Effect of genetic polymorphism on nasopharyngeal carcinoma chemoradiotherapy reaction

Nasopharyngeal carcinoma (NPC) is a malignant tumor with high geographic prevalence in parts of Asia, especially China. Radiotherapy is the mainstay treatment, with simple RT used for early-stage disease and concurrent chemoradiotherapy for intermediate/advanced stages. Despite RT’s central role, patient outcomes vary substantially and reliable pretreatment biomarkers to predict radiosensitivity are lacking, leading to risks of over- or undertreatment. Radiogenomics posits that RT efficacy is influenced not only by dose and technique but also by patients’ genetic backgrounds. Prior research has largely used candidate-gene approaches implicating variants in DNA repair and growth factor pathways (e.g., XRCC1, EGFR, rs3135001) in RT response, but comprehensive GWAS focused on NPC radiosensitivity are scarce. This study aimed to perform a two-stage GWAS to identify germline variants associated with RT response in NPC and to explore their potential functional mechanisms.

The paper situates its work within radiogenomics, highlighting that RT outcomes are modulated by genetic factors. It reviews candidate-gene associations previously linked to RT efficacy across cancers, including XRCC1 399Gln with RT resistance in esophageal cancer, EGFR R497K predicting RT outcomes in cervical cancer, and rs3135001 predicting head and neck squamous cell carcinoma radiosensitivity. It notes that head and neck RT GWAS have primarily addressed toxicity (e.g., xerostomia, mucositis, brain injury), and that NPC-focused GWAS on efficacy are lacking. This gap motivates a genome-wide approach beyond candidate genes to discover new loci relevant to NPC radiotherapy response.

Design: Two-stage GWAS of RT response in NPC. Initial cohort n=1645; after QC exclusions (n=479), 1166 patients remained: discovery cohort (n=319, Jiangxi Provincial Cancer Hospital) and two validation cohorts (n=662 Hunan Cancer Hospital; n=185 Jiangxi). Inclusion: pathologically confirmed NPC without distant metastasis or other concurrent tumors, treated primarily with IMRT; no prior surgery, targeted therapy, immunotherapy, or other antitumor therapy before RT. RT details: single dose ~2 Gy, 5 days/week; primary lesions 66–77 Gy; positive cervical nodes 54–60 Gy, over ~30–33 fractions. Chemotherapy: cisplatin-based regimens for adjuvant, neoadjuvant, or concurrent settings as indicated. Exclusions: pregnancy/breastfeeding; serious comorbidities. Outcome assessment: WHO RECIST via MRI for primary lesion and positive lymph nodes; categorized as CR, PR, SD, PD; dichotomized as CR (good) vs non-CR (PR/SD/PD). Ethics: Approved by ethics committees of Jiangxi and Hunan Cancer Hospitals; written informed consent; registered in Chinese Clinical Trial Register (ChiCTR-OPC-14005257). Genotyping and QC: Discovery genotyping with Illumina Infinium GSA-24 v1.0; initial 688,783 autosomal SNPs. Sample QC: call rate ≥95%, cryptic relatedness correction, heterozygosity within ±3 SD, sex check. SNP QC: call rate ≥95%, HWE ≥1×10−4, MAF ≥0.01, autosomal. Imputation with IMPUTE2 using 1000 Genomes phase 3 v4; post-imputation filters INFO ≥0.4, MAF ≥0.05, yielding 4,112,760 SNPs. Validation: Stage 1 genotyping with Sequenom MassARRAY; Stage 2 with TaqMan SNP assay on Roche LightCycler 480 for rs11130424; standard reaction conditions and probe sequences provided; QC included sample call rate ≥95%, SNP detection rate ≥95%, MAF ≥0.01, HWE ≥0.05, relatedness checks, heterozygosity within ±3 SD, and sex check. Statistical analysis: PCA on discovery cohort using SNPRelate; outliers removed and PCA repeated. Clinical covariates evaluated (t-tests for continuous, logistic tests for categorical) and significant factors adjusted in multivariable logistic regression GWAS under additive model, with sex, smoking, and drinking as covariates for both phenotypes (primary lesion and positive lymph node efficacy). Genomic inflation factor λ estimated via PLINK; Q-Q and Manhattan plots via R qqman. Association thresholds: variants with P<1×10−6 in discovery forwarded to validation. Bioinformatics: Regional plots with LocusZoom; LD/haplotype with PLINK and LDBlockShow; eQTL via GTEx; meQTL via Pancan-meQTL; functional mapping and chromatin interactions via FUMA; transcription regulatory context via ENCODE; population MAF via 1000 Genomes and gnomAD. Additional analyses: Stratified associations by smoking, EBV status, clinical stage, and treatment regimen.

- Study population: 1166 NPC patients analyzed in two stages (discovery n=319; validation total n=847 [662+185]). Initial enrollment n=1645 with 479 excluded after QC. - Population structure: Minimal inflation (λ=1.013 primary lesion; λ=1.011 positive lymph node); PCA showed no substantial stratification. - Discovery screen: 19 SNPs (P<1×10−6) advanced to validation; one (rs6414584) failed QC for validation genotyping. - Primary lesion efficacy: No SNP replicated in validation. - Positive lymph node efficacy: rs11130424 (intronic to CACNA2D3 at 3p21.1) replicated: • Discovery: P=8.71×10−6; OR=2.53 (95% CI: 1.68–3.81) • Validation 1: P=0.046; OR=1.26 (1.00–1.59) • Validation 2: P=1.33×10−4; OR=2.33 (1.51–3.62) • Combined: P=1.38×10−5; OR=1.70 (1.34–2.15) The efficacy (likelihood of CR) was better in carriers of the minor allele compared with major allele carriers. - Stratified analysis (positive lymph node efficacy): • EBV-positive: P=3.175×10−4; OR=2.107 (1.401–3.170) • EBV-negative: P=0.987; OR=0.998 (0.822–1.213) • Stage III+IV: P=0.0126; OR=1.837 (1.046–3.223) • Stage I+II: P=0.228; OR=1.368 (0.945–2.975) • Treatment CCRT+IC/AC: P=3.403×10−3; OR=2.214 (1.296–3.783) • Treatment CCRT alone: P=0.551; OR=1.064 (0.868–1.305) • Smoking subgroup showed an association signal (P=0.0103), while nonsmokers did not reach significance (P=0.107). - Fine-mapping/functional context: rs11130424 was not in strong LD with nearby SNPs across a ~49.41 kb block. GTEx eQTL did not show effects on CACNA2D3 expression; Pancan-meQTL indicated genotype-associated methylation differences at CpG cg04394267 in endometrial carcinoma (P=4.87×10−6). 3D chromatin interaction implicated interactions with CCDC66. The locus shows higher MAF in East Asian populations. - No genome-wide significant associations were found for primary lesion efficacy after validation.

The study addressed the need for genetic biomarkers of RT response in NPC through a two-stage GWAS, moving beyond candidate-gene approaches that had focused primarily on DNA repair and EGFR-related variants. The identification and replication of rs11130424 in CACNA2D3 for positive lymph node response suggest a germline contribution to radiosensitivity. The finding was more pronounced in clinically relevant subgroups (EBV-positive, advanced stage III/IV, and those receiving combined chemoradiotherapy with induction/adjuvant chemotherapy), indicating interaction between tumor biology, treatment intensity, and host genetics. While eQTL analyses did not show an effect on CACNA2D3 expression, meQTL and chromatin interaction data suggest potential regulatory mechanisms, including effects on methylation and long-range interactions with CCDC66. Biologically, CACNA2D3’s role in calcium influx and modulation of PI3K/AKT and MAPK signaling provides plausible pathways through which genetic variation could influence RT-induced apoptosis and therapeutic response. The results fit within the broader literature that implicates PI3K/AKT signaling in NPC progression and treatment sensitivity, offering a mechanistic rationale and potential targets for future functional validation and therapeutic modulation.

This two-stage GWAS in 1166 NPC patients identified and validated rs11130424 within CACNA2D3 as a germline variant associated with improved radiotherapy response of positive lymph nodes, with stronger effects in EBV-positive, advanced-stage patients and those receiving combined chemoradiotherapy plus induction/adjuvant chemotherapy. The locus’s functional context points to calcium signaling and PI3K/AKT and MAPK pathways, and potential interactions with CCDC66. These findings expand the genetic landscape of NPC radiosensitivity and may inform biomarker development for treatment personalization. Future work should include functional studies to define the causal mechanism, replication in independent multi-ethnic cohorts, and integration of genetic variants with clinical factors to build predictive models for individualized RT decision-making.

- Sample size, while moderate, may limit power to detect additional loci and to precisely estimate effects across subgroups. - Mechanistic link remains unresolved: rs11130424 is intronic; eQTL data did not show direct effects on CACNA2D3 expression; methylation and chromatin interaction signals require functional validation. - Potential inconsistencies in subgroup effect estimates warrant cautious interpretation; further replication is needed. - The study focused on Chinese cohorts; generalizability to other ancestries may be limited given population MAF differences.