Abstract
Background/Aim: Ovarian cancer is a lethal gynecological malignancy, with 5-year survival of only about one third of patients. The ABCB1 gene encodes the P-glycoprotein which is one of the multidrug efflux pumps. Its decreased activity may result in multidrug resistance of cancer cells. Drug-metabolizing enzymes Cyp3A4 and Cyp3A5 may affect success of chemotherapy. In this study we attempted to examine the effects of 12 single nucleotide polymorphisms (SNPs) of the ABCB1 gene and one SNP in each of CYP3A4 and CYP3A5 genes on the incidence of ovarian cancer in Polish women and their response to treatment. Materials and Methods: Our study included 276 patients and 369 healthy control women. Results: The results showed no significant differences between patients and controls in allele frequencies of the tested SNPs, with one exception: rs2157926T allele decreased cancer risk by 99.4% (odds ratio, 0.006). Moreover, rs2032582T increased fourfold the risk of metastasis. Finally, rs1128503CC genotype prolonged survival (p=0.024778). Conclusion: These findings may contribute to a better prediction of therapy outcome.
Ovarian cancer is a lethal gynecologic malignancy, with a mean age-standardized 5-year survival of 34.5% in Poland, and similar values for other European countries and in U.S.A. (1, 2). The patients' survival depends on many factors like, e.g., cancer stage and class, metastasis and the response to treatment. The ABCB1 gene (ATP-binding cassette, subfamily B, member 1; alternative name: MDR1, multidrug resistance 1; located in humans on chromosome 7q21.12) encodes the P-glycoprotein (Pgp), which functions as an ATP-dependent multidrug efflux pump, exporting exogenous and endogenous substrates from the cell (3). High Pgp levels are associated with a poor response of cancer cells to chemotherapeutic agents. In contrast, low Pgp levels characterize tumors that are sensitive to chemotherapy. Several single nucleotide polymorphisms (SNPs) in ABCB1 have been described in Caucasians. Association of these polymorphisms with the response of ovarian cancer to chemotherapy has been investigated in recent years, but results are conflicting (4-7).
Like cytochrome P450 3A4, P-glycoprotein is non-specific for a large variety of drugs, and there is a wide overlap with the drug-metabolizing enzyme CYP3A4, making P-glycoprotein and CYP3A4 a synergistic defense mechanism against the intrusion of xenobiotics (3). Therefore, we attempted to examine the effects of twelve SNPs (rs10245483, rs2157930, rs10246878, rs2157926, rs3213619, rs2214102, rs9282564, rs868755, rs2229109, rs1128503, rs1045642, rs2032582) of the ABCB1 gene and two polymorphisms of cytochrome P450 (CYP3A4, CYP3A5) on response to chemotherapy, metastasis and patients' survival in Polish women with ovarian cancer.
Materials and Methods
Study subjects. Two hundred and seventy-six women diagnosed with ovarian cancer and treated, as shown in Table I, in Lower Silesian Oncology Center, Wrocław or in WSS Hospital, Wrocław, were enrolled in our study. The inclusion criteria comprised histologically confirmed diagnosis of ovarian cancer and treatment with surgery and platinum-based chemotherapy; borderline tumor patients were excluded. For a control, we included two populations: Control #1 consisted of healthy blood donors. However, most of them were below 50 years of age, in contrast to our patients. Therefore, Control #2 was added, consisting of WSS Hospital patients with non-malignant diseases. Since SNP frequencies in both controls were very similar (results not shown), these two controls were combined in further analyses.
Our study has been approved by a Biomedical Committee of the Medical University in Wrocław, Poland.
DNA isolation and SNP typing. Genomic DNA was extracted from fresh or frozen blood samples using a QIAamp® DNA blood mini kit (Qiagen, Hilden, Germany). DNA concentration was adjusted to 20 μg/ml with the kit's AE solution (10 mM Tris, 0.5 mM EDTA, pH 9).
Genotyping of ABCB1 SNPs was based on PCR (polymerase chain reaction) amplification of gene fragments containing specific polymorphisms and an amplicon analysis using either RFLP (restriction fragment length polymorphism) or HRM (high resolution melting). 150-300 bp fragments containing the SNPs of interest were amplified with primers either described in the literature or designed using Primer3 program and either ABCB1 genomic reference sequences NG_011513.1 and NC_000007.13, CYP3A4 sequence NG_008421.1, or CYP3A5 sequence NG_007938.1. To overcome some limitations of conventional HRM (8), a modification by Zhou et al. 2004 (9) with unlabeled oligonucleotide probes was used. PCR followed by HRM were performed in the presence of 3’-end blocked ~30-nt probes and DNA saturating fluorescent dye EvaGreen (Biotium, Fremont, USA) in the Eco Real-Time PCR System (Illumina, San Diego, USA). The Eco system operations and HRM data processing were carried out using Eco System software v.5.0, which allowed the melting curves to be normalized and directly compared.
Statistical methods. Allele frequencies were assessed by gene counting and distribution of the polymorphic variants was tested with the Hardy-Weinberg equilibrium. Backward stepwise multivariate regression model was used to determine odds ratios (OR) and 95% confidence intervals (CI) for evaluating the risk association of ABCB1, CYP3A4, CYP3A5 and BRCA1 polymorphisms with the incidence rate of ovarian cancer. Discrepancies between controls and cases were determined using χ2 Pearson test. Survival time model was performed by multiple regression and confirmed by median test.
Statistical analysis was performed using Statistica ver. 10 (StatSoft) with Medical Package. The differences were considered significant when p-value <0.05.
Results
ABCB1, CYP3A4, CYP3A5 polymorphisms were tested in 276 women with ovarian cancer and 370 healthy volunteers. There was no evidence of any deviations from the Hardy-Weinberg equilibrium in the distribution of ABCB1, CYP3A4, CYP3A5 genotypes in the study and control groups (Table II).
There was no evidence of any deviation from the Hardy–Weinberg equilibrium (p=0.3589 and p=0.5441, chi-squared goodness of fit test) in the distribution of PTPN13 genotypes in either the control or the study group (Table II).
There was no evidence of any deviation from the Hardy–Weinberg equilibrium (p=0.3589 and p=0.5441, chi-squared goodness of fit test) in the distribution of PTPN13 genotypes in either the control or the study group (Table II).
Characteristics of patients and control subjects.
We didn't observe any statistically significant differences between cancer and control group for rs10245483, rs2157930, rs10246878, rs2157926, rs3213619, rs2214102, rs9282564, rs868755, rs2229109, rs1128503, rs1045642, rs2032582 allele frequencies analyzed by χ1 Pearson test (data not shown). Backward stepwise multivariate regression model showed that “T” allele occurrence in the rs2157926 polymorphism reduced cancer risk by 99.4% (p-value=0.0000, OR=0.006, 95%CI=0.002-0.021; Table III).
Furthermore, backward stepwise multivariate regression model showed that “T” allele of the rs2032582 polymorphism increases metastasis risk four times (p-value=0.038, OR=3.925, 95%CI=1.076-14.318; Table IV).
Distribution of genotypes of ABCB1, CYP3A4, CYP3A5 in the study and control groups.
Multiple regression showed that “CC” genotype in the rs1128503 polymorphism statistically significantly increases survival time (Table V). This result was confirmed by median test (χ2=7.602, p-value=0.0224) (Table VI).
Discussion
Multiple polymorphisms have been described for the ABCB1 gene. For some of them, their effects on mRNA and protein expression in different tissues as well as on the ABCB1 transporter activity have been observed in vitro. However, these effects so far seem too weak for reliable predictions with regard to prevalence of disease and a patient's response to treatment (3, 7), therefore further studies are necessary. Here, among two SNPs in CYP3A4 and CYP3A5 and 12 SNPs in ABCB1 gene we observed an association with ovarian cancer risk for ABCB1 rs2157926 SNP only. This SNP is located in intron 1 (-40106A>T) and has not been extensively studied so far (3, 7), therefore its effect on mRNA or protein expression or on treatment response is not established.
rs2032582 allele frequencies did not differ in our patients and controls. Similarly, in a large study of over 4,600 ovarian cancer patients from the Ovarian Cancer Association Consortium and The Cancer Genome Atlas, Johnatty et al. (5) did not observe any association of rs2032582 or any other SNP of the ABCB1 gene with the overall survival or progression-free survival, although in an earlier study with a much lower number of patients, the same authors observed an effect of rs2032582 on progression-free survival (10). However, we found that rs2032582T allele increased metastasis risk about fourfold. This SNP has three alleles (G, T and A) and encodes an amino acid change in position 893 from alanine to serine or threonine, respectively. Notably, the different alleles were described to have similar expression in normal tissues, unselected cell lines, and untreated malignant lymphomas, but there was loss of heterozygosity for rs2032582G>T in a number of selected cell lines and relapsed malignant lymphomas (3). The T allele (893Ser) was observed to increase Vmax for vincristine (3) and increase the response to imanitib in chronic myelocytic leukemia (7). However, many studies on rs2032582G>T-A role in anticancer chemotherapy of different types of malignancy brought conflicting results, and generally the effects of all ABCB1 polymorphisms were weak (7).
Backward stepwise multivariate regression model on ovarian cancer risk.
Backward stepwise multivariate regression model on risk of the metastasis.
Multiple regression model on survival time.
On the other hand, rs1128503CC genotype prolonged survival of our patients. This SNP is located in exon 12, but is synonymous (Gly412Gly). Its role was recently studied in several human diseases and their treatment outcome, but frequently with ambiguous results, depending on disease or ethnicity (11-15). No effect of rs1128503 or any of the SNPs tested here, but only effects of some other SNPs on treatment outcome were found in U.S. ovarian cancer patients (16).
Also, we did not detect any effect of rs1045642 on ovarian cancer risk or treatment outcome. This is a silent variant (Ile1145Ile), but nevertheless was suggested by some authors to affect Pgp protein expression, although it was questioned by others (3, 7). It was not examined in American ovarian cancer patients (16).
Recently, Tęcza et al. [2015] (17) published results showing associations of rs2032582 and rs1045642 with ovarian cancer, but only in patients positive and negative for BRCA1 mutations, respectively. In contrast, we have not observed these relations in similar numbers of patients and controls. Grzybowska et al. [2002] (18) described the role of the germline mutations in the BRCA1 gene in predisposition to breast and ovarian cancer in Upper Silesian population. They showed a very high frequency of single BRCA1 5382insC mutation and BRCA2 9631delC. They suggested that this particular mutation in BRCA2 gene is limited to the Southern part of Poland (18). We observed a significantly lower frequency in these two mutations in Lower Silesian population (data not shown). This can be a cause for differences between those two populations.
Median test model on survival time.
Altogether, our results suggest that genotyping for rs2157926, rs2032582 and rs1128503 may help predict ovarian cancer outcome.
Acknowledgements
This work was supported by a grant N N401 063735 from the Polish National Science Centre. The Authors would like to express their gratitude to all the patients and control volunteers for kindly donating their blood and their agreement to participate in this study.
- Received January 9, 2018.
- Revision received January 23, 2018.
- Accepted January 24, 2018.
- Copyright© 2018, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
In this issue
