checked HWE of each study and divided them into subgroups by whether the controls are deviated
from HWE.
Heterogeneity and model
All heterogeneity statistic I2 values were observed less than 25% in the present study, which indicated
that the appropriate pooling model should be fixed effects. Furthermore, using a suitable underlying
genetic model in genetic association studies is crucial for combining data biologically rather than just
statistically. Previous studies listed the association under different genetic models but without given a
best fit model, we here carefully selected the most likely genetic model for representing the association
between CASP8 D302H and breast cancer. The estimated OR(CC vs GG), OR(CG vs GG), and OR(CC vs CG)
were 0.73 (95% CI: 0.64-0.83), 0.90 (95% CI: 0.86-0.94) and 0.81 (95% CI: 0.71-0.93). According to
the methodology for genetic model selection developed by Thakkinstian et al(Thakkinstian et al., 2005),
the genetic model was most likely to be co-dominant. After a sensitivity analysis, no individual study
was found to affect the overall result robustly, which implied the magnitude of the summary evaluation.
Gene effect
The overall results of the genetic analysis indicated a significant association between CASP8 D302H
and a reduced breast cancer risk (OR=0.89, 95%CI: 0.86-0.92) (Figure 2 and Figure 3).
To test whether the studies whose controls were not in HWE affect the pooling result, we stratified two
subgroups. Significant association between breast cancer risk and CASP8 D302H are shown in both
subgroups which HWE p-value > 0.05 (OR=0.87, 95%CI: 0.84-0.91) and HWE p-value < 0.05
(OR=0.91, 95%CI: 0.86-0.96) (Figure 2), indicating that if the studies are in Hardy–Weinberg
equilibrium could be neglected in this study. To gather as much data as we could, we included the
studies no matter if its controls were in Hardy–Weinberg equilibrium.
We then performed the stratified analysis according to the countries where the cases were from.
Significant association was observed in the studies of U.K. (OR=0.89, 95%CI: 0.84-0.95), Germany
(OR=0.85, 95%CI: 0.75-0.97) and Poland (OR=0.82, 95%CI: 0.72-0.93). Campa et al performed a
multi-ethnicity case-control study within the National Cancer Institute’s Breast and Prostate Cancer
Cohort Consortium (BPC3), including European descent, Latino, African American, Asian American
(mostly of Japanese origin) and Native Hawaiian (Campa et al., 2011). Therefore, we identified it as
multi-ethnicity, not included in U.S. or any European countries. Result shows significant association in
this study. However, in Australia, Finland, Spain, Sweden and U.S., there is no significant association
between CASP8 D302H and breast cancer risk (Figure 3). No heterogeneity was found between groups
or between studies. Hence, the association between CASP8 D302H and breast cancer risk may be
country variable.
Previously, Sergentanis et al investigated 4 studies of association between CASP8 D302H and breast
cancer risk, the summary of each study was pooled into a meta-analysis by using dominant model, but
without distinguishing the counties where cases were from (Sergentanis & Economopoulos, 2010).
Overall result indicates the CASP8 D302H polymorphism may be associated with reduced breast
cancer risk in Caucasian population (OR=0.87, 95%CI: 0.83-0.92). A case-control study in Han
Chinese population found no polymorphism of CASP8 D302H (Guan et al., 2014). A recent study in
African-American population showed that CASP8 D302H were deviated from Hardy-Weinberg test
and had a MAF<5% in African-ancestry populations and it was not replicated in all previous studies of
African-ancestry populations (Long et al., 2013). Therefore, the association between CASP8 D302H
and breast cancer risk may only occur in some specific countries.
While meta-analysis shed light on the trend of association between SNP and disease risk,
gene-environment interactions have the potential to reveal the biological processes leading to disease,
identify the most relevant risk factors, and improve the accuracy of epidemiological risk models. Travis
et al tested gene-environment interactions in a large prospective UK cohort, studying the effects of 12
polymorphisms, including CASP8 D302H, in relation to ten established environmental risk factors (age
at menarche, parity, age at first birth, breastfeeding, menopausal status, age at menopause, use of
hormone replacement therapy, body-mass index, height, and alcohol consumption) (Travis et al., 2010).
Results showed that the risks of breast cancer associated with CASP8 D302H do not vary significantly
with most of the established environmental risk factors, except for alcohol consumption. Replication
studies for CASP8 D302H and alcohol consumption are reported, suggesting CASP8 D302H is a
complicated SNP and worth more investigation to explore the plausible biological mechanisms that can
explain this association (Fletcher & Dudbridge, 2014; Nickels et al., 2013). Some of the underlying
causes that can partly explain the differences observed by country may be as follows, such as weak
association, statistical power issues in some countries, possible different distribution of breast cancer
subtypes in the different countries, possible gene-environment interaction etc.
The publication bias was accessed using Begg’s (p = 0.495) and Egger’s (p = 0.517) tests. The funnel
plot displayed a symmetric shape (Figure 4), indicating the absence of a publication bias for both
positive and negative or non-significant findings from published studies.
In conclusion, CASP8 has long been considered as a breast cancer susceptibility gene, at least 168
single nucleotide polymorphisms (SNP) have been reported for CASP8, mostly rare or non-coding. We
here provide a comprehensive meta-analysis on CASP8 D302H polymorphism, also known as
rs1045485, including 27807 cases and 32332 controls from European countries, U.S. and Australia. An
overall trend indicates a protective effect of the polymorphism. This is in accordance with previous
ones, which had been performed on smaller numbers of studies (Breast Cancer Association, 2006;
Janssens et al., 2009). However, a recent GWAS study detects no association between rs1045485 and
breast cancer risk in European women (Michailidou et al., 2015). To address this problem, we gathered
as many studies as we could and divided them into subgroups by the countries where cases were from.
Stratified analysis implies the protective association seems to pertain only to part of Caucasian, which
mainly resident in U.K., Germany and Poland. To the people who are from Finland, Spain, Sweden,
Australia and U.S., no significant association is observed. In addition, rs1045485 lacks polymorphism
in Chinese and African-American population, suggesting that the association between CASP8 D302H
and breast cancer risk may be country-sensitive, the fact of which may be due to the effect of
gene-environment interaction (Guan et al., 2014; Long et al., 2013).
Since the early investigations were taken place in populations in UK or Germany, many follow-up
studies focused on these two countries too. Even we already excluded some large overlapping studies,
the majority source data are still from U.K. or Germany. On one hand, the lack of CASP8 D302H
polymorphism in ethnicity other than Caucasians should be confirmed by genotyping in other ethnicity.
On the other hand, in Caucasians, or in European descent, detailed investigation should be performed
to explain why the differences occurred in same ethnic group who lived in different countries. The
difference in the MAF and different LD structure in these populations may also partly explain the
difference between ethnic groups for the association between rs1045485 and breast cancer risk. Like
the study of Michailidou et al, increasing the size of the population can significantly improve the test
results and obtain a deeper and clearer view of the function of CASP8 D302H (Michailidou et al.,
2015).
Acknowledgement
We acknowledge financial support by the Scientific Research Foundation and Academic & Technology
Leaders Introduction Project, and 211 Project of Anhui University (10117700023,
02303203-32030081), and The Student Research Training Program of Anhui University (J18520131),
and Natural Science Foundation Project of Anhui Province (1508085MH189, 1508085QC63) as well
as The Education Revitalization Project of Anhui Province: Stem Cell and Translational Medicine
(Y05201374).
References
Breast Cancer Association, C. (2006). Commonly studied single-nucleotide polymorphisms and breast
cancer: results from the Breast Cancer Association Consortium. J Natl Cancer Inst 98,
1382-96.
Campa, D., Kaaks, R., Le Marchand, L., Haiman, C. A., Travis, R. C., Berg, C. D., et al. (2011).
Interactions between genetic variants and breast cancer risk factors in the breast and prostate
cancer cohort consortium. J Natl Cancer Inst 103, 1252-63.
Couch, F. J., Wang, X., McWilliams, R. R., Bamlet, W. R., de Andrade, M. &Petersen, G. M. (2009).
Association of breast cancer susceptibility variants with risk of pancreatic cancer. Cancer
Epidemiol Biomarkers Prev 18, 3044-8.
Cox, A., Dunning, A. M., Garcia-Closas, M., Balasubramanian, S., Reed, M. W., Pooley, K. A., et al.
(2007). A common coding variant in CASP8 is associated with breast cancer risk. Nat Genet
39, 352-8.
Fletcher, O.Dudbridge, F. (2014). Candidate gene-environment interactions in breast cancer. BMC Med
12, 195.
Frank, B., Bermejo, J. L., Hemminki, K., Klaes, R., Bugert, P., Wappenschmidt, B., et al. (2005). Re:
Association of a common variant of the CASP8 gene with reduced risk of breast cancer. J Natl
Cancer Inst 97, 1012; author reply 1012-3.
Guan, Y. P., Yang, X. X., Yao, G. Y., Qiu, F., Chen, J., Chen, L. J., et al. (2014). Breast cancer
association studies in a Han Chinese population using 10 European-ancestry-associated breast
cancer susceptibility SNPs. Asian Pac J Cancer Prev 15, 85-91.
Guo, S. W.Thompson, E. A. (1992). Performing the exact test of Hardy-Weinberg proportion for
multiple alleles. Biometrics 48, 361-72.
Higgins, J. P., Thompson, S. G., Deeks, J. J. &Altman, D. G. (2003). Measuring inconsistency in
meta-analyses. BMJ 327, 557-60.
Janssens, A. C., Gonzalez-Zuloeta Ladd, A. M., Lopez-Leon, S., Ioannidis, J. P., Oostra, B. A., Khoury,
M. J., et al. (2009). An empirical comparison of meta-analyses of published gene-disease
associations versus consortium analyses. Genet Med 11, 153-62.
Ji, G. H., Li, M., Cui, Y. &Wang, J. F. (2014). The relationship of CASP 8 polymorphism and cancer
susceptibility: a meta-analysis. Cell Mol Biol (Noisy-le-grand) 60, 20-8.
Long, J., Zhang, B., Signorello, L. B., Cai, Q., Deming-Halverson, S., Shrubsole, M. J., et al. (2013).
Evaluating genome-wide association study-identified breast cancer risk variants in
African-American women. PLoS One 8, e58350.
Lubahn, J., Berndt, S. I., Jin, C. H., Klim, A., Luly, J., Wu, W. S., et al. (2010). Association of CASP8
D302H polymorphism with reduced risk of aggressive prostate carcinoma. Prostate 70,
646-53.
MacPherson, G., Healey, C. S., Teare, M. D., Balasubramanian, S. P., Reed, M. W., Pharoah, P. D., et
al. (2004a). Association of a common variant of the CASP8 gene with reduced risk of breast
cancer. J Natl Cancer Inst 96, 1866-9.
MacPherson, G., Healey, C. S., Teare, M. D., Balasubramanian, S. P., Reed, M. W. R., Pharoah, P. D. P.,
et al. (2004b). Association of a common variant of the CASP8 gene with reduced risk of
breast cancer. Jnci-Journal of the National Cancer Institute 96, 1866-1869.
Michailidou, K. & Beesley, J. & Lindstrom, S. & Canisius, S. & Dennis, J. & Lush, M. J., et al. (2015).
Genome-wide association analysis of more than 120,000 individuals identifies 15 new
susceptibility loci for breast cancer. Nat Genet 47, 373-80.
Nickels, S., Truong, T., Hein, R., Stevens, K., Buck, K., Behrens, S., et al. (2013). Evidence of
gene-environment interactions between common breast cancer susceptibility loci and
established environmental risk factors. PLoS Genet 9, e1003284.
Pittman, A. M., Broderick, P., Sullivan, K., Fielding, S., Webb, E., Penegar, S., et al. (2008). CASP8
variants D302H and -652 6N ins/del do not influence the risk of colorectal cancer in the
United Kingdom population. Br J Cancer 98, 1434-6.
Ramus, S. J., Vierkant, R. A., Johnatty, S. E., Pike, M. C., Van Den Berg, D. J., Wu, A. H., et al. (2008).
Consortium analysis of 7 candidate SNPs for ovarian cancer. Int J Cancer 123, 380-8.
Sergentanis, T. N.Economopoulos, K. P. (2010). Association of two CASP8 polymorphisms with breast
cancer risk: a meta-analysis. Breast Cancer Res Treat 120, 229-34.
Shephard, N. D., Abo, R., Rigas, S. H., Frank, B., Lin, W. Y., Brock, I. W., et al. (2009). A breast cancer
risk haplotype in the caspase-8 gene. Cancer Res 69, 2724-8.
Sigurdson, A. J., Bhatti, P., Doody, M. M., Hauptmann, M., Bowen, L., Simon, S. L., et al. (2007).
Polymorphisms in apoptosis- and proliferation-related genes, ionizing radiation exposure, and
risk of breast cancer among U.S. Radiologic Technologists. Cancer Epidemiol Biomarkers
Prev 16, 2000-7.
Sun, T., Gao, Y., Tan, W., Ma, S., Shi, Y., Yao, J., et al. (2007). A six-nucleotide insertion-deletion
polymorphism in the CASP8 promoter is associated with susceptibility to multiple cancers.
Nat Genet 39, 605-13.
Thakkinstian, A., McElduff, P., D'Este, C., Duffy, D. &Attia, J. (2005). A method for meta-analysis of
molecular association studies. Stat Med 24, 1291-306.
Travis, R. C., Reeves, G. K., Green, J., Bull, D., Tipper, S. J., Baker, K., et al. (2010).
Gene-environment interactions in 7610 women with breast cancer: prospective evidence from
the Million Women Study. Lancet 375, 2143-51.
Received 6 June 2016, in revised form 22 August 2016; accepted 26 August 2016
Unedited version published online: 29 August 2016
Tables
Table 1. Characteristics of studies included in this meta-analysis.
Article (ref) Year Region Study Case Control MAF
HWE p-value GG CG CC GG CG CC Case Control
Campa, D. et al (Campa
et al., 2011)
2011 Europe and U.S. BPC3 6414 1539 110 8834 2345 197 21.8% 24.1% 0.004
Shephard, N. D. et al
(Shephard et al., 2009)
2009 U.K. SBCS 896 292 17 839 314 27 27.1% 31.2% 0.708
Germany GC-HBOC 275 76 4 815 260 23 23.7% 27.9% 0.672
U.S. UBCS 557 130 15 338 74 10 22.8% 22.3% 0.019
Sigurdson, A. J. et
al(Sigurdson et al.,
2007)
2007
U.S. RT 660 185 7 802 232 22 23.4% 26.1% 0.283
Cox, A. et al(Cox et al.,
2007)
2007 Australia ABCFS/kConFaB 1117 307 22 433 142 8 24.3% 27.1% 0.339
U.K. BBC 440 135 8 435 142 15 25.9% 29.1% 0.407
Germany GENICA 466 122 11 464 137 15 24.0% 27.1% 0.206
Germany HBCS 771 205 15 745 246 15 23.7% 27.4% 0.295
Finland Helsinki 680 135 8 712 160 5 18.3% 19.4% 0.212
U.K. ICR_FBCS 772 238 12 1082 352 31 25.6% 28.3% 0.706
Finland Kuopio 374 70 3 349 80 0 17.0% 18.6% 0.033
U.S. Mayo Clinic 600 176 14 603 201 24 25.8% 30.1% 0.151
Poland Poland 1590 430 25 1714 555 45 23.5% 27.9% 0.993
U.K. SEARCH 3117 827 66 3330 949 81 23.9% 25.5% 0.164
U.K. Sheffield* 672 212 14 675 265 24 26.7% 32.5% 0.739
Sweden SASBAC 1164 328 20 1139 310 37 24.3% 25.8% 0.005
Spain CNIO 403 97 14 417 137 8 24.3% 27.2% 0.387
U.S. USRT* 578 158 7 783 224 20 23.1% 25.7% 0.398
Frank, B. et al (Frank et
al., 2005)
2005 Germany Frank's* 275 76 4 815 260 23 23.7% 27.9% 0.672
MacPherson, G. et al
(MacPherson et al.,
2004b)
2004 U.K. Sheffield* 718 221 15 675 265 24 26.3% 32.5% 0.739
U.K. East Anglia 1468 358 22 1591 450 41 21.8% 25.6% 0.168
*Excluded from meta-analysis for large overlapping with other study.
Figure 1. The flowchart of our analysis.
Figure 2. Stratified analysis based on HWE for the association between rs1045485 polymorphism
and breast cancer risk using a co-dominant genetic model.
Figure 3. Stratified analysis based on the country of sources for the association between
rs1045485 polymorphism and breast cancer risk using a co-dominant genetic model.
Figure 4. Begg’s funnel plot displaying a symmetric shape.
The horizontal axis represents the standard error of OR value, and the vertical axis represents the OR
value of each study.