Dynamic Tumor Mutational Profiling Using a Novel Non-Invasive Swab Technique for Serial Whole Exome Sequencing of Cervical Tumors During Chemoradiation Therapy.

Purpose/Objective(s) Clinically relevant genetic predictors of radiation response for cervical cancer are understudied due to the morbidity of repeat invasive biopsies required to obtain genetic material. We have developed a novel noninvasive cervical swab technique to collect tumor DNA with adequate throughput to perform whole exome sequencing (WES) at serial timepoints over the course of chemoradiation therapy (CRT). We hypothesize that characterization of dynamic changes in the mutational landscape of cervical tumors during the course of CRT will allow for identification of molecular signatures correlating with therapy response. Materials/Methods A total of 69 patients with FIGO stage IB-IVA cervical cancer who underwent definitive CRT were prospectively enrolled on an IRB approved clinical trial. Cervical tumor DNA samples were collected by a noninvasive swab technique at baseline, and after 1, 3, and 5 weeks of CRT. Matched buccal DNA swabs were obtained to filter somatic mutations for each patient. An initial cohort included 24 patients (81 samples) and a second cohort included 45 patients (135 samples). WES was performed on the Illumina HiSeq 4000 series. Raw sequencing data were aligned to a human reference genome and MuTect was applied to detect somatic single nucleotide variants (SNV). Tumor purity (TP) was estimated from the distribution of variant allelic fractions of SNVs. In patients with pelvic MRIs at baseline and week 5 of CRT, 3D tumor volumes (TV) were generated (n = 6,8). Regression analysis was used to evaluate the goodness of fit between TP and TV. Results Estimated TP was highly correlated with TV in both cohorts (R2 = 0.969 and 0.858). Between the two cohorts, there was overlap in 4 of the top 5 genes (80%) containing the greatest number of exonic mutations across all timepoints (PIK3CA, MUC4, TPRXL, TTN). The number of altered genes present at baseline decreased over time in both cohorts. When normalized to TP, relative mutational burden decreased among standard and exceptional responders ( 20% TV remaining at week 5). The top ranked genes by magnitude of change at week 5 included MUC2, PARP10 and PIK3CA. Pathway analysis of these top ranked genes mapped to key radiation response pathways including regulation of the G1/S transition. Conclusion We validated this method of performing serial WES over the course of CRT in 2 independent prospective patient cohorts, highlighting the robustness of our novel noninvasive technique. This allowed for the identification of a panel of potential genes involved in CRT resistance. Future directions include examination of dynamic changes in copy number variation, indels, and tumor mutational burden towards a comprehensive analysis of clinically relevant drivers of CRT resistance. This may allow future targeted treatment for patients with a radioresistant genetic profile, or concurrent targeted therapy for tumors with actionable mutations.