Abstract
Background/Aim: Breast cancer (BCa) remains one of the most prevalent malignancies and a leading cause of cancer-related deaths globally. Understanding the genetic underpinnings of BCa is critical for advancing precision medicine, including the development of predictive biomarkers and repurposed therapies. This study aimed to identify and functionally annotate BCa-associated single nucleotide polymorphisms (SNPs) to identify biological risk genes and assess drug repositioning opportunities.
Materials and Methods: We extracted BCa-related SNPs from the GWAS Catalog, applying a genome-wide significance threshold (p-value <10−8) to identify 1,219 SNPs. From these, 14 missense variants were prioritized and evaluated using six complementary tools: missense annotation, cis-expression quantitative trait loci (eQTL) mapping, combined annotation dependent depletion (CADD), sorting intolerant from tolerant (SIFT), polymorphism phenotyping v2 (PolyPhen-2) and AlphaMissense. Genes were scored across these criteria, with those scoring ≥2 considered biologically relevant. GTEx data was used to assess tissue-specific gene expression. Allele frequencies across populations were obtained from the Ensembl database, and druggability was evaluated using DrugBank.
Results: We identified nine genes achieving the maximum score of 4 as the highest-priority candidates SLCO1B1 (rs4149056), ARHGEF38 (rs61751053), EXO1 (rs4149909), KDELC2 (rs74911261), MAPT (rs63750417), PHLDA3 (rs35383942), AKAP9 (rs6964587), ATXN7 (rs1053338) and DCLRE1B (rs11552449). SLCO1B1 (rs4149056) and ARHGEF38 (rs61751053), supported by functional and regulatory evidence. SLCO1B1, predominantly expressed in the liver, may influence BCa metastasis and drug metabolism; its variant shows population-specific allele distribution, being particularly higher in Europeans (16%). ARHGEF38, though variably expressed across tissues, may play regulatory roles relevant to tumorigenesis. Among the prioritized genes, MAPT was identified as the only druggable target, with existing therapeutics such as paclitaxel and docetaxel indirectly linked to its function, suggesting potential for drug repurposing. These findings provide a foundation for further studies on SNP-guided biomarkers and repositioned therapies targeting key BCa-related genes.
Conclusion: This integrative bioinformatics approach prioritized functionally significant BCa-associated SNPs and identified promising candidates for biomarker development and drug repositioning. The nine high-scoring variants, including SLCO1B1 and ARHGEF38 emerge as biologically impactful genes, while MAPT’s known drug interactions highlight its translational potential in repurposing existing anticancer agents.
- Breast cancer
- genome-wide association study (GWAS)
- SNP. functional annotation
- MAPT
- population genetics
- SLCO1B1
- rs4149056
- ARHGEF38
- rs61751053
Introduction
Cancer represents a significant global health burden, with Indonesia being particularly affected (1, 2). Breast cancer (BCa) is one of the most diagnosed cancers among women worldwide (3), though men can also be affected (4). It ranks as the fourth leading cause of cancer-related deaths, with 2,308,897 cases and 665,684 deaths annually (5). BCa etiology involves multiple factors, including genetics, environment, and lifestyle (6). Key risk factors include age, family history, and mutations in BRCA1 and BRCA2, which increase risk by up to 85% in familial cases. Other risk determinants include hormonal factors, excess weight, alcohol, tobacco, physical inactivity, and radiation exposure, all shaping individual risk profiles influencing BCa initiation and progression (7).
Estrogen and lifestyle factors such as obesity, high-fat diets, alcohol, and inactivity play significant roles in BCa development. Prevention targets healthier lifestyles and early screening (8, 9). Invasive ductal and lobular carcinoma are the most frequent BCa subtypes (10). BCa often metastasizes to lymph nodes, bones, lungs, liver, and brain (11, 12). Treatment typically combines surgery either breast-conserving or mastectomy with radiation, chemotherapy, and targeted therapies like trastuzumab and pertuzumab for specific tumor types, plus hormonal treatments when estrogen receptors are expressed (13). Treatment plans are personalized based on tumor characteristics, stage, and patient health.
BCa incidence and mortality vary among ethnicities and regions. Globally, people of Black African ancestry have higher rates and poorer outcomes (14) In the U.S., Black women show the highest incidence and mortality, influenced by genetic risk loci prevalence and socioeconomic factors (15, 16). Rising BCa rates are seen especially in poor and rural areas among non-Hispanic Black women, who are often diagnosed later and face worse survival, underscoring the need for targeted interventions and equitable care access (17). Despite progress, interactions among genetic variants and their role in BCa progression remain unclear.
Advances in bioinformatics enable better identification and analysis of genetic variants as potential biomarkers for early detection, prognosis, and therapy improvement (18). This study employs a bioinformatic approach using The Genome-Wide Association Study (GWAS) Catalog which identifies associations between genome-wide genetic variations and complex traits (19-21), Focusing on single nucleotide polymorphisms (SNPs), the most common human genetic variants (22). GWAS helps pinpoint risk variants contributing to disease (4). The study’s aim is to identify and prioritize BCa-associated SNPs from GWAS data and assess their functional relevance using multiple annotation tools, ranking candidate genes by integrated biological scores to highlight those most implicated in BCa pathogenesis. This approach aims to support improved prevention and therapeutic strategies (23).
Materials and Methods
Genes associated with breast cancer. This study commenced by systematically searching for SNPs associated with BCa using the GWAS Catalog database, a comprehensive resource available at https://www.ebi.ac.uk/gwas/ (24, 25). On January 15, 2025, a query using the keyword “breast cancer” yielded a total of 2,454 Associations linked to BCa susceptibility.
Functional annotation was subsequently performed using HaploReg version 4.2, accessible at https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php to characterize these variants. HaploReg was used specifically for positional mapping to verify that the missense variants identified from the GWAS Catalog were correctly annotated and properly located within their target genes. Haploreg integrates diverse genomic and epigenomic datasets, including linkage disequilibrium information from the 1000 Genomes Project, chromatin state annotations from the Roadmap Epigenomics and ENCODE projects, and expression quantitative trait loci (eQTL) data from GTEx, enabling comprehensive evaluation of noncoding regulatory variants (26, 27). The analysis specifically focused on SNPs with a high level of statistical significance, defined by a p-value threshold of less than 10−8, to prioritize variants most likely to have biological relevance in BCa. This approach facilitates the identification of candidate genes potentially implicated in disease pathogenesis by linking SNPs to regulatory elements and gene expression changes. The integration of multiple annotation layers provided by HaploReg version 4.2 allows for the generation of mechanistic hypotheses regarding how these genetic variants may influence BCa risk and progression.
Functional annotation analysis to prioritize genes associated with breast cancer. To systematically identify potential gene targets associated with BCa, a comprehensive scoring system was developed that integrates six distinct sets of functional annotations. Genes achieving a cumulative score of ≥2 were classified as biologically relevant risk genes (26). The functional annotation framework employed in this study included the components mentioned thereafter. (i) Missense mutations: These are single nucleotide changes in the DNA sequence that result in the substitution of one amino acid for another in the encoded protein, potentially altering its structure and function. Genes harboring missense SNPs were assigned one point due to their likely impact on protein functionality. (ii) eQTL analysis: This approach elucidates the relationship between genetic variants and gene expression levels across tissues (27). The analysis was performed using data from the Genotype-Tissue Expression (GTEx) Portal (https://gtexportal.org/home/) (26). Specifically, cis-eQTLs variants affecting gene expression in nearby genomic regions were determined by querying gene names in the “eQTL significance” section. Genes with significant SNPs (p-value <0.05) in BCa relevant tissues received one point, identifying tissue specific regulatory patterns associated with BCa. (iii) Combined annotation dependent depletion (CADD): CADD provides a quantitative measure of the deleteriousness of genetic variants by integrating multiple annotations into a single score, facilitating the ranking of variants according to their predicted functional impact. This analysis utilized data from the Ensembl database (http://useast.ensembl.org/index.html) (24, 28). (iv) Sorting intolerant from tolerant (SIFT) prediction: SIFT predicts whether amino acid substitutions resulting from coding variants are likely to affect protein function deleteriously (29). Variants predicted to be damaging were scored one point, reflecting their potential pathogenicity. (v) Polymorphism phenotyping v2 (PolyPhen-2): This tool classifies SNPs based on their predicted impact on protein structure and function, categorizing variants as “probably damaging” or “possibly damaging.” SNPs falling into these categories were assigned one point. (vi) AlphaMissense: AlphaMissense is a deep learning method that classifies missense variants into pathogenic, benign, or ambiguous categories using population data, amino acid distribution, and AlphaFold structural context. This tool integrates multiple machine learning models to generate pathogenicity score for comprehensive variant classification (30). Variants predicted to be pathogenic were assigned one point, contributing to the overall risk assessment scoring system.
The functional characterization of gene variants using SIFT and PolyPhen-2 was conducted through the SNP-Nexus platform (https://www.snp-nexus.org/v4/), which integrates multiple bioinformatics resources for variant annotation. All databases and annotation tools were accessed on January 15, 2025, except AlphaMissense (accessed September 7, 2025), ensuring the use of the most current genomic data and annotations available at the time of analysis. Schematic workflow diagrams presented in Figure 1 were created using BioRender Software (biorender.com, Toronto, Canada) under academic license.
Schematic workflow illustrating the identification of biologically relevant genes associated with BCa. In this study, the prioritization of SNPs was conducted utilizing data from the GWAS Catalog, HaploReg v4.2, and Ensembl databases.
DrugBank-based target gene identification. Breast cancer risk-associated genes were systematically mapped to the DrugBank database (version released on June 6, 2025) to identify potential therapeutic candidates. DrugBank is a comprehensive online bioinformatics and cheminformatics platform that integrates detailed information on drugs and their molecular targets (31). The search was refined using specific criteria, including pharmacologically active compounds with documented human relevance, approval status, clinical trial involvement, or classification as experimental drugs.
Results
This study aimed to identify single nucleotide polymorphisms (SNPs) associated with BCa by querying the Genome-Wide Association Study (GWAS) Catalog database, a comprehensive repository of SNP-trait associations (accessed on January 15, 2025). The initial database search yielded 2,454 SNPs reported to be associated with BCa. To refine this dataset and focus on variants with a higher likelihood of functional significance, a stringent statistical threshold was applied, retaining only SNPs with a p-value <10−8. This filtering step reduced the dataset to 1,219 SNPs. Subsequent removal of duplicate entries resulted in a refined list of SNPs, from which 14 unique SNPs exhibiting missense variants changes in the DNA sequence that lead to amino acid substitutions in the encoded protein were identified. To further characterize the potential biological impact of these missense SNPs, a multifaceted functional analysis was conducted. This analysis integrated information from missense mutation predictions, cis-eQTL mapping, CADD scores, SIFT predictions, PolyPhen-2 scores and AlphaMissense. Based on the cumulative scores derived from these analyses, the SNPs were categorized into three distinct groups, as detailed in Table I. This comprehensive approach allowed for the prioritization of SNPs with the greatest potential to influence BCa risk and progression through alterations in protein structure, gene expression, and functional consequences.
Biological annotation priority for breast cancer candidate genes.
Table I shows the prioritization of SNPs associated with BCa using a comprehensive integrative scoring system. This system combines six distinct functional annotations to evaluate the potential biological relevance of each SNP. The annotations include: (1) identification of missense mutations, which result in amino acid changes that may affect protein function; (2) cis-eQTL effects, indicating SNPs that influence gene expression levels in nearby genes; (3) CADD scores, which predict the deleteriousness of variants by integrating multiple genomic features; (4) SIFT predictions, which assess the impact of amino acid substitutions on protein function based on sequence homology; (5) PolyPhen-2 predictions, which evaluate the possible damaging effects of missense mutations on protein structure and function; and (6) AlphaMissense predictions, which classify missense variants using deep learning models that integrate population data, amino acid distribution and AlphaFold structural context. Each SNP was assigned one point for each annotation in which it met predefined significance criteria, resulting in a cumulative score that reflects the overall potential biological impact of the variant. Higher scores indicate SNPs with stronger functional evidence and greater likelihood of contributing to BCa pathogenesis. This integrative approach enables the prioritization of candidate SNPs for further experimental validation and potential clinical relevance in BCa research.
Among the identified candidates, multiple SNPs achieved the highest total score of 4, including SLCO1B1 (rs4149056) and ARHGEF38 (rs61751053), EXO1 (rs4149909), KDELC2 (rs74911261) MAPT (rs63750417), PHLDA3 (rs35383942), AKAP9 (rs6964587), ATXN7 (rs1053338) and DCLRE1B (rs11552449). These variants reflecting strong evidence from their missense mutation status, deleterious predictions by multiple computational tools including SIFT, PolyPhen-2 and AlphaMissense. Other genes scored between 1-3 points indicating varying levels of functional support. The Scoring framework was based on six sets of functional annotations, each contributing essential insights into variant pathogenicity.
Missense mutations: The high scores of our top candidates are largely attributed to their missense mutation status. The consistent presence of missense mutations among prioritized SNPs underscores their potential to alter protein structure and function, which may contribute to BCa development by disrupting critical cellular processes. Beyond structural effect, these variants also demonstrate regulatory functional.
Cis-eQTL effects: In addition to their protein-coding impact, variants showing cis-eQTL activity, such as those in PHLDA3, AKAP9, ATXN7, DCLRE1B, CASP8, MTMR11, and CCDC170 suggest a regulatory role in modulating gene expression within tissues relevant to BCa, potentially influencing pathways involved in tumor initiation and progression. These regulatory effects are further validated by computational prediction tools.
Functional prediction tools (SIFT and PolyPhen-2): Supporting the missense and regulatory findings, both tools assess the potential harmfulness of amino acid substitutions, with concordant high scores indicating a strong likelihood that these variants impair protein function, thereby reinforcing their biological significance. Integrative scoring approaches showed different patterns.
CADD scores: Despite strong evidence from individual tools, although none of the top-ranking SNPs surpassed the CADD pathogenicity threshold in this analysis, CADD remains an important integrative tool that synthesizes multiple annotations to evaluate variant impact. The lack of high CADD scores may be due to variant-specific characteristics or current limitations in the scoring algorithms. AlphaMissense provided pathogenic predictions for most top-ranking variants.
This integrative methodology successfully refines the selection of SNPs most likely to have functional significance in BCa, emphasizing variants that may affect both protein function and gene regulation. The nine variants achieving the maximum score of 4 represent the most promising candidates for experimental validation, each demonstrating consistent evidence across multiple prediction algorithms. These findings provide a foundation for future mechanistic studies and may contribute to biomarker discovery and therapeutic target identification in BCa research.
Expression analysis of SLCO1B1 across human tissues. To investigate the tissue-specific expression patterns of the solute carrier organic anion transporter family member 1B1 (SLCO1B1) gene (ENSG00000134538.3), we utilized data from the GTEx Portal, a comprehensive resource for gene expression profiling across a wide range of human tissues (https://gtexportal.org/home/; accessed 27/02/2025). Figure 2 displays a violin plot summarizing the bulk tissue expression levels of SLCO1B1, measured as transcripts per million (TPM) values, in several representative human tissues. The plot highlights both the distribution and intensity of gene expression across samples for each tissue type. SLCO1B1 expression across human tissues was evaluated using data from the GTEx Portal (accessed 27/02/2025), which provides comprehensive gene expression profiles. Analysis revealed that SLCO1B1 is predominantly and highly expressed in the liver, consistent with its role as a liver-specific organic anion transporter involved in the hepatic uptake and metabolism of endogenous compounds and drugs such as statins. Expression levels in other tissues, including testis, tibial nerve, cervix, and minor salivary gland, were substantially lower, indicating tissue-specific expression. This liver-specific pattern suggests that genetic variants in SLCO1B1 may have significant functional consequences in hepatic physiology and pathology, including BCa metastasis to the liver. Supporting evidence from multiple studies shows that SLCO1B1 expression is influenced by genetic variants, notably the c.388A>G missense variant, which is associated with increased protein expression and altered drug pharmacokinetics, such as atorvastatin clearance. These findings emphasize the importance of SLCO1B1 in liver function and its potential impact on disease processes and therapeutic responses. The GTEx-based analysis confirms that SLCO1B1 is highly and specifically expressed in the liver, with minimal expression in other tissues. This tissue-specific profile underscores the importance of considering organ context when interpreting the functional impact of SLCO1B1 variants in BCa research and therapy.
Tissue-specific expression profile of SLCO1B1 gene across human tissues. Violin plots display the distribution of SLCO1B1 expression levels across multiple tissue types, with data expressed as transcripts per million (TPM) values. The analysis revealed SLCO1B1 highest expression was present in liver tissue and minimal expression in testis, nerve-tibial, cervix-endocervix and minor salivary gland tissues. Data source: Genotype-Tissue Expression (GTEx) Portal V10 accessed on 27/02/2025.
Expression analysis of ARHGEF38 across human tissues. Figure 3 shows a violin plot summarizing the tissue expression levels of Rho Guanine Nucleotide Exchange Factor 38 (ARHGEF38), measured as TPM values, in five representative human tissues: minor salivary glands, prostate, breast (mammary tissue), pancreas, and pituitary gland. This plot highlights the distribution and intensity of gene expression in various samples for each tissue type based on data from the GTEx Portal. The analysis results show that ARHGEF38 is expressed at the highest level in the minor salivary glands, followed sequentially by the prostate, pancreas, breast, and pituitary gland. These data indicate a tissue-specific expression pattern for the ARHGEF38 gene, with varying expression levels in each tissue type studied. Various supporting studies have shown that the rs61751053 genetic variant can alter ARHGEF38 protein function and Rho GTPase activity, as predicted by the SIFT and PolyPhen-2 algorithms.
Tissue-specific expression profile of ARHGEF38 gene across human tissues. Violin plots illustrate the distribution of ARHGEF38 expression levels across multiple tissue types, with data expressed as transcripts per million (TPM) values. The analysis revealed elevated ARHGEF38 expression in minor salivary gland, prostate, breast-mammary tissues, pancreas and pituitary tissues. Data source: Genotype-Tissue Expression (GTEx) Portal V10 accessed on 27/02/2025.
Global allele frequency distribution of rs4149056 (SLCO1B1) and rs61751053 (ARHGEF38). Figure 4 and Table II illustrate the global distribution of allele frequencies for two SNPs, rs4149056 located in the SLCO1B1 gene and rs61751053 within the ARHGEF38 gene, across diverse continental populations including Admixed American (AMR), Europe (EUR), Africa (AFR), East Asia (EAS), and South Asia (SAS). Each population’s allele composition is represented by pie charts, with rs4149056 alleles depicted in red and rs61751053 alleles in blue. For rs4149056 (SLCO1B1), the T allele predominates globally but exhibits notable variation in frequency among populations. African populations show an overwhelming majority of the T allele at 99%, with the alternative C allele present at a minimal 1%. In contrast, European populations display a lower T allele frequency of 84%, accompanied by a higher C allele frequency of 16%. East Asian populations have a T allele frequency of 88%, with the C allele at 12%, while South Asian populations demonstrate a T allele frequency of 96% and a C allele frequency of 4%. The American population exhibits an intermediate distribution with 87% T allele and 13% C allele frequencies. This variability in allele frequencies indicates population-specific genetic diversity at this locus, which may influence phenotypic traits such as drug metabolism and disease susceptibility, given SLCO1B1 role in hepatic transport of endogenous and exogenous compounds.
Distribution of the rs4149056 allele of the SLCO1B1 gene and the rs61751053 allele of the ARHGEF38 gene affecting different populations across continents. Data were obtained from Ensembl 113 on 27/02/2025 and processed with BioRender.
Allele frequencies of rs4149056 and rs61751053 in various populations across continents, obtained from Ensembl on 27/02/2025.
In contrast, rs61751053 (ARHGEF38) shows minimal genetic variation across populations, with the C allele almost fixed at a frequency of 99-100% in all populations studied. The alternative T allele is only found in African populations at a low frequency of 1%, while it is completely absent (0%) in Mixed American, European, East Asian, and South Asian populations, where the C allele reaches 100% fixation. This limited pattern of allelic variation indicates that rs61751053 is highly conserved globally, with population-specific variation limited primarily to African populations. The near-fixation of the C allele in most populations may imply that the functional consequences associated with this SNP may have relatively uniform effects, while the specific presence of the T allele in African populations may contribute to population-specific patterns of genetic susceptibility. The observed differences in allele frequency distribution between these two SNPs underscore the importance of considering population genetics in genetic association studies and personalized medicine. Variants such as rs4149056 with significant frequency differences may contribute to population-specific disease risk or therapeutic response, whereas variants such as rs61751053 with minimal variation may have more uniform effects across most populations, with population-specific effects limited to African descent groups.
Drug-gene interaction analysis via DrugBank. Among the ten identified BCa risk genes, only Microtubule-Associated Protein Tau (MAPT) was found to be druggable according to DrugBank data. Three drugs are associated with MAPT: paclitaxel and docetaxel, well-known chemotherapeutic agents that function by stabilizing microtubules and are commonly used to treat breast and other cancers; and flortaucipir F-18, a radiolabeled compound developed primarily for PET imaging of tau pathology in neuro-degenerative diseases. Importantly, paclitaxel and docetaxel do not directly target MAPT, but their mechanism of microtubule stabilization is linked to MAPT’s biological role. These drug associations highlight MAPT’s potential relevance in both therapeutic and diagnostic contexts, underscoring its multifaceted function in cancer treatment strategies.
Discussion
The integrative prioritization analysis (Table I) highlighted nine genes achieving the maximum score of 4, as the most significant genetic contributors to BCa development: SLCO1B1 (rs4149056), ARHGEF38 (rs61751053), EXO1 (rs4149909), KDELC2 (rs74911261) MAPT (rs63750417), PHLDA3 (rs35383942), AKAP9 (rs6964587), ATXN7 (rs1053338) and DCLRE1B (rs11552449). These variants demonstrated consistent evidence across multiple functional annotations such as missense mutations, and pathogenicity predictions by SIFT, PolyPhen-2 and Alphamissense. SLCO1B1 is known for its role in hepatic transport of estrogen metabolites and drugs, influencing BCa risk and treatment response, including aromatase inhibitor efficacy and statin pharmacokinetics (32-34). Recent pharmacogenetic studies have further characterized the impact of SLCO1B1 variants on drug toxicities and treatment outcomes in diverse populations (35). ARHGEF38, a regulator of cytoskeletal dynamics, may affect tumor progression through cell signaling pathways, although its direct role in BCa requires further investigation. Additional genes like EXO1, KDELC2, MAPT, and PHLDA3 showed moderate evidence, emphasizing the complexity of BCa genetics. These findings underscore the value of combining genetic and functional data to identify candidate variants for future validation and therapeutic targeting.
SLCO1B1 encodes the organic anion transporter OATP1B1, which facilitates hepatic uptake of endogenous compounds (e.g., estrogen conjugates) and drugs such as statins and chemotherapeutic agents. The rs4149056 variant (c.521T>C, p.V174A) reduces OATP1B1 transport activity, leading to elevated systemic concentrations of substrates, including estrogen metabolites linked to BCa risk (36). This SNP has been associated with increased BCa susceptibility in estrogen-progestin therapy use (33). Moreover, recent studies have reported associations between ABCC2 variants and nab-paclitaxel-induced peripheral neuropathy, highlighting the importance of transporter gene polymorphisms in chemotherapy-related toxicities that may relate to SLCO1B1 functions (37). The variant causes altered pharmacokinetics of chemotherapeutics (e.g., methotrexate) and hormonal agents (38), and it also results in higher plasma estrogen levels in postmenopausal women, potentially influencing BCa progression and treatment resistance (32, 39). While this variant did not show cis-eQTL activity in our analysis, further investigation is needed to understand its regulatory mechanisms. In terms of tissue expression SLCO1B1 shows highest expression in liver tissue (Figure 2) which aligns with its role in hepatic estrogen metabolism, suggesting that dysregulated transport of estrogen conjugates may contribute to BCa pathogenesis (33, 36).
ARHGEF38, a Rho guanine nucleotide exchange factor, regulates cytoskeletal dynamics and cell signaling. While direct evidence in BCa is limited, its high prioritization score suggests potential roles in tumor cell migration, invasion, or metastasis. Like SLCO1B1, this variant achieved a high score through missense mutation status and consistent pathogenicity predictions across computational tools rather than regulatory effects. However, based on our analysis, this variant did not demonstrate cis-eQTL activity, suggesting its primary impact is through protein function alteration rather than gene expression regulation. Recent research has emphasized the complex genetic landscape of hereditary cancers, reinforcing the importance of further investigation into genes like ARHGEF38 that may contribute to tumor progression pathways (40). Several other high-scoring genes such as EXO1, KDELC2, MAPT, PHLDA3, AKAP9, ATXN7, and DCLRE1B demonstrated substantial functional evidence supported by missense mutations and pathogenicity predictions. Among these PHLDA3, AKAP9, ATXN7, and DCLRE1B showed cis-eQTL activity in BCa. MAPT has been implicated in hormone receptor signaling and chemotherapy resistance, though its direct association with BCa requires further validation. Lower-scoring genes include MN1 (rs45589338) with a score of 3, and genes with scores like DCLRE1B exhibited cis-eQTL effects without PolyPhen-2-predicted pathogenicity, indicating potential regulatory rather than protein-disrupting roles. This highlights the genes scoring ≤2 points such as CASP8, MTMR11, CCDC170 and MCM8. Some of lower scoring genes including CASP8, MTMR11, CCDC170 exhibited cis-eQTL effect without consistent pathogenicity predictions across all tools excluding missense mutation. This highlights the complexity of BCa genetics, where variants may influence disease through multiple mechanisms. These insights align with current perspectives on BCa complexity, which underscores the multifaceted ecosystem of breast cancer etiology and treatment challenges (35). The nine variants achieving the maximum score of 4 represent the most promising candidates for experimental validations, each demonstrating consistent evidence across multiple prediction algorithms and providing a foundation for future mechanistic studies in BCa research.
Extensive research has demonstrated that SLCO1B1 is highly expressed in liver tissue as shown in Figure 2. significantly more than in other tissues such as breast, lung, and thyroid (11, 36, 41, 42). This elevated hepatic expression is particularly relevant given the liver’s role as a common site of BCa metastasis, ranking third after lymph nodes and bones. The liver’s unique sinusoidal vascular structure and supportive microenvironment facilitate cancer cell colonization and growth (43). Interactions between metastatic BCa cells and hepatocytes involve specialized adhesion mechanisms, with hepatocytes secreting growth factors like IGF-1 and HGF that promote tumor progression (11). Clinically, patients with liver metastases often suffer from impaired liver function due to cancer burden, which can be life-threatening. The high expression of SLCO1B1 in the liver underscores its potential role in mediating metastatic processes and influencing disease outcomes.
In contrast, ARHGEF38 (rs61751053) ranked third after the minor salivary glands and prostate in terms of expression levels based on median TPM values in mammary breast tissue, which is consistent with its presumed involvement in the pathogenesis of BCa. Breast cancer predominantly arises as ductal carcinoma, with infiltrating ductal carcinoma accounting for 65% to 80% of invasive breast lesions, while invasive lobular carcinomas represent about 5% to 10% of the disease. Both ductal and lobular carcinomas can present as in situ or invasive forms (44, 45). SLCO1B1 encodes the organic anion-transporting polypeptide OATP1B1, a transporter critical for hepatic uptake of endogenous substances such as estrogen conjugates and various drugs, including chemotherapeutic agents (32, 36). The rs4149056 variant (c.521T>C, p.V174A) reduces transporter activity, resulting in increased plasma concentrations of substrates like statins and methotrexate, which has been linked to altered drug efficacy and toxicity profiles (36). Furthermore, this variant has been associated with increased BCa risk, especially in estrogen-progestin therapy users, and influences estrogen metabolism, affecting aromatase inhibitor treatment outcomes in estrogen receptor-positive BCa patients (32, 36, 46, 47). The pharmacogenetic impact of SLCO1B1 polymorphisms extends to bleeding risk in patients on anticoagulants, highlighting the gene’s broad clinical relevance (47). Recent studies have reported a high frequency of BRCA2 pathogenic variants in certain Japanese populations, emphasizing the importance of population-specific genetics that may also apply to SLCO1B1 and other BCa-associated genes in pharmacogenomic contexts (48). In summary, the integrative functional annotation approach effectively prioritized SLCO1B1 and ARHGEF38 as key genetic contributors to BCa, supported by their tissue-specific expression profiles and functional impacts. These findings provide a foundation for further mechanistic studies and may inform personalized therapeutic strategies targeting these genes in BCa management.
The allele frequency distribution of the two-breast cancer (BCa)-associated single nucleotide polymorphisms (SNPs), SLCO1B1 (rs4149056) and ARHGEF38 (rs61751053), varies significantly across global populations, as illustrated in Figure 4. This geographic variability highlights the importance of population-specific genetic backgrounds in understanding disease susceptibility and progression. For SLCO1B1 (rs4149056), the frequency of the risk-associated C allele differs markedly among populations, with the highest prevalence observed in Europeans at 16%, followed by the American population at 13%, East Asians at 12%, South Asians at 4%, and the lowest frequency in Africans at just 1%. This gradient suggests that genetic predisposition linked to this variant may contribute differentially to BCa risk and metastatic patterns across populations. In the United States alone, over 168,000 women were living with metastatic BCa in 2020, with stage IV patients exhibiting distinct patterns of distant metastasis: bone being the most common site (68.8%), followed by lung (16.0%), liver (13.3%), and brain (1.9%) (49). The relatively high frequency of the C allele in the American population corresponds with the notable incidence of metastatic BCa, implying a potential genetic influence on disease progression and metastatic behavior within this demographic. Conversely, the ARHGEF38 (rs61751053) variant shows a contrasting allele distribution pattern. The T allele is rare, with the highest frequency detected only in African populations at approximately 1%, whereas other major populations, including those from the Americas, East Asia, Europe, and South Asia, exhibit near fixation of the C allele, approaching 100%. This lack of variation outside Africa suggests that the ARHGEF38 SNP may have limited population-specific impact globally, though its functional significance within African populations warrants further investigation. These findings underscore the critical role of population genetics in BCa research and highlight the necessity of incorporating diverse ancestral backgrounds in genetic association studies.
Table II details allele frequencies of rs4149056 (SLCO1B1) and rs61751053 (ARHGEF38) across populations. For SLCO1B1, the risk-associated C allele shows highest frequency in Europeans (16%), followed by Admixed American (13%), East Asia (12%), South Asia (4%) and lowest in Africans (1%), while the T allele shows frequencies of 84%, 87%, 88%, 96% and 99% respectively in the same populations. The ARHGEF38 T allele is rare globally, present only in African populations at 1% (with C allele at 99%), while completely absent in other major populations where the C allele is fixed at 100% (Admixed American, East Asia, Europe and South Asia). These variations suggest population-specific genetic susceptibility to BCa. Epidemiological data (Figure 5) shows Asia has the highest BCa incidence (42.9%), mortality (47.3%), and prevalence (39.1%), followed by Europe. Africa, despite lower incidence (8.6%), exhibits a disproportionately high mortality-to-incidence ratio (13.7%), indicating greater disease severity. This disparity is potentially attributable to healthcare access limitations and advanced stage presentation, as more than 70% of BCa cases in sub-Saharan (Nigeria) are diagnosed at stages III-IV compared to ≤46% in European settings (50). These findings highlight the interplay of genetic diversity and regional factors in global BCa disparities.
Breast cancer incidence, mortality and 5-year prevalence by continent in 2020. Pie charts show continental distribution based on GLOBOCAN 2020 estimates from the International Agency for Research on Cancer (IARC) Global Cancer Observatory. Asia accounts for the largest proportion across all metrics (42.9% incidence, 47.3% mortality, 39.1% prevalence), followed by Europe and Northern America. Based on data available at https://gco.iarc.fr/.
The observed variation in allele frequencies of breast cancer-associated SNPs such as SLCO1B1 (rs4149056) and ARHGEF38 (rs61751053) across different populations underscores the significant role of genetic diversity in influencing BCa susceptibility and epidemiology worldwide. These differences may partly explain the disparities in BCa incidence, mortality, and disease severity seen among ethnic groups and geographic regions (51-53). For example, the higher frequency of the SLCO1B1 risk allele in European and American populations may correlate with increased metastatic BCa prevalence, while the near fixation of the ARHGEF38 allele in most populations except Africa suggests population-specific genetic effects. Understanding these allele distributions can inform personalized risk assessment, screening strategies, and therapeutic approaches tailored to diverse populations, ultimately improving BCa outcomes globally (54, 55).
From a therapeutic standpoint, our analysis of drug-gene interactions using DrugBank identified MAPT as the only druggable gene among the prioritized candidates. Three drugs are associated with MAPT: paclitaxel and docetaxel, well-known chemotherapeutic agents that function by stabilizing microtubules and are commonly used to treat breast and other cancers; and flortaucipir F-18, a radiolabeled compound developed primarily for PET imaging of tau pathology. Importantly, paclitaxel and docetaxel do not directly target MAPT, but their mechanism of microtubule stabilization is indirectly linked to MAPT’s biological role. MAPT’s ability to compete with taxanes for microtubule binding sites suggests that patients exhibiting different MAPT expression levels might benefit from personalized dosing strategies or alternative therapies targeting microtubules. Recent computational studies further reinforce the potential of MAPT inhibition as an effective anticancer strategy, advocating for combination treatments that include MAPT-targeting agents alongside standard therapies to overcome chemotherapy resistance (56).
This therapeutic promise is especially relevant in the context of population-specific medicine, where variations in pharmacokinetics and pharmacodynamics across ethnic groups highlight the need for individualized treatment plans that consider both genetic backgrounds and pharmacogenomic profiles.
This study integrates genetic variant data with epidemiological statistics, providing a comprehensive view of how allele frequency variations correspond with BCa incidence and mortality across continents. Utilizing large-scale databases such as the 1000 Genomes Project and WHO epidemiological data enhances the robustness and generalizability of the findings. The focus on multiple populations addresses a critical gap in BCa genetics research, which has historically been Eurocentric, thereby contributing valuable insights into genetic risk factors in underrepresented groups (53, 57). Moreover, the combined analysis of functional annotation and population genetics strengthens the biological plausibility of the prioritized SNPs as contributors to BCa susceptibility.
Despite these strengths, several limitations must be acknowledged. The allele frequency data are derived from population-level databases that may not capture intrapopulation heterogeneity or rare variants with significant effects (55). The study’s reliance on correlation between allele frequencies and epidemiological data does not establish causation; environmental, lifestyle, and healthcare access factors also critically influence BCa outcomes and may confound genetic associations (51, 52). Additionally, the functional impact of some SNPs, particularly in non-coding regions or with modest effect sizes, remains uncertain without experimental validation. The underrepresentation of certain populations, such as those from Africa and indigenous groups, limits the comprehensiveness of the genetic risk assessment and may bias conclusions (54, 57). Finally, polygenic risk scores and gene–environment interactions, which are important in BCa risk prediction, were not fully addressed in this analysis (54).
Conclusion
This study highlighted nine genetic variants achieving the maximum score of 4 as potentially involved in BCa development: rs4149056 in the SLCO1B1 gene and rs61751053 in the ARHGEF38 gene, along with variants EXO1 (rs4149909), KDELC2 (rs74911261) MAPT (rs63750417), PHLDA3 (rs35383942), AKAP9 (rs6964587), ATXN7 (rs1053338) and DCLRE1B (rs11552449). The allele frequencies of these variants vary across populations and align with global BCa epidemiology. Specifically, the C allele of rs4149056 is most prevalent in European populations (16%), while the T allele of rs61751053 is most common in African populations (1%). Notably, Africa shows the highest mortality-to-incidence ratio for BCa. These results emphasize the need to incorporate population-specific genetic differences into BCa prevention, screening, and treatment strategies. Additionally, MAPT emerged as the only druggable target among the prioritized candidates with existing therapeutic connections to paclitaxel and docetaxel, though these represent indirect interactions through microtubule stabilization mechanisms rather than. Direct drug targets, highlighting promising opportunities for drug repurposing in BCa treatment. However, further clinical studies are required to validate the functional significance of these variants for their use in precision medicine.
Acknowledgements
None.
Footnotes
Authors’ Contributions
Putri Permata Suka Admanegara and Lalu Muhammad Irham conceived and designed the study. Putri Permata Suka Admanegara and Lalu Muhammad Irham performed the computational analysis. Putri Permata Suka Admanegara, Rista Yulianti, Desti Rahmawati, Siska Widiastuti, Wirawan Adikusuma, Brilliant Citra Wirashada, Danang Prasetyaning Amukti, Darmawi Darmawi, Baik Heni Rispawati, Benni Iskandar. Rockie Chong, Ates Ilker, and Lalu Muhammad Irham wrote the manuscript. Lalu Muhammad Irham provided the funding. Putri Permata Suka Admanegara, Rista Yulianti, Desti Rahmawati, Siska Widiastuti, Wirawan Adikusuma, Brilliant Citra Wirashada, Danang Prasetyaning Amukti, Darmawi Darmawi, Baik Heni Rispawati, Benni Iskandar, Rockie Chong, Ates Ilker, and Lalu Muhammad Irham revised the manuscript. All Authors have read and approved the manuscript and have made significant contributions to this study.
Supplementary Material
Supplementary materials are available online at Zenodo: https://doi.org/10.5281/zenodo.17072325
Conflicts of Interest
The Authors disclose no conflicts of interest.
Funding
We would like to thank the Directorate of Research and Community Service, Directorate General of Research and Development, Ministry of Higher Education, Science, and Technology of the Republic of Indonesia for their support of this study under grant numbers 126/C3/DT.05.00/PL/2025, 0498.12/ll5-INT/AL.04/2025, 72/PTM/LPPM.UAD/V/2025.
Artificial Intelligence (AI) Disclosure
During the preparation of this manuscript, DeepL (DeepL SE, Cologne, Germany) was used solely for language translation and stylistic improvements in select paragraphs. No sections were generated by AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.
- Received August 5, 2025.
- Revision received September 8, 2025.
- Accepted September 22, 2025.
- Copyright © 2025 The Author(s). Published by the International Institute of Anticancer Research.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.