Association of the hypoxia-inducible factor-1α (HIF-1α) gene polymorphisms with prognosis in ovarian clear cell carcinoma

Ovarian cancer is the leading cause of death among gynecological malignancies, as well as is the fourth most common malignancy in women in developed countries, following breast, lung, and colorectal cancer [1, 2]. Each of the ovarian cancers, represented by serous carcinoma, endometrioid carcinoma, clear cell carcinoma, and mucinous carcinoma, are known to have specific clinicopathological features and molecular or genetic characteristics. In Japan, ovarian clear cell carcinoma (OCCC) is the second most common ovarian cancer, following serous carcinoma [3, 4]. OCCC arises from endometriosis in 50–70% of the cases [5, 6] and has a more unfavorable prognosis due to a poor response to platinum-based chemotherapy, compared with serous carcinoma [3, 4].

HIF-1α is a key regulator of cellular response to hypoxia and plays an important role in tumor growth by trans-activating various genes that are related to regulation of angiogenesis, energy metabolism, survival, resistance to anti-tumor therapy, and cell survival, apoptosis, and proliferation [7‐9]. In our previous studies of OCCC and other ovarian epithelial cancers, we found an increased nuclear expression of HIF-1α in OCCC and have identified the HIF-1α regulating factors [10, 11]. Genetic polymorphisms are responsible for inter-individual variation and diversity, and have been recently considered as the main genetic elements involved in the development and progression of cancer [12]. HIF-1α gene SNPs are more frequent in several cancers than in healthy groups [13‐29]. Furthermore, they are associated with a poor prognosis in some cancers, including non-small cell lung cancer [13, 14], breast cancer [15, 16], head and neck squamous cell carcinoma [17], prostate cancer [18], bladder cancer [19], and glioma [20]. A total of 35 SNPs have been located within the HIF-1α gene. Three of the 35 SNPs were located in coding regions, one in exon 2, and the others in exon 12 [30]. The two SNPs located within exon 12 (codon 582 and 588) were associated with transcriptional activity [9, 30]. The C to T transition at nucleotide 1772 leads to an amino acid change of proline to serine at codon 582 (C1772T/P582S/rs11549465), and the G to A nucleotide substitution at point 1790 gives rise to an alanine/threonine variation at codon 588 (G1790A/A588T/rs11549467).

This study was conducted to investigate the impact and susceptibility of HIF-1α gene SNPs (C1772T and G1790A) on the prognosis of OCCCs because there have been no reports to analyze the association of the SNPs with outcome. In particular, the two SNPs associated with transcriptional activity were the focus of the study because they were associated with transcriptional activity.

The genotypes of the homozygous wild-type HIF-1α gene SNPs C1772T (CC) and G1790A (GG) as well as heterozygous/homozygous SNPs C1772T (CT + TT) and G1790A (GA + AA) were identified (Fig. 1). Among the 89 OCCC patients, 23.6 and 3.3% showed the presence of C1772T and G1790A SNPs in the HIF-1α gene, respectively. Results were compared with those for the Japanese healthy population group; prevalence of C1772T and G1790A SNPs was 9.1–11.0% and 8.2–8.7%, respectively [19, 31‐33]. All clinicopathological results (age, FIGO stage, surgical residual tumor, recurrence, and death) failed to show a significant relationship with the SNPs (Table 1).

In Kaplan-Meier survival curves, C1772T SNPs (CT + TT genotype) had no significant adverse effect on OS (p = 0.673, Fig. 2a) and PFS (p = 0.318, Fig. 2b). G1790A SNPs (GA + AA genotype) also had no significant adverse effect on OS (p = 0.643, Fig. 2c) and PFS (p = 0.748, Fig. 2d). Additionally, C1772T and/or G1790A SNPs (CT + TT and/or GA + AA) had no significant adverse effect on OS (p = 0.845, Fig. 2e) and PFS (p = 0.400, Fig. 2f). However, FIGO stage and surgical residual tumor had a significant adverse effect on OS (p = < 0.001; p = < 0.001, respectively) and PFS (p = < 0.001; p = < 0.001, respectively).

In the univariate analysis using the Cox proportional hazard model, OS was associated with FIGO stage (hazard ratio (HR) = 15.62; 95% confidence interval (CI) = 4.949 to 49.31; p = < 0.001) and surgical residual tumor (HR = 16.13; 95% CI = 5.780 to 45.00; p = < 0.001), but not with C1772T (HR = 0.762; 95% CI, 0.215 to 2.701; p = 0.674), G1790A (HR = 1.609; 95% CI, 0.211 to 12.28; p = 0.647), C1772T and/or G1790A (HR = 0.892; 95% CI, 0.284 to 2.803; p = 0.845), and age (HR = 1.463; 95% CI, 0.529 to 4.049; p = 0.463) (Table 2). In the multivariate survival analysis, FIGO stage (HR = 7.527; 95% CI, 1.808 to 31.33; p = 0.006) and surgical residual tumor (HR = 4.030; 95% CI, 1.127 to 14.41; p = 0.032) were found to be the independent prognostic factors (Table 2).

HIF-1α expression represents an important biomarker in the evaluation of ovarian carcinoma prognosis [34]. In our study, OCCCs are characterized by a nuclear expression of HIF-1α compared to other histological types. It is believed that HIF-1α is one of the key factors closely associated with chemo-resistance or unfavorable OCCC prognosis [10, 11]. Overexpression of HIF-1α may be attributed to transcriptional and/or translational activity, nuclear transition of the protein, and its degradation.

This study was conducted to assess whether there is an association of the HIF-1α gene SNPs with the prognosis and clinicopathological characteristics of OCCCs. A significant association was observed between prognosis and clinicopathological factors such as FIGO stage and surgical residual tumor. However, any variations of the SNPs were proven not to be associated with the prognosis. The previous studies of variable cancers with a focus on the relationship between HIF-1α gene SNPs and patient prognosis are summarized in Table 3 [13‐29]. OCCC patients had more frequent C1772T SNPs than the healthy Japanese population [19, 31‐33] and many other carcinomas. OCCC prognosis as well as colorectal cancer [21, 22], thymic malignancy [27], and cervical cancer [28, 29] prognoses had no association with C1772T and G1790A SNPs. However, the T allele of C1772T and A allele of G1790A are a poor or good prognostic factor in several cancers [17, 23]. The effects of HIF-1α SNPs on the prognosis with cancers are not uniform.

The C1772T SNP has been reported to increase HIF-1α protein expression in some cancers [13, 15]. Twenty specimens, which were randomly selected out of the 89 OCCCs examined in this study, were subjected to immunohistochemical staining for HIF-1α. The results failed to show the associated between HIF-1α staining and presence of SNPs (data not shown). In our previous studies, OCCCs showed the highest frequency of HIF-1α, histone deacetylase (HDAC) 6, and HDAC7 compared to other ovarian epithelial cancer [10, 11, 35]. HDAC6 and HDAC7 induced not only HIF-1α transcriptional activity, but also stabilized HIF-1α protein via interaction with von Hippel Lindau and ubiquitin-independent proteasomal degradation of HIF-1α [36‐38]. In OCCCs, post-translational modification may be more important for the HIF-1α expressions than upregulated transcription activity by HIF-1α gene SNPs.

Our study has several limitations. The sample size used in this study was small and the survival analysis was only performed with a few events. However, when considering the low incidence of OCCC, the present study included a relatively large number of patients. Secondly, normal controls were not recruited in the present study; instead, we compared the frequencies of HIF-1α SNPs using the normal Japanese population reported in the past studies [19, 31‐33].