Abstract
In the UK, neoadjuvant chemotherapy (NAC) for locally advanced esophageal adenocarcinoma (EAC) is the standard of care. Unfortunately, response to NAC, following surgery is often low (<30%) and survival benefit at 2 years is only 5.1%. The EAC genome is complex and heterogeneous between patients, where specific mutagenic processes and mutations may result in chemotherapy resistance. Here we report preliminary results from whole-genome sequencing (WGS) of 48 patients who were subsequently treated with NAC.
Methods
We defined response as Mandard Tumor Regression Grade (TRG) 1-2 (n = 15) and non-response as TRG4-5 (n = 33). WGS of 50x coverage and calling of genomic events (Strelka2, SCAT/GISTIC) was performed according to Frankell Nat Genet 2019 with modifications, to differentiate driver from passenger mutations (dNdScv, oncodriveCLUST), determine variant allele frequencies (VAF; copy number-adjusted) and mutation signatures (NMF R-package). Following stringent variant QC (Phred score ≥ 30), filtering (VAF >0.02) and annotation (ANNOVAR, cancer census/helper, the 77 known EAC drivers and false positive genes) and visualisation in maftools, we evaluated the data for associations between genomic events and response to NAC.
Results
COSMIC mutation signatures 2 (TRG1-2; Cosθ = 0.89) and 6 (TRG4-5; Cosθ = 0.82) were enriched, suggesting defective mismatch repair in non-responders. Using 5,193 high-confidence non-silent mutations (TRG1-2,n = 1969; TRG4-5, n = 3224), we identified 39 mutated driver genes (Figure-1) with a median of 2.9(0-5) (TRG1-2) and 2.7(0-9) (TRG4-5) driver events/patient. Shared dysregulated pathways, included DNA-damage (TP53), cell cycle (CDKN2A), TGF-β (SMAD4) and chromatin regulation (ARID1A). There was no difference in the prognostic of SMAD4 and GATA4 genes. Interestingly, NAV3 mutations were only present in non-responders (21%,7/33); and in responders TP53 incidence was higher (93% vs. 58%) with a concomitant reduction in clonal cell fraction (0.25 vs. 0.39;Wilcoxon, p < 0.0001).
Conclusion
The data suggest that specific genomic events, such as NAV3 mutation and the intra-tumoral heterogeneity of mutated DNA-damage response genes may offer additional predictive value. Ongoing work includes analysis of the impact of non-coding variation on response. While our analysis is limited by sample size and requires validation in additional cohorts, it demonstrates the potential of genomic sequencing to identify NAC response biomarkers and may guide alternative or novel treatment modalities for chemo-resistant tumours.