- Email: [email protected]
Ovarian cancer survival and polymorphisms in hormone and DNA repair pathway genes
Cancer Letters 251 (2007) 96–104 www.elsevier.com/locate/canlet
Ovarian cancer survival and polymorphisms in hormone and DNA repair pathway genes Christina M. Nagle a, Georgia Chenevix-Trench b, Penelope M. Webb a, Amanda B. Spurdle c,* a
c
Cancer and Population Studies, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia b Cancer Genetics, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia Molecular Cancer Epidemiology, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia Received 14 September 2006; accepted 6 November 2006
Abstract We evaluated the association between 21 polymorphisms in hormone and DNA repair pathway genes and survival among 454 Australian women diagnosed with invasive epithelial ovarian cancer. The cohort was followed for mortality using personal identifiers which were linked to state cancer registry records and the Australian National Death Index. The mean follow-up time after ovarian cancer diagnosis was 4.63 years (all women) and 8.07 years for the censored group (those alive or dead from non-ovarian cancer causes). Two hundred and eighty-eight (63%) ovarian cancer deaths occurred during the follow-up period. No association was observed for the vast majority of polymorphisms, but there was suggestive evidence for altered risk of ovarian cancer death associated with the CYP17 5 0 UTR C allele (HR 1.30; 95% CI = 1.02–1.68, p = 0.04), and for the SRD5A2 V89L C allele (HR 0.79; 95% CI = 0.62–1.01, p = 0.06). These results are interesting given tentative evidence that both of these variants are also associated with increased predisposition to ovarian cancer in our extended Australian study, and in other published studies. However, given the marginal significance of these associations and the large number of tests performed, independent replication will be necessary to validate these novel findings. Ó 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: DNA repair genes; Epithelial ovarian cancer; Hormone pathway genes; Polymorphisms; Survival
1. Introduction Ovarian cancer is a significant cause of morbidity and mortality in Australia and worldwide [1–3]. It is estimated that 1 in 104 Australian women will
* Corresponding author. Tel.: +61 7 3362 0371; fax +61 7 3362 0105. E-mail address: [email protected] (A.B. Spurdle).
develop ovarian cancer in their lifetime, and over half of women with the disease are expected to die within 5 years of diagnosis [1]. Most women present with advanced disease at diagnosis and despite high rates of clinical response to first-line treatment, many women experience relapse due to sub-clinical disease that persists after completion of initial treatment [4]. This variability of response is thought to be related, in part, to individual genetic variation [5].
0304-3835/$ - see front matter Ó 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2006.11.011
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Variants in hormone pathway genes are good candidates for ovarian cancer risk [6]. Several epidemiological studies suggest a role for polymorphisms in hormone genes in ovarian cancer risk [7]. Support for a role of hormones in ovarian tumour progression is also growing. There is some evidence from the literature that tumour progesterone receptor (PR) positivity is an independent favourable predictor of progression-free [8] or total survival [9,10] particularly when the estrogen receptor (ER) is not expressed [11]. In addition, improved survival rate has been reported for patients with high serum progesterone, especially in combination with expression of PR [12]. Estrogen has been shown to stimulate growth of ovarian tumour cell lines expressing ER, and numerous estrogen-induced proteins have been shown to be upregulated in epithelial ovarian cancer [13]. AR levels have been associated with epidermal growth factor receptor (EGFR) levels, and overexpression of EGFR is in turn associated with poor prognosis [14,15]. Similarly, there is much evidence from known familial predisposition genes to suggest that DNA repair genes are involved in ovarian cancer risk, and accumulating evidence that alterations in DNA damage recognition and DNA repair may also play a role in outcome. Gene expression studies indicate that gene families involved in regulating DNA damage play a role in development of drug resistance, with upregulation of repair genes invariably associated with chemoresistance [16,17]. We have extended our ovarian cancer case-control candidate gene association analyses to test the hypothesis that polymorphisms in hormone and DNA repair pathway genes are associated with survival after a diagnosis of ovarian cancer. We assessed association between survival and 21 polymorphisms in hormone pathway and DNA repair pathway genes, for a large group of Australian women previously diagnosed with ovarian cancer. 2. Materials and methods 2.1. Study population We included 454 women diagnosed with primary invasive epithelial ovarian cancer between 1985 and 1997. Two thirds of the women (n = 296 [65%]) had participated in a large population-based case-control study of the etiology of ovarian cancer, the results of which have been published previously [18]. The remaining women (n = 158 [35%]) were ascertained as incident cases from the Royal Brisbane Hospital, Queensland, Australia [19]. The study was approved by the Institutional ethics review board of
97
the University of Queensland and all participating hospitals where the women were originally diagnosed and treated. 2.2. Clinical and pathologic data Clinical details including operative and histologic data were obtained retrospectively from the women’s medical records and pathology reports. The following clinical and pathologic data were collected: International Federation of Gynecology and Obstetrics (FIGO) stage, tumor histology (subtype), grade (well, moderate, poor, undifferentiated) and treatment details. Three quarters of the women (340/454) received platinum based chemotherapy as first-line treatment. The women were followed up for survival from the date of diagnosis until the date of death (from ovarian cancer) or censor date using personal identifiers which were linked to the Australian National Death Index (NDI) and state cancer registry records. Both the NDI and cancer registries used probabilistic record linkage software to match the women to their databases. To identify all ovarian cancer related deaths we obtained full details, including immediate and underlying cause/s of death and contributing factors, directly from death certificates for those decedents that had participated in the case-control study (n = 179), or indirectly through linkage to death records held at the Queensland cancer registry for those decedents recruited from the Royal Brisbane Hospital (n = 109). 2.3. Genotype detection DNA preparation and genotyping was as described previously [19]. Briefly, germline DNA was extracted from peripheral blood (collected at diagnosis) by a salt-precipitation method for women recruited through the Royal Brisbane Hospital and from uninvolved tissue from archival paraffin blocks, made at the time of diagnosis, for women ascertained through the population-based casecontrol study. Genotyping was performed in the course of ongoing analysis of candidate low penetrance ovarian cancer susceptibility genes [19–27]. ABI Taqman methodology was used for single nucleotide polymorphisms (all samples), and fluorescent PCR–PAGE methodology for the AR CAG repeat (peripheral blood DNA only (n = 239), due to archival DNA quality limitations). Additional assay details are available on request. Polymorphisms are described in Tables 2 and 3. 2.4. Statistical analysis The Kaplan–Meier technique was used to plot crude survival curves and estimate crude survival probabilities, and adjusted hazard ratios (HRs) and 95% confidence
98
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
intervals (95% CIs) were obtained from Cox regression models. The p-value for linear trend was calculated by the change in likelihood ratio statistic for entry of a linear term in the model, and thus was a v2 test on 1 degree of freedom. Analyses were adjusted for age at diagnosis (10-year age groups), FIGO stage, histologic subtype and histologic grade (moderate, poor and undifferentiated tumors were combined into a single group because survival did not vary for these three sub-groups but was significantly different from that for women with well differentiated tumors). All statistical analyses were performed using the Statistical Packages for Social Sciences for Windows, version 13.0 (SPSS Inc., Chicago, IL).
3. Results The characteristics of the study participants are shown in Table 1. Age at diagnosis ranged from 19 to 87 years, and the average age was 59 years. The most common subtype was serous (n = 274), followed by endometrioid (n = 62), clear cell (n = 31), mucinous (n = 29) and 58 ‘other’ histology which included mostly mixed mullerian and undifferentiated tumors. Patients were staged at laparotomy according to FIGO criteria. Of the 446 women
with known stage at diagnosis there were 83, 44, 278, and 41 with FIGO stages I, II, III, and IV, respectively. The group included 56 women with well differentiated tumors, 128 with moderately differentiated tumors, and 241 patients with poorly differentiated and undifferentiated tumors. The mean follow-up time after ovarian cancer diagnosis was 4.63 years (all women) and 8.07 years for the censored group (those alive or dead from another cause). Two hundred and eighty-eight (63%) ovarian cancer deaths occurred during the follow-up period, and the crude 5-year survival rate was 44%. Late FIGO stage, older age at diagnosis, histologic subtype and higher tumor grade were associated with survival and were included in all subsequent models. The majority of polymorphisms were not significantly associated with altered survival (Tables 2 and 3). Among the hormone pathway polymorphisms (Table 2), there was suggestive evidence for an association of similar magnitude with the TC and CC genotypes defined by the CYP17 5 0 UTR T > C polymorphism (Fig. 1), and carriers of the C allele showed a 30% increase in death hazard rate after adjustment for stage, age, grade and histology (HR 1.30, 95% CI 1.02–1.68; p = 0.04). We also found evidence of a modest, borderline significant, protective effect against early death among women who carried the GC
Table 1 Association between clinical and pathological factors and ovarian cancer survival
Age group <50 years 50–59 years 60–69 years 70+
Total no.a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
106 123 131 94
57 49 38 29
1.0 1.37 (0.97–1.95) 1.84 (1.31–2.59) 2.30 (1.60–3.30)
1.0 1.09 (0.74–1.62) 1.34 (0.91–1.97) 1.57 (1.04–2.36)
pc
0.16 FIGO stage I II III IV
83 44 278 41
86 76 31 11
1.0 2.43 (1.12–5.26) 10.32 (5.76–18.51) 19.17 (9.96–36.90)
1.0 1.92 (0.82–4.47) 6.99 (3.49–13.98) 11.79 (5.37–25.86) <0.01
Histological subgroup Serous Mucinous Endometrioid Clear cell Other
274 29 62 31 58
35 78 67 57 42
1.0 0.25 0.32 0.51 0.77
(0.12–0.51) (0.20–0.49) (0.29–0.87) (0.54–1.10)
1.0 1.02 0.56 1.88 0.78
(0.52–2.34) (0.33–0.87) (1.04–3.40) (0.52–1.15) 0.01
Histologic grade Well differentiated Moderately differentiated Poorly and undifferentiated
56 128 241
74 49 35
1.0 2.54 (1.49–4.31) 3.86 (2.34–6.37)
1.0 1.80 (1.01–3.19) 1.96 (1.12–3.44) 0.08
a b c
Column numbers (n) do not all sum to total because some data missing. Adjusted for age group, FIGO stage, histologic subgroup, histologic grade (1 vs. 2–4). p-value for heterogeneity for histological subgroup, p trend for all other comparisons.
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
99
Table 2 Association between hormone-metabolising genes and ovarian cancer survival Total no. (%)a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
CYP17 C > T -34 promoter (rs743572) TT CT CC CT/CC
153 218 73 291
(35) (49) (16) (65)
47 43 42 43
1.0 1.27 (0.98–1.65) 1.22 (0.86–1.73) 1.26 (0.98–1.61)
1.0 1.29 (0.99–1.69) 1.34 (0.94–1.92) 1.30 (1.02–1.68)
0.12 0.04
CYP3A4 A > G 5 0 NFSE (rs2740574) AA GA
369 (93) 27 (7)
45 54
1.0 0.63 (0.36–1.11)
1.0 0.77 (0.44–1.36)
0.67
CYP19UTR T > C 3 UTR (rs10046) TT CC TC
102 (23) 115 (26) 228 (51)
49 46 41
1.0 0.96 (0.68–1.35) 1.18 (0.88–1.58)
1.0 0.98 (0.68–1.39) 1.25 (0.91–1.68)
0.19
PR G > T V660L (rs1042838) GG TG/TT
321 (71) 130 (29)
44 44
1.0 1.11 (0.86–1.43)
1.0 1.10 (0.84–1.44)
0.47
PR G > A -331 promoter (rs10895068) GG GA
410 (92) 37 (8)
44 36
1.0 1.47 (1.00–2.16)
1.0 1.14 (0.77–1.69)
0.82
HsdA238V C > T A238V CC CT
443 (99) 6 (1)
44 50
1.0 0.90 (0.33–2.41)
1.0 0.85 (0.31–3.32)
0.74
HsdS313G A > G S313G (rs605059) AA GA GG
122 (27) 235 (52) 91 (21)
45 43 44
1.0 1.01 (0.76–1.32) 1.13 (0.82–1.58)
1.0 1.04 (0.78–1.38) 0.99 (0.71–1.38)
0.93
Hsd517B4 T > C W511R (rs17145454) TT TC
382 (85) 67 (15)
45 40
1.0 1.09 (0.80–1.50)
1.0 0.97 (0.70–1.34)
0.85
361 (81) 86 (19)
46 38
1.0 1.22 (0.92–1.61)
1.0 1.03 (0.77–1.39)
0.81
207 194 46 240
40 45 57 48
1.0 0.79 (0.62–1.01) 0.72 (0.47–1.09) 0.77 (0.61–0.98)
1.0 0.80 (0.62–1.03) 0.75 (0.48–1.15) 0.79 (0.62–1.01)
0.16 0.06
169 70
33 37
1.0 0.86 (0.62–1.20)
1.0 1.13 (0.80–1.60)
0.46
621 P22
57 182
30 35
1.0 0.93 (0.66–1.32)
1.0 0.87 (0.61–1.25)
0.45
AR Average allele size <22 P22
102 137
34 35
1.0 0.96 (0.71–1.30)
1.0 1.06 (0.77–1.45)
0.71
57
32
Gene polymorphism
Genotype
pc
0
0
Srd5A2 TA repeat 3 UTR TA0/TA0 TA0/TA9 and TA9/TA9 Srd5A2V89L G > C V89L (rs523349) GG GC CC GC/CC
(46) (43) (10) (54)
AR exon 1 CAG repeat, cutpoint P22 AR Smaller allele size 621 P22 AR Larger allele size
AR genotype number of alleles 0d
(continued on next page)
100
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Table 2 (continued) Gene polymorphism
a b c d e f
Genotype
Total no. (%)a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
1e 2f
112 70
34 37
0.98 (0.68–1.43) 0.86 (0.57–1.29)
0.81 (0.55–1.19) 0.98 (0.64–1.52)
pc
0.45
Column numbers (n) do not all sum to total because some data missing. Adjusted for age group, FIGO stage, histologic subgroup, histologic grade (1 vs. 2–4). p-value for v2 test for dominant model (2 levels) from v2 test for trend for genotypes analysed under a codominant model (3 levels). 0 = ARAL1 6 22 and ARAL2 < 22. 1 = ARAL1 P 22 or ARAL2 P 22. 2 = ARAL1 P 22 and ARAL2 P 22.
or CC genotypes defined by the SRD5A2 G > C V89L polymorphism (Fig. 2), with a 21% decreased risk of early death compared with the common homozygote (adjusted HR 0.79, 95% CI 0.62–1.01; p = 0.06). Among the DNA repair pathway polymorphisms, only the XRCC3 T241M variant showed statistically significant evidence for an association with outcome, but this was driven by the heterozygote genotype, with no overall association under a dominant model (adjusted HR 1.17, 95% CI 0.91–1.51; p = 0.20). For those variants with a non-significant elevated HR observed for the rare homozygote but not the heterozygote genotype (XPD D312N BRCA2 N372H, XRCC1 R399Gln), we also assessed risk under a recessive model. There was no convincing evidence that the rare homozygote genotype for any of these polymorphisms modified survival (p > 0.20).
4. Discussion We assessed the association between ovarian cancer survival and 21 polymorphisms in hormone and DNA repair pathway genes. Polymorphisms were selected as they are candidate risk factors for ovarian cancer predisposition, based on their existence in hormone and DNA repair pathway genes, and their predicted or known functional relevance. The CYP19 exon 10 3 0 UTR snp has been reported to be associated with increased mRNA levels [28] and is also significantly associated with serum estradiol levels [29]; the PGR V660L polymorphism has been reported to increase transcriptional activity [30]; the PGR -331 promoter region variant has been shown to have functional relevance [31] and to be associated with risk of endometrioid ovarian cancer [24]; the SRD5A2 V89L variant has been shown to exhibit decreased testosterone production in in vitro assays [32]; the AR CAG repeat length has been shown to modulate AR transactivation [33]. Evidence to support functional relevance for the DNA gene polymorphisms included in this study has been largely indirect, often based on small
sample sizes, and is very often conflicting. However in vitro assays assessing the effect of the variant directly have indicated altered function for some variants: the XRCC2 188H variant has been shown to slightly increase sensitivity to damage in cellular complementation assays [34], the XRCC1 280H variant has been associated with increased single-strand breaks in transfection assays [35]; and the antiproliferative activity of the RNA molecule encoded by the PHB gene is inactivated by the 3 0 UTR T allele polymorphism [36]. We found suggestive evidence that only two of the 21 polymorphisms might be independent predictors of ovarian cancer survival, the 5 0 UTR T > C CYP17 gene polymorphism and the G > C V89L SRD5A2 polymorphism. Our results showed that women with the CYP17 C allele had a 30% increase in death hazard rate compared to those with the common homozygote (T allele), a finding consistent with similar analysis of association with breast cancer survival in a large population-based study [37]. However, the direct biological relevance of the CYP17 polymorphism is questionable given that a recent large study found no evidence for altered hormone levels [29]. We also observed a modest, but statistically non-significant, inverse association between the SRD5A2 V89L rare C allele and survival, with a 21% reduction in risk of early death for women with the C allele. However, we would intuitively hypothesize that this variant would be associated with poor survival since this amino acid substitution variant has been reported to increase testosterone production, and given the observed associations between AR status, EGFR levels, and poor prognosis [14,15]. Indeed, our result is in contrast to that observed in relation to breast and prostate cancer, where a number of studies have reported the Leu allele is associated with a worse prognostic profile and shorter survival time [38–40]. Given the large number of polymorphisms tested, it is highly
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
101
Table 3 Association between DNA repair genes and ovarian cancer survival Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
XRCC2 G > A R188H (rs3218536) GG 362 (84) AA and GA 69 (16)
42 50
1.0 0.97 (0.70–1.34)
1.0 1.05 (0.75–1.49)
0.74
XRCC3 C > T T241M (rs861539) CC CT TT
196 (44) 191 (42) 61 (14)
46 39 53
1.0 1.23 (0.95–1.57) 0.93 (0.64–1.35)
1.0 1.34 (1.02–1.75) 0.76 (0.52–1.12)
<0.01
XPD312 G > A D312N (rs1799793) GG 200 (45) GA 180 (41) AA 63 (14)
42 48 47
1.0 0.81 (0.62–1.05) 0.80 (0.56–1.15)
1.0 0.98 (0.74–1.29) 1.15 (0.78–1.69)
0.73
RAD51 C > G -135 promoter (rs1801320) CC 379 (84) GG/GC 72 (16)
44 43
1.0 1.16 (0.84–1.57)
1.0 1.06 (0.76–1.47)
0.75
RAD52 T > G Y415X (rs4987208) TT TG
404 (97) 13 (3)
43 27
1.0 1.31 (0.65–2.66)
1.0 0.82 (0.41–1.69)
0.60
BRCA2 T > G N372H (rs144848) NN NH HH
221 (50) 174 (40) 42 (10)
43 46 36
1.0 0.84 (0.65–1.08) 1.09 (0.73–1.63)
1.0 0.91 (0.70–1.19) 1.10 (0.71–1.69)
0.68
PHB C > T 3 UTR (rs6917) CC TT TC
294 (66) 18 (4) 134 (30)
46 50 39
1.0 0.84 (0.45–1.54) 1.00 (0.77–1.29)
1.0 0.90 (0.47–1.71) 0.99 (0.75–1.31)
0.95
APE T > G D148E (rs3136820) TT TG GG
120 (30) 206 (51) 78 (19)
41 45 53
1.0 0.93 (0.71–1.23) 0.82 (0.56–1.18)
1.0 1.07 (0.79–1.44) 0.82 (0.55–1.22)
0.36
XRCC1 G > A R399Gln (rs25487) GG GA AA
196 (44) 192 (44) 51 (12)
41 50 36
1.0 0.80 (0.62–1.03) 1.17 (0.82–1.68)
1.0 0.84 (0.65–1.11) 1.14 (0.77–1.67)
0.25
XRCC1 C > T R194W (rs1799782) CC 391 (88) CT 49 (11) TT 6 (1)
46 37 33
1.0 1.23 (0.86–1.74) 1.08 (0.40–2.91)
1.71 (0.86–1.74) 0.68 (0.24–1.89)
0.71
Gene polymorphism
Genotype
Total no. (%)a
pc
0
a b c
Column numbers (n) do not sum to total because some data missing. Adjusted for FIGO stage, age group, histologic subgroup and grade. p trend for genotypes analysed under a codominant model.
likely that these marginally significant findings occurred by chance. It is interesting to note, however, that we have tentative evidence that both of these variants are also associated with increased predisposition to ovarian cancer in our Australian sample set (unpublished data, including an extension of our published work on CYP17 [20], and there is at least one published study supporting an association
between the CYP17 variant and increased risk of invasive ovarian cancer [41]). Independent replication is thus required to validate these findings. There is no convincing evidence that our results were confounded by known prognostic factors. In most instances there was no marked difference between crude and adjusted hazard ratios suggesting confounding was minimal. One exception was
102
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Cumulative survival
1.0 0.8 0.6 CYP17 TT
0.4
CYP17 CT CYP17 CC
0.2 0.0 0.0
5.0
10.0
15.0
20.0
Survival time (years)
Fig. 1. Kaplan–Meier survival curve for the CYP17 C > T -34 promoter (rs743572) polymorphism.
Cumulative survival
1.0 0.8 0.6 0.4
SRD5A2 V89L CC SRD5A2 V89L GC
0.2
SRD5A2 V89L GG
0.0 0.0
5.0
10.0
15.0
20.0
Survival time (years)
Fig. 2. Kaplan–Meier survival curve for the Srd5A2 G > C V89L (rs523349) polymorphism.
the rare PR -331 variant, which has been shown to cause increased expression of the PRB protein form of the progesterone receptor over the PRA form. While we observed a borderline significant 1.5-fold increased HR for the GA allele in the crude analysis, this diminished considerably with adjustment for clinicopathological factors, including histology. Of note, this variant has been shown to be associated with decreased risk of the endometrioid subtype of ovarian cancer in a pooled study which included samples in this study [24]. This is particularly interesting, given evidence from several small studies of ovarian tumors that PR form B is the predominant form expressed in advanced ovarian cancers of serous, endometrioid and clear cell subtype [42], and the only form expressed in metastases from ovarian cancers of these subtypes [43]. The RAD52 rare truncation variant also showed marked difference between crude and adjusted estimates, but given the rarity of this variant, and the imprecision of the estimates, this is likely to reflect chance rather than confounding.
The women in this analysis were unselected with regard to genotype and follow-up was essentially complete [44]. The baseline clinical measures were made independently of laboratory analyses, and are likely to have been accurate. Laboratory analyses were unbiased with respect to outcome and, although some non-differential misclassification (random error) will be inevitable, the likely effect of this would be to bias our results towards the null, thus the true effects might be greater than those seen here. Similarly ascertainment of outcome is thought to be almost complete and did not vary by genotype or other clinical factors. A potential limitation of the study is the possibility of selection bias introduced by cases who did not participate in the original case-control study due to illness or death (n = 69), but this is unlikely to have been an issue for incident cases recruited through the Royal Brisbane Hospital, since recruitment was at the time of clinical diagnosis with no additional requirements for participation such as collection of detailed questionnaire information. It is also likely that women who did not participate in the original case-control study would have been more ill and had worse survival than those who did take part, thus the overall survival proportions observed here might over-estimate survival among all women with ovarian cancer. This would not, however, have affected our results unless the association between CYP17 and SRD5A2 V89L genotypes and survival somehow differed between participants and non-participants. We have adjusted for the major clinical factors that affect survival but there is likely to be some residual confounding because of the imperfect nature of some of the clinical measures. For CYP17 the association became stronger after adjustment and, given the similarity of the crude and adjusted estimates for SRD5A2 V89L, it is unlikely that residual confounding could explain the observed associations. It is also impossible to completely exclude chance as an explanation. In conclusions, our results suggest that polymorphisms in genes in hormone and DNA repair pathways are not major determinants of survival among women with ovarian cancer. We found some evidence that the CYP17 and SRD5A2 V89L genetic polymorphisms may be associated with ovarian cancer outcome, however our preliminary results require confirmation by further association studies and in vitro investigations concerning the mecha-
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
nisms of the effect before we can conclude that this is real. Acknowledgements This study was supported in part by a grant from the Mayne Bequest Fund, University of Queensland. Initial funding to recruit the cases for the Survey of Women’s Health was obtained from the National Health & Medical Research Council (NHMRC) of Australia and Queensland Cancer Fund. P. Webb is funded by a Queensland Cancer Fund Senior Research Fellowship, A. Spurdle is funded by an NHMRC Career Development Award and G. Chenevix-Trench is an NHMRC Principal Research Fellow. References [1] Australian Institute of Health and Welfare (AIHW) and Australasian Association of Cancer Registries (AACR) Cancer in Australia 2001. AIHW cat. no. CAN 23. Canberra: AIHW, 2004. [2] A. Jemal, T. Murray, E. Ward, A. Samuels, R.C. Tiwari, A. Ghafoor, et al., Cancer statistics, 2005, CA Cancer J. Clin. 55 (2005) 10–30. [3] T. Bjorge, A. Engeland, K. Sundfor, C.G. Trope, Prognosis of 2800 patients with epithelial ovarian cancer diagnosed during 1975–94 and treated at the Norwegian Radium Hospital, Acta. Obstet. Gynecol. Scand. 77 (1998) 777–781. [4] S.A. Cannistra, Cancer of the ovary, N. Engl. J. Med. 351 (2004) 2519–2529. [5] A. Berchuck, E.S. Iversen, J.M. Lancaster, J. Pittman, J. Luo, P. Lee, et al., Patterns of gene expression that characterize long-term survival in advanced stage serous ovarian cancers, Clin. Cancer Res. 11 (2005) 3686–3696. [6] H.A. Risch, Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone, J. Natl. Cancer Inst. 90 (1998) 1774–1786. [7] F. Modugno, Ovarian cancer and polymorphisms in the androgen and progesterone receptor genes: a HuGE review, Am. J. Epidemiol. 159 (2004) 319–335. [8] R.E. Hempling, M.S. Piver, G.H. Eltabbakh, F.O. Recio, Progesterone receptor status is a significant prognostic variable of progression-free survival in advanced epithelial ovarian cancer, Am. J. Clin. Oncol. 21 (1998) 447–451. [9] B.R. Rao, B.J. Slotman, A.A. Geldof, W.N. Dinjens, Correlation between tumor histology, steroid receptor status, and adenosine deaminase complexing protein immunoreactivity in ovarian cancer, Int. J. Gynecol. Pathol. 9 (1990) 47–54. [10] M. Harding, S. Cowan, D. Hole, L. Cassidy, H. Kitchener, J. Davis, R. Leake, Estrogen and progesterone receptors in ovarian cancer, Cancer 65 (1990) 486–491. [11] K. Munstedt, J. Steen, A.G. Knauf, T. Buch, R. von Georgi, F.E. Franke, Steroid hormone receptors and long term survival in invasive ovarian cancer, Cancer 89 (2000) 1783–1791.
103
[12] P. Lindgren, T. Backstrom, C.G. Mahlck, M. Ridderheim, S. Cajander, Steroid receptors and hormones in relation to cell proliferation and apoptosis in poorly differentiated epithelial ovarian tumors, Int. J. Oncol. 19 (2001) 31–38. [13] S. Cunat, P. Hoffmann, P. Pujol, Estrogens and epithelial ovarian cancer, Gynecol. Oncol. 94 (2004) 25–32. [14] J.V. Ilekis, J.P. Connor, G.S. Prins, K. Ferrer, C. Niederberger, B. Scoccia, Expression of epidermal growth factor and androgen receptors in ovarian cancer, Gynecol. Oncol. 66 (1997) 250–254. [15] G. Scambia, P. Benedetti-Panici, G. Ferrandina, M. Distefano, G. Salerno, M.E. Romanini, et al., Epidermal growth factor, oestrogen and progesterone receptor expression in primary ovarian cancer: correlation with clinical outcome and response to chemotherapy, Br. J. Cancer 72 (1995) 361–366. [16] A. Richardson, S.B. Kaye, Drug resistance in ovarian cancer: the emerging importance of gene transcription and spatio-temporal regulation of resistance, Drug Resist. Updat. 8 (2005) 311–321. [17] K.V. Ferry, T.C. Hamilton, S.W. Johnson, Increased nucleotide excision repair in cisplatin-resistant ovarian cancer cells: role of ERCC1-XPF, Biochem. Pharmacol. 60 (2000) 1305–1313. [18] D. Purdie, A. Green, C. Bain, V. Siskind, B. Ward, N. Hacker, Reproductive and other factors and risk of epithelial ovarian cancer: an Australian case-control study. Survey of Women’s Health Study Group, Int. J. Cancer 62 (1995) 678–684. [19] A.B. Spurdle, B. Goodwin, E. Hodgson, J.L. Hopper, X. Chen, D. Purdie, et al., The CYP3A4*1B polymorphism has no functional significance and is not associated with risk of breast or ovarian cancer, Pharmacogenetics 12 (2002) 355–366. [20] A.B. Spurdle, X.Q. Chen, M. Abbazadegan, N. Martin, S.K. Khoo, T. Hurst, B. Ward, CYP17 promotor polymorphism and ovarian cancer risk, Int. J. Cancer 86 (2000) 436–439. [21] A.B. Spurdle, J.L. Hopper, X. Chen, G.S. Dite, M.R. McCredie, G.G. Giles, et al., The steroid 5alpha-reductase type II TA repeat polymorphism is not associated with risk of breast or ovarian cancer in Australian women, Cancer Epidemiol. Biomarkers Prev. 10 (2001) 1287–1293. [22] A.B. Spurdle, P.M. Webb, X. Chen, N. Martin, G.G. Giles, J.L. Hopper, et al., Androgen receptor exon 1 CAG repeat length and risk of ovarian cancer, Int. J. Cancer 87 (2000) 637–643. [23] A.B. Spurdle, P.M. Webb, D.M. Purdie, X. Chen, A. Green, G. Chenevix-Trench, No significant association between progesterone receptor exon 4 Val660Leu G/T polymorphism and risk of ovarian cancer, Carcinogenesis 22 (2001) 717–721. [24] A. Berchuck, J.M. Schildkraut, R.M. Wenham, B. Calingaert, S. Ali, A. Henriott, et al., Progesterone receptor promoter +331A polymorphism is associated with a reduced risk of endometrioid and clear cell ovarian cancers Cancer Epidemiol, Biomarkers Prev. 13 (2004) 2141–2147. [25] P.M. Webb, J.L. Hopper, B. Newman, X. Chen, L. Kelemen, G.G. Giles, et al., Double-strand break repair gene polymorphisms and risk of breast or ovarian cancer, Cancer Epidemiol. Biomarkers Prev. 14 (2005) 319–323. [26] L. Kelemen, A.B. Spurdle, D.M. Purdie, D. Gertig, G. Chenevix-Trench, RAD52 Y415X truncation polymorphism
104
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104 and epithelial ovarian cancer risk in Australian women, Cancer Lett. 218 (2005) 191–197. A. Auranen, A.B. Spurdle, X. Chen, J. Lipscombe, D.M. Purdie, J.L. Hopper, et al., BRCA2 Arg372His polymorphism and epithelial ovarian cancer risk, Int. J. Cancer 103 (2003) 427–430. V.N. Kristensen, N. Harada, N. Yoshimura, E. Haraldsen, P.E. Lonning, B. Erikstein, et al., Genetic variants of CYP19 (aromatase) and breast cancer risk, Oncogene 19 (2000) 1329–1333. A.M. Dunning, M. Dowsett, C.S. Healey, L. Tee, R.N. Luben, E. Folkerd, et al., Polymorphisms associated with circulating sex hormone levels in postmenopausal women, J. Natl. Cancer Inst. 96 (2004) 936–945. I.U. Agoulnik, Z.W. Tong, D.C. Fischer, K. Korner, N.E. Atkinson, D.P. Edwards, et al., A germline variation in the progesterone receptor gene increases transcriptional activity and may modify ovarian cancer risk, J. Clin. Endocrinol. Metab. 89 (2004) 6340–6347. I. De Vivo, G.S. Huggins, S.E. Hankinson, P.J. Lescault, M. Boezen, G.A. Colditz, et al., A functional polymorphism in the promoter of the progesterone receptor gene associated with endometrial cancer risk, Proc. Natl. Acad. Sci. 99 (2002) 12263–12268. N.M. Makridakis, E. di Salle, J.K. Reichardt, Biochemical and pharmacogenetic dissection of human steroid 5 alphareductase type II, Pharmacogenetics 10 (2000) 407–413. N.L. Chamberlain, E.D. Driver, R.L. Miesfeld, The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function, Nucleic Acids Res. 22 (1994) 3181–3186. S. Rafii, P. O’Regan, G. Xinarianos, I. Azmy, T. Stephenson, M. Reed, et al., A potential role for the XRCC2 R188H polymorphic site in DNA-damage repair and breast cancer, Hum. Mol. Genet. 11 (2002) 1433–1438. T. Takanami, J. Nakamura, Y. Kubota, S. Horiuchi, The Arg280His polymorphism in X-ray repair cross-comple-
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
menting gene 1 impairs DNA repair ability, Mutat. Res. 582 (2005) 135–145. E.R. Jupe, A.A. Badgett, B.R. Neas, M.A. Craft, D.S. Mitchell, R. Resta, et al., Single nucleotide polymorphism in prohibition 39 untranslated region and breast-cancer susceptibility, Lancet 357 (2001) 1588–1589. E.L. Goode, A.M. Dunning, B. Kuschel, C.S. Healey, N.E. Day, B.A.J. Ponder, et al., Effect of germ-line genetic variation on breast cancer survival in a population-based study, Cancer Res. 62 (2002) 3052–3057. C.H. van Gils, N.C. Onland-Moret, M. Roest, P.A.H. van Noord, P.H.M. Peeters, The V89L polymorphism in the 5alpha-reductase type 2 gene and risk of breast cancer, Cancer Epidemiol. Biomarkers Prev. 12 (2003) 1194–1199. A. Scorilas, B. Bharaj, M. Giai, E.P. Diamandis, Codon 89 polymorphism in the human 5 alpha-reductase gene in primary breast cancer, Br. J. Cancer 84 (2001) 760–767. A. Shibata, M.I. Garcia, I. Cheng, T.A. Stamey, J.E. McNeal, J.D. Brooks, et al., Polymorphisms in the androgen receptor and type II 5 alpha-reductase genes and prostate cancer prognosis, Prostate 52 (2002) 269–278. E.I. Garner, E.E. Stokes, R.S. Berkowitz, S.C. Mok, D.W. Cramer, Polymorphisms of the estrogen-metabolizing genes CYP17 and catechol-O-methyltransferase and risk of epithelial ovarian cancer, Cancer Res. 62 (2002) 3058–3062. J. Fujimoto, S. Ichigo, M. Hori, M. Nishigaki, T. Tamaya, Expression of progesterone receptor form A and B mRNAs in gynecologic malignant tumors, Tumour Biol. 16 (1995) 254–260. J. Fujimoto, S. Ichigo, R. Hirose, H. Sakaguchi, T. Tamaya, Clinical implication of expression of progesterone receptor form A and B mRNAs in secondary spreading of gynecologic cancers, J. Steroid Biochem. Mol. Biol. 62 (1997) 449– 454. C. Nagle, P. Webb, D. Purdie, A. Green, C. Bain, Searching for cancer deaths on Australia: NDI vs. cancer registries, Asian Pac. J. Cancer Prev. 7 (2006) 41–46.
Ovarian cancer survival and polymorphisms in hormone and DNA repair pathway genes Christina M. Nagle a, Georgia Chenevix-Trench b, Penelope M. Webb a, Amanda B. Spurdle c,* a
c
Cancer and Population Studies, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia b Cancer Genetics, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia Molecular Cancer Epidemiology, Queensland Institute of Medical Research, PO Royal Brisbane Hospital, Queensland 4029, Australia Received 14 September 2006; accepted 6 November 2006
Abstract We evaluated the association between 21 polymorphisms in hormone and DNA repair pathway genes and survival among 454 Australian women diagnosed with invasive epithelial ovarian cancer. The cohort was followed for mortality using personal identifiers which were linked to state cancer registry records and the Australian National Death Index. The mean follow-up time after ovarian cancer diagnosis was 4.63 years (all women) and 8.07 years for the censored group (those alive or dead from non-ovarian cancer causes). Two hundred and eighty-eight (63%) ovarian cancer deaths occurred during the follow-up period. No association was observed for the vast majority of polymorphisms, but there was suggestive evidence for altered risk of ovarian cancer death associated with the CYP17 5 0 UTR C allele (HR 1.30; 95% CI = 1.02–1.68, p = 0.04), and for the SRD5A2 V89L C allele (HR 0.79; 95% CI = 0.62–1.01, p = 0.06). These results are interesting given tentative evidence that both of these variants are also associated with increased predisposition to ovarian cancer in our extended Australian study, and in other published studies. However, given the marginal significance of these associations and the large number of tests performed, independent replication will be necessary to validate these novel findings. Ó 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: DNA repair genes; Epithelial ovarian cancer; Hormone pathway genes; Polymorphisms; Survival
1. Introduction Ovarian cancer is a significant cause of morbidity and mortality in Australia and worldwide [1–3]. It is estimated that 1 in 104 Australian women will
* Corresponding author. Tel.: +61 7 3362 0371; fax +61 7 3362 0105. E-mail address: [email protected] (A.B. Spurdle).
develop ovarian cancer in their lifetime, and over half of women with the disease are expected to die within 5 years of diagnosis [1]. Most women present with advanced disease at diagnosis and despite high rates of clinical response to first-line treatment, many women experience relapse due to sub-clinical disease that persists after completion of initial treatment [4]. This variability of response is thought to be related, in part, to individual genetic variation [5].
0304-3835/$ - see front matter Ó 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2006.11.011
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Variants in hormone pathway genes are good candidates for ovarian cancer risk [6]. Several epidemiological studies suggest a role for polymorphisms in hormone genes in ovarian cancer risk [7]. Support for a role of hormones in ovarian tumour progression is also growing. There is some evidence from the literature that tumour progesterone receptor (PR) positivity is an independent favourable predictor of progression-free [8] or total survival [9,10] particularly when the estrogen receptor (ER) is not expressed [11]. In addition, improved survival rate has been reported for patients with high serum progesterone, especially in combination with expression of PR [12]. Estrogen has been shown to stimulate growth of ovarian tumour cell lines expressing ER, and numerous estrogen-induced proteins have been shown to be upregulated in epithelial ovarian cancer [13]. AR levels have been associated with epidermal growth factor receptor (EGFR) levels, and overexpression of EGFR is in turn associated with poor prognosis [14,15]. Similarly, there is much evidence from known familial predisposition genes to suggest that DNA repair genes are involved in ovarian cancer risk, and accumulating evidence that alterations in DNA damage recognition and DNA repair may also play a role in outcome. Gene expression studies indicate that gene families involved in regulating DNA damage play a role in development of drug resistance, with upregulation of repair genes invariably associated with chemoresistance [16,17]. We have extended our ovarian cancer case-control candidate gene association analyses to test the hypothesis that polymorphisms in hormone and DNA repair pathway genes are associated with survival after a diagnosis of ovarian cancer. We assessed association between survival and 21 polymorphisms in hormone pathway and DNA repair pathway genes, for a large group of Australian women previously diagnosed with ovarian cancer. 2. Materials and methods 2.1. Study population We included 454 women diagnosed with primary invasive epithelial ovarian cancer between 1985 and 1997. Two thirds of the women (n = 296 [65%]) had participated in a large population-based case-control study of the etiology of ovarian cancer, the results of which have been published previously [18]. The remaining women (n = 158 [35%]) were ascertained as incident cases from the Royal Brisbane Hospital, Queensland, Australia [19]. The study was approved by the Institutional ethics review board of
97
the University of Queensland and all participating hospitals where the women were originally diagnosed and treated. 2.2. Clinical and pathologic data Clinical details including operative and histologic data were obtained retrospectively from the women’s medical records and pathology reports. The following clinical and pathologic data were collected: International Federation of Gynecology and Obstetrics (FIGO) stage, tumor histology (subtype), grade (well, moderate, poor, undifferentiated) and treatment details. Three quarters of the women (340/454) received platinum based chemotherapy as first-line treatment. The women were followed up for survival from the date of diagnosis until the date of death (from ovarian cancer) or censor date using personal identifiers which were linked to the Australian National Death Index (NDI) and state cancer registry records. Both the NDI and cancer registries used probabilistic record linkage software to match the women to their databases. To identify all ovarian cancer related deaths we obtained full details, including immediate and underlying cause/s of death and contributing factors, directly from death certificates for those decedents that had participated in the case-control study (n = 179), or indirectly through linkage to death records held at the Queensland cancer registry for those decedents recruited from the Royal Brisbane Hospital (n = 109). 2.3. Genotype detection DNA preparation and genotyping was as described previously [19]. Briefly, germline DNA was extracted from peripheral blood (collected at diagnosis) by a salt-precipitation method for women recruited through the Royal Brisbane Hospital and from uninvolved tissue from archival paraffin blocks, made at the time of diagnosis, for women ascertained through the population-based casecontrol study. Genotyping was performed in the course of ongoing analysis of candidate low penetrance ovarian cancer susceptibility genes [19–27]. ABI Taqman methodology was used for single nucleotide polymorphisms (all samples), and fluorescent PCR–PAGE methodology for the AR CAG repeat (peripheral blood DNA only (n = 239), due to archival DNA quality limitations). Additional assay details are available on request. Polymorphisms are described in Tables 2 and 3. 2.4. Statistical analysis The Kaplan–Meier technique was used to plot crude survival curves and estimate crude survival probabilities, and adjusted hazard ratios (HRs) and 95% confidence
98
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
intervals (95% CIs) were obtained from Cox regression models. The p-value for linear trend was calculated by the change in likelihood ratio statistic for entry of a linear term in the model, and thus was a v2 test on 1 degree of freedom. Analyses were adjusted for age at diagnosis (10-year age groups), FIGO stage, histologic subtype and histologic grade (moderate, poor and undifferentiated tumors were combined into a single group because survival did not vary for these three sub-groups but was significantly different from that for women with well differentiated tumors). All statistical analyses were performed using the Statistical Packages for Social Sciences for Windows, version 13.0 (SPSS Inc., Chicago, IL).
3. Results The characteristics of the study participants are shown in Table 1. Age at diagnosis ranged from 19 to 87 years, and the average age was 59 years. The most common subtype was serous (n = 274), followed by endometrioid (n = 62), clear cell (n = 31), mucinous (n = 29) and 58 ‘other’ histology which included mostly mixed mullerian and undifferentiated tumors. Patients were staged at laparotomy according to FIGO criteria. Of the 446 women
with known stage at diagnosis there were 83, 44, 278, and 41 with FIGO stages I, II, III, and IV, respectively. The group included 56 women with well differentiated tumors, 128 with moderately differentiated tumors, and 241 patients with poorly differentiated and undifferentiated tumors. The mean follow-up time after ovarian cancer diagnosis was 4.63 years (all women) and 8.07 years for the censored group (those alive or dead from another cause). Two hundred and eighty-eight (63%) ovarian cancer deaths occurred during the follow-up period, and the crude 5-year survival rate was 44%. Late FIGO stage, older age at diagnosis, histologic subtype and higher tumor grade were associated with survival and were included in all subsequent models. The majority of polymorphisms were not significantly associated with altered survival (Tables 2 and 3). Among the hormone pathway polymorphisms (Table 2), there was suggestive evidence for an association of similar magnitude with the TC and CC genotypes defined by the CYP17 5 0 UTR T > C polymorphism (Fig. 1), and carriers of the C allele showed a 30% increase in death hazard rate after adjustment for stage, age, grade and histology (HR 1.30, 95% CI 1.02–1.68; p = 0.04). We also found evidence of a modest, borderline significant, protective effect against early death among women who carried the GC
Table 1 Association between clinical and pathological factors and ovarian cancer survival
Age group <50 years 50–59 years 60–69 years 70+
Total no.a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
106 123 131 94
57 49 38 29
1.0 1.37 (0.97–1.95) 1.84 (1.31–2.59) 2.30 (1.60–3.30)
1.0 1.09 (0.74–1.62) 1.34 (0.91–1.97) 1.57 (1.04–2.36)
pc
0.16 FIGO stage I II III IV
83 44 278 41
86 76 31 11
1.0 2.43 (1.12–5.26) 10.32 (5.76–18.51) 19.17 (9.96–36.90)
1.0 1.92 (0.82–4.47) 6.99 (3.49–13.98) 11.79 (5.37–25.86) <0.01
Histological subgroup Serous Mucinous Endometrioid Clear cell Other
274 29 62 31 58
35 78 67 57 42
1.0 0.25 0.32 0.51 0.77
(0.12–0.51) (0.20–0.49) (0.29–0.87) (0.54–1.10)
1.0 1.02 0.56 1.88 0.78
(0.52–2.34) (0.33–0.87) (1.04–3.40) (0.52–1.15) 0.01
Histologic grade Well differentiated Moderately differentiated Poorly and undifferentiated
56 128 241
74 49 35
1.0 2.54 (1.49–4.31) 3.86 (2.34–6.37)
1.0 1.80 (1.01–3.19) 1.96 (1.12–3.44) 0.08
a b c
Column numbers (n) do not all sum to total because some data missing. Adjusted for age group, FIGO stage, histologic subgroup, histologic grade (1 vs. 2–4). p-value for heterogeneity for histological subgroup, p trend for all other comparisons.
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
99
Table 2 Association between hormone-metabolising genes and ovarian cancer survival Total no. (%)a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
CYP17 C > T -34 promoter (rs743572) TT CT CC CT/CC
153 218 73 291
(35) (49) (16) (65)
47 43 42 43
1.0 1.27 (0.98–1.65) 1.22 (0.86–1.73) 1.26 (0.98–1.61)
1.0 1.29 (0.99–1.69) 1.34 (0.94–1.92) 1.30 (1.02–1.68)
0.12 0.04
CYP3A4 A > G 5 0 NFSE (rs2740574) AA GA
369 (93) 27 (7)
45 54
1.0 0.63 (0.36–1.11)
1.0 0.77 (0.44–1.36)
0.67
CYP19UTR T > C 3 UTR (rs10046) TT CC TC
102 (23) 115 (26) 228 (51)
49 46 41
1.0 0.96 (0.68–1.35) 1.18 (0.88–1.58)
1.0 0.98 (0.68–1.39) 1.25 (0.91–1.68)
0.19
PR G > T V660L (rs1042838) GG TG/TT
321 (71) 130 (29)
44 44
1.0 1.11 (0.86–1.43)
1.0 1.10 (0.84–1.44)
0.47
PR G > A -331 promoter (rs10895068) GG GA
410 (92) 37 (8)
44 36
1.0 1.47 (1.00–2.16)
1.0 1.14 (0.77–1.69)
0.82
HsdA238V C > T A238V CC CT
443 (99) 6 (1)
44 50
1.0 0.90 (0.33–2.41)
1.0 0.85 (0.31–3.32)
0.74
HsdS313G A > G S313G (rs605059) AA GA GG
122 (27) 235 (52) 91 (21)
45 43 44
1.0 1.01 (0.76–1.32) 1.13 (0.82–1.58)
1.0 1.04 (0.78–1.38) 0.99 (0.71–1.38)
0.93
Hsd517B4 T > C W511R (rs17145454) TT TC
382 (85) 67 (15)
45 40
1.0 1.09 (0.80–1.50)
1.0 0.97 (0.70–1.34)
0.85
361 (81) 86 (19)
46 38
1.0 1.22 (0.92–1.61)
1.0 1.03 (0.77–1.39)
0.81
207 194 46 240
40 45 57 48
1.0 0.79 (0.62–1.01) 0.72 (0.47–1.09) 0.77 (0.61–0.98)
1.0 0.80 (0.62–1.03) 0.75 (0.48–1.15) 0.79 (0.62–1.01)
0.16 0.06
169 70
33 37
1.0 0.86 (0.62–1.20)
1.0 1.13 (0.80–1.60)
0.46
621 P22
57 182
30 35
1.0 0.93 (0.66–1.32)
1.0 0.87 (0.61–1.25)
0.45
AR Average allele size <22 P22
102 137
34 35
1.0 0.96 (0.71–1.30)
1.0 1.06 (0.77–1.45)
0.71
57
32
Gene polymorphism
Genotype
pc
0
0
Srd5A2 TA repeat 3 UTR TA0/TA0 TA0/TA9 and TA9/TA9 Srd5A2V89L G > C V89L (rs523349) GG GC CC GC/CC
(46) (43) (10) (54)
AR exon 1 CAG repeat, cutpoint P22 AR Smaller allele size 621 P22 AR Larger allele size
AR genotype number of alleles 0d
(continued on next page)
100
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Table 2 (continued) Gene polymorphism
a b c d e f
Genotype
Total no. (%)a
Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
1e 2f
112 70
34 37
0.98 (0.68–1.43) 0.86 (0.57–1.29)
0.81 (0.55–1.19) 0.98 (0.64–1.52)
pc
0.45
Column numbers (n) do not all sum to total because some data missing. Adjusted for age group, FIGO stage, histologic subgroup, histologic grade (1 vs. 2–4). p-value for v2 test for dominant model (2 levels) from v2 test for trend for genotypes analysed under a codominant model (3 levels). 0 = ARAL1 6 22 and ARAL2 < 22. 1 = ARAL1 P 22 or ARAL2 P 22. 2 = ARAL1 P 22 and ARAL2 P 22.
or CC genotypes defined by the SRD5A2 G > C V89L polymorphism (Fig. 2), with a 21% decreased risk of early death compared with the common homozygote (adjusted HR 0.79, 95% CI 0.62–1.01; p = 0.06). Among the DNA repair pathway polymorphisms, only the XRCC3 T241M variant showed statistically significant evidence for an association with outcome, but this was driven by the heterozygote genotype, with no overall association under a dominant model (adjusted HR 1.17, 95% CI 0.91–1.51; p = 0.20). For those variants with a non-significant elevated HR observed for the rare homozygote but not the heterozygote genotype (XPD D312N BRCA2 N372H, XRCC1 R399Gln), we also assessed risk under a recessive model. There was no convincing evidence that the rare homozygote genotype for any of these polymorphisms modified survival (p > 0.20).
4. Discussion We assessed the association between ovarian cancer survival and 21 polymorphisms in hormone and DNA repair pathway genes. Polymorphisms were selected as they are candidate risk factors for ovarian cancer predisposition, based on their existence in hormone and DNA repair pathway genes, and their predicted or known functional relevance. The CYP19 exon 10 3 0 UTR snp has been reported to be associated with increased mRNA levels [28] and is also significantly associated with serum estradiol levels [29]; the PGR V660L polymorphism has been reported to increase transcriptional activity [30]; the PGR -331 promoter region variant has been shown to have functional relevance [31] and to be associated with risk of endometrioid ovarian cancer [24]; the SRD5A2 V89L variant has been shown to exhibit decreased testosterone production in in vitro assays [32]; the AR CAG repeat length has been shown to modulate AR transactivation [33]. Evidence to support functional relevance for the DNA gene polymorphisms included in this study has been largely indirect, often based on small
sample sizes, and is very often conflicting. However in vitro assays assessing the effect of the variant directly have indicated altered function for some variants: the XRCC2 188H variant has been shown to slightly increase sensitivity to damage in cellular complementation assays [34], the XRCC1 280H variant has been associated with increased single-strand breaks in transfection assays [35]; and the antiproliferative activity of the RNA molecule encoded by the PHB gene is inactivated by the 3 0 UTR T allele polymorphism [36]. We found suggestive evidence that only two of the 21 polymorphisms might be independent predictors of ovarian cancer survival, the 5 0 UTR T > C CYP17 gene polymorphism and the G > C V89L SRD5A2 polymorphism. Our results showed that women with the CYP17 C allele had a 30% increase in death hazard rate compared to those with the common homozygote (T allele), a finding consistent with similar analysis of association with breast cancer survival in a large population-based study [37]. However, the direct biological relevance of the CYP17 polymorphism is questionable given that a recent large study found no evidence for altered hormone levels [29]. We also observed a modest, but statistically non-significant, inverse association between the SRD5A2 V89L rare C allele and survival, with a 21% reduction in risk of early death for women with the C allele. However, we would intuitively hypothesize that this variant would be associated with poor survival since this amino acid substitution variant has been reported to increase testosterone production, and given the observed associations between AR status, EGFR levels, and poor prognosis [14,15]. Indeed, our result is in contrast to that observed in relation to breast and prostate cancer, where a number of studies have reported the Leu allele is associated with a worse prognostic profile and shorter survival time [38–40]. Given the large number of polymorphisms tested, it is highly
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
101
Table 3 Association between DNA repair genes and ovarian cancer survival Crude 5-year survival (%)
Crude HR (95% CI)
Adjustedb HR (95% CI)
XRCC2 G > A R188H (rs3218536) GG 362 (84) AA and GA 69 (16)
42 50
1.0 0.97 (0.70–1.34)
1.0 1.05 (0.75–1.49)
0.74
XRCC3 C > T T241M (rs861539) CC CT TT
196 (44) 191 (42) 61 (14)
46 39 53
1.0 1.23 (0.95–1.57) 0.93 (0.64–1.35)
1.0 1.34 (1.02–1.75) 0.76 (0.52–1.12)
<0.01
XPD312 G > A D312N (rs1799793) GG 200 (45) GA 180 (41) AA 63 (14)
42 48 47
1.0 0.81 (0.62–1.05) 0.80 (0.56–1.15)
1.0 0.98 (0.74–1.29) 1.15 (0.78–1.69)
0.73
RAD51 C > G -135 promoter (rs1801320) CC 379 (84) GG/GC 72 (16)
44 43
1.0 1.16 (0.84–1.57)
1.0 1.06 (0.76–1.47)
0.75
RAD52 T > G Y415X (rs4987208) TT TG
404 (97) 13 (3)
43 27
1.0 1.31 (0.65–2.66)
1.0 0.82 (0.41–1.69)
0.60
BRCA2 T > G N372H (rs144848) NN NH HH
221 (50) 174 (40) 42 (10)
43 46 36
1.0 0.84 (0.65–1.08) 1.09 (0.73–1.63)
1.0 0.91 (0.70–1.19) 1.10 (0.71–1.69)
0.68
PHB C > T 3 UTR (rs6917) CC TT TC
294 (66) 18 (4) 134 (30)
46 50 39
1.0 0.84 (0.45–1.54) 1.00 (0.77–1.29)
1.0 0.90 (0.47–1.71) 0.99 (0.75–1.31)
0.95
APE T > G D148E (rs3136820) TT TG GG
120 (30) 206 (51) 78 (19)
41 45 53
1.0 0.93 (0.71–1.23) 0.82 (0.56–1.18)
1.0 1.07 (0.79–1.44) 0.82 (0.55–1.22)
0.36
XRCC1 G > A R399Gln (rs25487) GG GA AA
196 (44) 192 (44) 51 (12)
41 50 36
1.0 0.80 (0.62–1.03) 1.17 (0.82–1.68)
1.0 0.84 (0.65–1.11) 1.14 (0.77–1.67)
0.25
XRCC1 C > T R194W (rs1799782) CC 391 (88) CT 49 (11) TT 6 (1)
46 37 33
1.0 1.23 (0.86–1.74) 1.08 (0.40–2.91)
1.71 (0.86–1.74) 0.68 (0.24–1.89)
0.71
Gene polymorphism
Genotype
Total no. (%)a
pc
0
a b c
Column numbers (n) do not sum to total because some data missing. Adjusted for FIGO stage, age group, histologic subgroup and grade. p trend for genotypes analysed under a codominant model.
likely that these marginally significant findings occurred by chance. It is interesting to note, however, that we have tentative evidence that both of these variants are also associated with increased predisposition to ovarian cancer in our Australian sample set (unpublished data, including an extension of our published work on CYP17 [20], and there is at least one published study supporting an association
between the CYP17 variant and increased risk of invasive ovarian cancer [41]). Independent replication is thus required to validate these findings. There is no convincing evidence that our results were confounded by known prognostic factors. In most instances there was no marked difference between crude and adjusted hazard ratios suggesting confounding was minimal. One exception was
102
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
Cumulative survival
1.0 0.8 0.6 CYP17 TT
0.4
CYP17 CT CYP17 CC
0.2 0.0 0.0
5.0
10.0
15.0
20.0
Survival time (years)
Fig. 1. Kaplan–Meier survival curve for the CYP17 C > T -34 promoter (rs743572) polymorphism.
Cumulative survival
1.0 0.8 0.6 0.4
SRD5A2 V89L CC SRD5A2 V89L GC
0.2
SRD5A2 V89L GG
0.0 0.0
5.0
10.0
15.0
20.0
Survival time (years)
Fig. 2. Kaplan–Meier survival curve for the Srd5A2 G > C V89L (rs523349) polymorphism.
the rare PR -331 variant, which has been shown to cause increased expression of the PRB protein form of the progesterone receptor over the PRA form. While we observed a borderline significant 1.5-fold increased HR for the GA allele in the crude analysis, this diminished considerably with adjustment for clinicopathological factors, including histology. Of note, this variant has been shown to be associated with decreased risk of the endometrioid subtype of ovarian cancer in a pooled study which included samples in this study [24]. This is particularly interesting, given evidence from several small studies of ovarian tumors that PR form B is the predominant form expressed in advanced ovarian cancers of serous, endometrioid and clear cell subtype [42], and the only form expressed in metastases from ovarian cancers of these subtypes [43]. The RAD52 rare truncation variant also showed marked difference between crude and adjusted estimates, but given the rarity of this variant, and the imprecision of the estimates, this is likely to reflect chance rather than confounding.
The women in this analysis were unselected with regard to genotype and follow-up was essentially complete [44]. The baseline clinical measures were made independently of laboratory analyses, and are likely to have been accurate. Laboratory analyses were unbiased with respect to outcome and, although some non-differential misclassification (random error) will be inevitable, the likely effect of this would be to bias our results towards the null, thus the true effects might be greater than those seen here. Similarly ascertainment of outcome is thought to be almost complete and did not vary by genotype or other clinical factors. A potential limitation of the study is the possibility of selection bias introduced by cases who did not participate in the original case-control study due to illness or death (n = 69), but this is unlikely to have been an issue for incident cases recruited through the Royal Brisbane Hospital, since recruitment was at the time of clinical diagnosis with no additional requirements for participation such as collection of detailed questionnaire information. It is also likely that women who did not participate in the original case-control study would have been more ill and had worse survival than those who did take part, thus the overall survival proportions observed here might over-estimate survival among all women with ovarian cancer. This would not, however, have affected our results unless the association between CYP17 and SRD5A2 V89L genotypes and survival somehow differed between participants and non-participants. We have adjusted for the major clinical factors that affect survival but there is likely to be some residual confounding because of the imperfect nature of some of the clinical measures. For CYP17 the association became stronger after adjustment and, given the similarity of the crude and adjusted estimates for SRD5A2 V89L, it is unlikely that residual confounding could explain the observed associations. It is also impossible to completely exclude chance as an explanation. In conclusions, our results suggest that polymorphisms in genes in hormone and DNA repair pathways are not major determinants of survival among women with ovarian cancer. We found some evidence that the CYP17 and SRD5A2 V89L genetic polymorphisms may be associated with ovarian cancer outcome, however our preliminary results require confirmation by further association studies and in vitro investigations concerning the mecha-
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104
nisms of the effect before we can conclude that this is real. Acknowledgements This study was supported in part by a grant from the Mayne Bequest Fund, University of Queensland. Initial funding to recruit the cases for the Survey of Women’s Health was obtained from the National Health & Medical Research Council (NHMRC) of Australia and Queensland Cancer Fund. P. Webb is funded by a Queensland Cancer Fund Senior Research Fellowship, A. Spurdle is funded by an NHMRC Career Development Award and G. Chenevix-Trench is an NHMRC Principal Research Fellow. References [1] Australian Institute of Health and Welfare (AIHW) and Australasian Association of Cancer Registries (AACR) Cancer in Australia 2001. AIHW cat. no. CAN 23. Canberra: AIHW, 2004. [2] A. Jemal, T. Murray, E. Ward, A. Samuels, R.C. Tiwari, A. Ghafoor, et al., Cancer statistics, 2005, CA Cancer J. Clin. 55 (2005) 10–30. [3] T. Bjorge, A. Engeland, K. Sundfor, C.G. Trope, Prognosis of 2800 patients with epithelial ovarian cancer diagnosed during 1975–94 and treated at the Norwegian Radium Hospital, Acta. Obstet. Gynecol. Scand. 77 (1998) 777–781. [4] S.A. Cannistra, Cancer of the ovary, N. Engl. J. Med. 351 (2004) 2519–2529. [5] A. Berchuck, E.S. Iversen, J.M. Lancaster, J. Pittman, J. Luo, P. Lee, et al., Patterns of gene expression that characterize long-term survival in advanced stage serous ovarian cancers, Clin. Cancer Res. 11 (2005) 3686–3696. [6] H.A. Risch, Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone, J. Natl. Cancer Inst. 90 (1998) 1774–1786. [7] F. Modugno, Ovarian cancer and polymorphisms in the androgen and progesterone receptor genes: a HuGE review, Am. J. Epidemiol. 159 (2004) 319–335. [8] R.E. Hempling, M.S. Piver, G.H. Eltabbakh, F.O. Recio, Progesterone receptor status is a significant prognostic variable of progression-free survival in advanced epithelial ovarian cancer, Am. J. Clin. Oncol. 21 (1998) 447–451. [9] B.R. Rao, B.J. Slotman, A.A. Geldof, W.N. Dinjens, Correlation between tumor histology, steroid receptor status, and adenosine deaminase complexing protein immunoreactivity in ovarian cancer, Int. J. Gynecol. Pathol. 9 (1990) 47–54. [10] M. Harding, S. Cowan, D. Hole, L. Cassidy, H. Kitchener, J. Davis, R. Leake, Estrogen and progesterone receptors in ovarian cancer, Cancer 65 (1990) 486–491. [11] K. Munstedt, J. Steen, A.G. Knauf, T. Buch, R. von Georgi, F.E. Franke, Steroid hormone receptors and long term survival in invasive ovarian cancer, Cancer 89 (2000) 1783–1791.
103
[12] P. Lindgren, T. Backstrom, C.G. Mahlck, M. Ridderheim, S. Cajander, Steroid receptors and hormones in relation to cell proliferation and apoptosis in poorly differentiated epithelial ovarian tumors, Int. J. Oncol. 19 (2001) 31–38. [13] S. Cunat, P. Hoffmann, P. Pujol, Estrogens and epithelial ovarian cancer, Gynecol. Oncol. 94 (2004) 25–32. [14] J.V. Ilekis, J.P. Connor, G.S. Prins, K. Ferrer, C. Niederberger, B. Scoccia, Expression of epidermal growth factor and androgen receptors in ovarian cancer, Gynecol. Oncol. 66 (1997) 250–254. [15] G. Scambia, P. Benedetti-Panici, G. Ferrandina, M. Distefano, G. Salerno, M.E. Romanini, et al., Epidermal growth factor, oestrogen and progesterone receptor expression in primary ovarian cancer: correlation with clinical outcome and response to chemotherapy, Br. J. Cancer 72 (1995) 361–366. [16] A. Richardson, S.B. Kaye, Drug resistance in ovarian cancer: the emerging importance of gene transcription and spatio-temporal regulation of resistance, Drug Resist. Updat. 8 (2005) 311–321. [17] K.V. Ferry, T.C. Hamilton, S.W. Johnson, Increased nucleotide excision repair in cisplatin-resistant ovarian cancer cells: role of ERCC1-XPF, Biochem. Pharmacol. 60 (2000) 1305–1313. [18] D. Purdie, A. Green, C. Bain, V. Siskind, B. Ward, N. Hacker, Reproductive and other factors and risk of epithelial ovarian cancer: an Australian case-control study. Survey of Women’s Health Study Group, Int. J. Cancer 62 (1995) 678–684. [19] A.B. Spurdle, B. Goodwin, E. Hodgson, J.L. Hopper, X. Chen, D. Purdie, et al., The CYP3A4*1B polymorphism has no functional significance and is not associated with risk of breast or ovarian cancer, Pharmacogenetics 12 (2002) 355–366. [20] A.B. Spurdle, X.Q. Chen, M. Abbazadegan, N. Martin, S.K. Khoo, T. Hurst, B. Ward, CYP17 promotor polymorphism and ovarian cancer risk, Int. J. Cancer 86 (2000) 436–439. [21] A.B. Spurdle, J.L. Hopper, X. Chen, G.S. Dite, M.R. McCredie, G.G. Giles, et al., The steroid 5alpha-reductase type II TA repeat polymorphism is not associated with risk of breast or ovarian cancer in Australian women, Cancer Epidemiol. Biomarkers Prev. 10 (2001) 1287–1293. [22] A.B. Spurdle, P.M. Webb, X. Chen, N. Martin, G.G. Giles, J.L. Hopper, et al., Androgen receptor exon 1 CAG repeat length and risk of ovarian cancer, Int. J. Cancer 87 (2000) 637–643. [23] A.B. Spurdle, P.M. Webb, D.M. Purdie, X. Chen, A. Green, G. Chenevix-Trench, No significant association between progesterone receptor exon 4 Val660Leu G/T polymorphism and risk of ovarian cancer, Carcinogenesis 22 (2001) 717–721. [24] A. Berchuck, J.M. Schildkraut, R.M. Wenham, B. Calingaert, S. Ali, A. Henriott, et al., Progesterone receptor promoter +331A polymorphism is associated with a reduced risk of endometrioid and clear cell ovarian cancers Cancer Epidemiol, Biomarkers Prev. 13 (2004) 2141–2147. [25] P.M. Webb, J.L. Hopper, B. Newman, X. Chen, L. Kelemen, G.G. Giles, et al., Double-strand break repair gene polymorphisms and risk of breast or ovarian cancer, Cancer Epidemiol. Biomarkers Prev. 14 (2005) 319–323. [26] L. Kelemen, A.B. Spurdle, D.M. Purdie, D. Gertig, G. Chenevix-Trench, RAD52 Y415X truncation polymorphism
104
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
C.M. Nagle et al. / Cancer Letters 251 (2007) 96–104 and epithelial ovarian cancer risk in Australian women, Cancer Lett. 218 (2005) 191–197. A. Auranen, A.B. Spurdle, X. Chen, J. Lipscombe, D.M. Purdie, J.L. Hopper, et al., BRCA2 Arg372His polymorphism and epithelial ovarian cancer risk, Int. J. Cancer 103 (2003) 427–430. V.N. Kristensen, N. Harada, N. Yoshimura, E. Haraldsen, P.E. Lonning, B. Erikstein, et al., Genetic variants of CYP19 (aromatase) and breast cancer risk, Oncogene 19 (2000) 1329–1333. A.M. Dunning, M. Dowsett, C.S. Healey, L. Tee, R.N. Luben, E. Folkerd, et al., Polymorphisms associated with circulating sex hormone levels in postmenopausal women, J. Natl. Cancer Inst. 96 (2004) 936–945. I.U. Agoulnik, Z.W. Tong, D.C. Fischer, K. Korner, N.E. Atkinson, D.P. Edwards, et al., A germline variation in the progesterone receptor gene increases transcriptional activity and may modify ovarian cancer risk, J. Clin. Endocrinol. Metab. 89 (2004) 6340–6347. I. De Vivo, G.S. Huggins, S.E. Hankinson, P.J. Lescault, M. Boezen, G.A. Colditz, et al., A functional polymorphism in the promoter of the progesterone receptor gene associated with endometrial cancer risk, Proc. Natl. Acad. Sci. 99 (2002) 12263–12268. N.M. Makridakis, E. di Salle, J.K. Reichardt, Biochemical and pharmacogenetic dissection of human steroid 5 alphareductase type II, Pharmacogenetics 10 (2000) 407–413. N.L. Chamberlain, E.D. Driver, R.L. Miesfeld, The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function, Nucleic Acids Res. 22 (1994) 3181–3186. S. Rafii, P. O’Regan, G. Xinarianos, I. Azmy, T. Stephenson, M. Reed, et al., A potential role for the XRCC2 R188H polymorphic site in DNA-damage repair and breast cancer, Hum. Mol. Genet. 11 (2002) 1433–1438. T. Takanami, J. Nakamura, Y. Kubota, S. Horiuchi, The Arg280His polymorphism in X-ray repair cross-comple-
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
menting gene 1 impairs DNA repair ability, Mutat. Res. 582 (2005) 135–145. E.R. Jupe, A.A. Badgett, B.R. Neas, M.A. Craft, D.S. Mitchell, R. Resta, et al., Single nucleotide polymorphism in prohibition 39 untranslated region and breast-cancer susceptibility, Lancet 357 (2001) 1588–1589. E.L. Goode, A.M. Dunning, B. Kuschel, C.S. Healey, N.E. Day, B.A.J. Ponder, et al., Effect of germ-line genetic variation on breast cancer survival in a population-based study, Cancer Res. 62 (2002) 3052–3057. C.H. van Gils, N.C. Onland-Moret, M. Roest, P.A.H. van Noord, P.H.M. Peeters, The V89L polymorphism in the 5alpha-reductase type 2 gene and risk of breast cancer, Cancer Epidemiol. Biomarkers Prev. 12 (2003) 1194–1199. A. Scorilas, B. Bharaj, M. Giai, E.P. Diamandis, Codon 89 polymorphism in the human 5 alpha-reductase gene in primary breast cancer, Br. J. Cancer 84 (2001) 760–767. A. Shibata, M.I. Garcia, I. Cheng, T.A. Stamey, J.E. McNeal, J.D. Brooks, et al., Polymorphisms in the androgen receptor and type II 5 alpha-reductase genes and prostate cancer prognosis, Prostate 52 (2002) 269–278. E.I. Garner, E.E. Stokes, R.S. Berkowitz, S.C. Mok, D.W. Cramer, Polymorphisms of the estrogen-metabolizing genes CYP17 and catechol-O-methyltransferase and risk of epithelial ovarian cancer, Cancer Res. 62 (2002) 3058–3062. J. Fujimoto, S. Ichigo, M. Hori, M. Nishigaki, T. Tamaya, Expression of progesterone receptor form A and B mRNAs in gynecologic malignant tumors, Tumour Biol. 16 (1995) 254–260. J. Fujimoto, S. Ichigo, R. Hirose, H. Sakaguchi, T. Tamaya, Clinical implication of expression of progesterone receptor form A and B mRNAs in secondary spreading of gynecologic cancers, J. Steroid Biochem. Mol. Biol. 62 (1997) 449– 454. C. Nagle, P. Webb, D. Purdie, A. Green, C. Bain, Searching for cancer deaths on Australia: NDI vs. cancer registries, Asian Pac. J. Cancer Prev. 7 (2006) 41–46.