Polymorphisms of T- cell leukemia 1A gene loci are not related to the development of adjuvant letrozole-induced adverse events in breast cancer

Letrozole, an aromatase inhibitor (AI), is the first-line adjuvant drug for treating hormone receptor-positive (HR+) breast cancer in postmenopausal women. However, harmful adverse events (AEs) and significant differences in drug response among individuals remain a significant problem in clinical application. Current evidence suggests that the observed individual variation in the treatment outcomes of AI is conferred by genetic variants. Hence, in this study, we examined the association of TCL1A gene polymorphisms with letrozole-induced AEs. The study subjects were postmenopausal HR+ breast cancer patients who were receiving adjuvant letrozole. Genomic DNA was isolated by a routine standard phenol-chloroform method. In total, 198 South Indian patients were genotyped for four single nucleotide polymorphisms (SNPs) in the TCL1A gene loci by the TaqMan allelic discrimination assay using the RT-PCR system. We used the odds ratio and 95% confidence interval to assess the genetic association. Musculoskeletal (MS) AEs and vasomotor symptoms (VMSs) are the most common side effects observed in the study cohort. Among 198 patients, 81 experienced musculoskeletal toxicity, reporting MS-AEs, 57 had VMSs, and 33 of them had both. The most frequently identified polymorphic variants in the patient series were rs11849538 (G), with an allele frequency of about 27.3%, followed by rs7158782-G (27.3%), rs7159713-G (25.8%), and rs2369049-G (22.5%). The genetic association analysis indicated no significant difference in the proportion of TCL1A gene variants between patients with and without AEs on either MS-AEs or VMSs. Though we observed high LD in all patient groups, the inferred haplotypes displayed a non-significant association with letrozole-induced specific AEs. However, the SNP functionality analysis by RegulomeDB provided a 2b rank score for rs7158782, suggesting a potential biological function. Our findings suggest that TCL1A gene polymorphisms may not play any role in the prediction of letrozole-induced AEs in South Indian HR+ breast cancer patients.

Introduction confirm its effect on other independent populations. As evidenced by our earlier reports, the South Indians represent a unique set of ethnicity-specific genetic signatures within the pattern of genetic variants in genes that governs drug metabolism and transport as compared to other global races [18].
Till now, no research into TCL1A gene polymorphisms and AIs treatment outcomes has been described in Indians. Therefore, we indended to understand the clinical relevance of TCL1A genetic polymorphisms on the development of letrozole-induced 'specific' AEs in postmenopausal HR+ breast cancer patients of South Indian ancestry.

Patient population
We identified a total of 198 breast cancer patients (median age 56 years with a range from 38 to 85 years) who were getting adjuvant treatment (Letrozole, 2.5mg OD) at Tumor Clinic, Outpatient Departments of General Surgery and Medical Oncology, Jawaharlal Institute of Postgraduate Medical Education and Research (JIPMER), Pondicherry from January 2009 through January 2014. The research protocol was approved by the Institutional Scientific Advisory and Research Ethics Committees of JIPMER, Pondicherry. Signed informed consent was obtained from each study subject. All of them were of South Indian origin, living in the southern states of Tamil Nadu, Puducherry, Karnataka, Kerala, Andhra Pradesh, and Telangana for more than three generations. The diagnosis of breast cancer was confirmed by histopathological examination. Details of this study design, selection criteria, toxicity assessment, questionnaire, and treatment have been described previously [14].

DNA extraction and genotyping
We extracted the genomic DNA from stored peripheral blood mononucleated cells by the routine standard phenol-chloroform method and used as DNA templates for genotyping. Genotypes of rs7158782 (assay ID C_29078024_10), rs7159713 (assay ID C_1927662_10), rs2369049 (assay ID C_1927663_20), and rs11849538 (assay ID C_1927667_10), SNPs were determined with predesigned TaqMan SNP genotyping assays obtained from Life Technologies on ABI 7300 quantitative RT-PCR system based on allelic discrimination according to the manufacturer's instructions. The details of the RT-PCR procedure have been described previously [19]. For quality control, replication was performed by other laboratory personnel in 30% of the randomly chosen samples to confirm the authenticity of the genotype results obtained earlier, and they were found to be in 100% concordance.

Statistical analyses
Haplotype frequency and linkage disequilibrium (LD) patterns were analyzed using Haploview program version 4.2 (http://www.broad.tamit.edu/mpg/haploview). The allele and genotype frequencies were calculated by the direct gene count method. Hardy-Weinberg equilibrium (HWE) was achieved by the chi-square test to assess the bias in the genotype distribution. Patients were classified into five subgroups based on the presence or absence of 'specific' AEs: MS-AEs (patients with MS-AEs), VMSs (patients with VMSs), Both (patients with MS-AEs and VMSs), Any AEs (patients with either MS-AEs or VMSs), and No AEs (patients without any AEs). For the genetic association analysis, we applied logistic regression to estimate odds ratio (OR) and 95% confidential interval (CI) for each group vs. patients without any AEs using six different genetic models (allele, co-dominant, dominant, recessive, over-dominant, and additive). The covariates adjusted were age, type/regime of chemotherapy, reasons for menopause, and time since the menopause (LMP). The above statistical tests were carried out using SPSS version 19.0 (SPSS Inc., Chicago, IL, USA). All the p values were two-sided, and the statistical significance threshold was set at P< 0.05. To predict the functional significance of each SNP, we employed RegulomeDB (http://www.regulomedb.org) online database.
We calculated the sample size to be 198 using PS Power & Sample Size Calculations software version 3.1.2 and the QUANTO program. The power was set at 0.80; α = 0.05 based on the least variant allele frequency of TCL1A gene polymorphism (rs2369049-G allele, 21.2%) in healthy South Indians [19] to demonstrate an association between the variant genotypes and letrozole-related 'specific' adverse events. Further, genetic heterogeneity could lead to either reduced power or spurious associations in pharmacogenetic epidemiological studies. Hence, we carried out a population-specific investigation restricting to South Indian subjects to avoid the ethnicity variations leading to a pharmacogenetic anomaly.

Baseline characteristic features of the study cohort
The baseline histopathological and clinical characteristics of all the study subjects are summarized and presented in Table 1. The median age of the series was 56 years, with a range from 38 to 85 years; the median BMI was 25.8 kg/m2 (range 14.8-45.2). As shown, most of the cases were grade I-II (83.8%), and most of them had a T2 tumor (38.9%). All the tumors expressed hormone receptor positivity, either ER +ve (89.9%) or PR +ve (67.2%) or both (55.5%), and 70 patients were positive for Her2. Among the diagnosed patients, chemotherapy was given to a total of 163 patients. Additionally, 80.8% of patients (160/198) had been treated with radiotherapy, and 22.2% of patients (44/198) completed prior tamoxifen therapy. Twenty-seven patients (13.6%) underwent BCS, and the remaining 171 cases had undergone a radical mastectomy. We have distinguished patients' education status into two groups: those with a high school or lower education (43.9%) and those with college or higher studies (9.6%). Regarding menopause, 141 patients attained it naturally, and the rest 57 cases were induced either by surgery or chemotherapy. We previously reported the clinical and pathological characteristics of patients stratified by 'specific' adverse events for this patient cohort [14]. Table 2 describes the genotype and allele frequency distributions of TCL1A genetic variants in the study population. According to HWE, the observed genotype frequencies of all the tested genetic variants were in agreement with the expected frequencies for both patients with and without 'specific' AEs. The most frequently identified polymorphic variant in the patient series was rs11849538 (G), with an allele frequency of about 27.3%. The observed frequency of cases with other minor alleles were rs7158782-G (27.3%), rs7159713-G (25.8%), and rs2369049-G (22.5%). All the polymorphisms described in this study were evaluated with Regulome RD online portal for their in silico functional prediction. Of them, the SNP rs7158782 has a less score of 2b, indicates its potential functional significance.

Association between letrozole-induced 'specific' AEs and TCL1A genetic variants
To evaluate the genetic effects of TCL1A genotypes on MS toxicity risk, the distribution of the dataset for each SNP was analyzed for co-dominant, dominant, recessive, over-dominant, and additive genetic models ( Table 3). None of the individual TCL1A genetic polymorphisms were found to be significantly associated with MS-AEs risk. As evident from Table 3, the association analyses of TCL1A genetic polymorphisms with VMSs also revealed statistically no significant difference (P>0.05) in the proportion of the genotype frequency among patients with VMSs and without any AEs. The different genetic model predictions also displayed a non-significant association with VMSs risk. We also used multiple logistic regression modeling to estimate the degree of pharmacogenetic association of 'specific' AEs explained by the genotypes of the chosen SNPs. For this multivariate model, we considered adjustments for the following potential confounders: age, BMI, year since last menopause, year since the start of letrozole, and chemotherapy. The patterns of allelic association were statistically similar, as observed for the genotypes.
We also assessed the influence of TCL1A gene haplotypes with letrozole-induced 'specific' AEs. Fig 1 shows the linkage disequilibrium (LD) patterns of TCL1A gene polymorphisms. Overall, we observed high LD in all patient groups. Four marker haplotype analyses inferred four different haplotypes with a frequency of above 1%. There was no significant difference in the proportion of these haplotype frequencies between patients with and without 'specific' AEs (Table 4). Further, we have performed two-marker sliding window analyses, which also indicated similar outcomes (Table 5).

Discussion
Breast cancer, a highly dreadful disease, is globally escalating [20]. It is often associated with poor prognosis and adverse outcomes. This is because of the striking variability in response to modern anti-cancer medications. Inter-individual differences underpinning pharmacological responses are driven for the most part by common genetic variants, the so-called polymorphisms in the DNA sequence of the genes encoding proteins, which controls drug absorption, distribution, metabolism, and excretion [21]. Over the decade, accelerated developments in molecular biology have brought a revolution in molecular medicine. One of the significant areas of biomarker research using pharmacogenomics as a screening tool is cancer.
As with most anti-cancer drugs, it is necessary to identify a potent pharmacogenetic marker that could determine the treatment outcomes of AIs. In addition, to a lesser extent, researchers investigated the role of TCL1A gene polymorphisms in the etiology of AIs mediated toxicity [15,22]. Genetic polymorphisms in the TCL1A loci gained attention after the indication of its association with MS toxicity risk in breast cancer patients undergoing anti-hormonal therapy with third-generation AIs [15]. Existing epidemiological investigations have implicated that race or ethnicity is an important factor inducing gene-drug relationships [23]. As of now, the normative frequency of these TCL1A gene polymorphisms was established only in postgenome project populations, namely HapMap (http://www.Hapmap.org) and 1000 genomes (http://www.1000Genomes.org). Nevertheless, more recently, we studied the distribution of four clinically important genetic polymorphisms in TCL1A gene loci and described significant inter-ethnic variation [19]. The variant allele frequency of the TCL1A gene differs by ethnicity. The SNPs namely, rs7158782, rs7159713, rs2369049, and rs11849538 were found to occur at a higher frequency of 75-84.2%, 75-84.1%, 73-84.1%, and 34-44.1%, respectively, in African populations, but were identified with very less frequencies in Asians (13.1-29%, 13.6-29%, 3.5-27%, and 17-30%, respectively), Caucasians (13.7-15.9%, 13.7-15.9%, 13.3-15.3%, and 8.8-14.0%, respectively), and Hispanics (21-25%, 21-25%, 13-24%, and 17-18%, respectively). This may be the first study to evaluate the association of TCL1A gene variants in South Indians to the best of our knowledge and with the available evidence. We observed no significant SNP-level or haplotype-based associations of letrozole-induced 'specific' AEs for TCL1A genetic variants. Only two studies have been carried out to elucidate the possible role of TCL1A genetic polymorphisms on AI-associated outcomes [15,22]. A case-control GWAS identified four polymorphisms (rs7158782, rs7159713, rs2369049, and rs11849538) corresponding to the TCL1A gene at chromosome 14, which were attributed to increased MS-AEs [15]. In contrast, the other study explored the potential association between toxicity-related discontinuation of AI medication and genetic variants in the TCL1A gene. In that study, the authors unable to confirm the findings of GWAS in their prospective observation performed on breast cancer patients enrolled from Exemestane and Letrozole Pharmacogenetics (ELPH) clinical trial [22]. Notably, despite close LD between these polymorphisms, their impact on the occurrence of MS-AEs and VMSs could not be evaluated in our study, which supported the previous findings by Henry et al [22]. Furthermore, a significant interaction between TCL1A polymorphism (rs7158772) and drug effectiveness of AIs was not found in 308 women with metastatic breast cancer [24]. Similarly, in a study of anastrozole versus tamoxifen following five years of adjuvant use, the polymorphism rs11849538 was thought to be related to AIs efficacy; however, no association was found [12]. Moreover, the variant genotypes at SNP rs11849538 created an estrogen response element (ERE). Besides, this gene was found to be involved in the regulation of a series of genes encoding inflammatory response proteins (cytokines and its receptors) including the IL13RA1 (interleukin 13 receptor, a1), IL18R1 (interleukin 18 receptor 1), IL1R2 (interleukin 1 receptor, type 2), IL17RA (interleukin receptor A), and IL12RB2 (interleukin 12 receptor, b2). Subsequent in vitro experiments confirmed the estrogen-dependent transcription of the TCL1A gene; however, the mechanism by which the change in the TCL1A expression induces MS-AEs remained unclear [15]. However, recently, Ho and co-workers performed RNA-seq and Chipseq experiments in the lymphoblastoid cell lines model system. They demonstrated the transcriptional factor role of the TCL1A gene and the molecular mechanism for SNP-estrogendependent regulation across the genome [25]. Additionally, it has also been reported that genetic variants in genes such as RANKL (rs7984870), OPG (rs2073618), and ESR1 (rsrs2234693 and rs9340799) were associated with AI-related MS-AEs [26][27][28].
In this cohort, we could not evaluate the relationship between TCL1A variants and letrozole-induced 'specific' AEs. However, we cannot exclude the biological significance of TCL1A on patients' risk of AEs. One of the possible explanations for this non-significant effect is the sample size. Our sample size (n = 198) for the association analyses was smaller than those in the existing studies [15,22]. The GWAS which discovered the significant role of TCL1A SNPs was primarily conducted in 878 subjects of white descent [15]. Importantly, there were no such significant associations observed in the following ELPH cohort involving 467 patients [22]. Other potential factors, including ethnicity, disease stage, and study design, could contribute to the mixed results among the studies. It should be noted that the current investigation has some limitations. First, we could not do a time-to-event analysis, which is one of the limitations of the study. Another limitation is that in the present study, the analytical plan did not envisage the association of genetic variation with some grades of AEs. Although the study has enough power to reflect the probability of a statistical meaning, we cannot exclude the possibility of false-negative outcomes due to the small sample size; thus, the results for these analyses should be interpreted with caution.

Conclusions
In summary, we found no statistically significant association between the four studied TCL1A SNPs and letrozole-induced 'specific' AEs. Thus, it appears unlikely that polymorphisms in TCL1A loci may be a robust predictor for AEs in HR+ breast cancer patients treated with letrozole. However, in order to further refine the role of these TCL1A genetic variants on the etiology of AI-induced AEs, future studies should include a greater number of patients sample for analysis. Furthermore, future investigations could systematically examine the interactive effects (epistatic) between candidate genes and AI-induced AE.