Evolutionary Analysis of Pre- and Post-treatment Molecular Diversity in Rectal Cancer Patients Receiving Neoadjuvant Chemoradiation (CRT) and a KRAS Mutation-targeted Radiosensitizer

Abstract

Neoadjuvant CRT for locally advanced rectal cancer results in variable pathologic complete response (pCR) rates that may be in part genomically determined. Prior preclinical and clinical data established KRAS/TP53 mutations (mut) as biomarkers of radiation resistance. Lab data suggested that KRAS mut tumors may be preferentially radiosensitized by midostaurin, an oral multi-target tyrosine kinase inhibitor that is FDA-approved for FLT3-mutated leukemia. Using pre- and post-treatment samples from a phase Ib trial of midostaurin + neoadjuvant CRT (mCRT), we assessed the evolution of somatic mutations over the course of treatment and identified molecular determinants of responsiveness. Patients with locally advanced (T3/T4 or N+, M0) rectal adenocarcinoma were enrolled on an IRB-approved protocol (clinicaltrials.gov registered). Neoadjuvant fluoropyrimidine-based mCRT (50.4 Gy) was administered with midostaurin dosed according to a 3+3 dose escalation study design. The MTD for midostaurin was previously reported as 75 mg BID. Total mesorectal excision (TME) was performed at 6-9 weeks post-treatment. DNA was extracted from FFPE samples and whole exome sequencing was performed on pre/post-treatment tumor and matched germline DNA. Data were analyzed by standard mutation analysis pipelines to determine tumor purity and identify somatic mutations, insertion/deletions, and copy number variations. From 10/2011 to 11/2016, 19 patients were enrolled (median age 53 years, 84.2% female, 73.6% stage III), all of whom completed mCRT + TME. Eight patients were classified as non-responders (NR) while 11 patients were partial (n = 8) or complete responders (n = 3) (R). FFPE samples from 17 patients were successfully sequenced and passed quality control metrics (7 NR, 10 R). Mean somatic tumor mutational burden pre- and post-mCRT was 6.7 (range, 3.3-24.3) and 4.8 (range, 0.8-15.4) mutations/MB, respectively. Exploratory comparison of mutational profiles pre- and post-treatment supported varied models of evolutionary pressure and subclonal selection in response to mCRT. Unexpectedly, of 4 KRAS mut tumors (2 with concurrent TP53 mut), 3 were in the R group, and the only NR had no KRAS mut identified in the post-treatment sample, suggesting selection against KRAS mut subclones during treatment. Further molecular analyses supported broad trends in the evolution of putative molecular subclones over the course of treatment with respect to major colorectal cancer driver pathways, including RTK/RAS, PI3K/AKT, WNT/APC, TGFb/EMT. Data suggest successful clinical translation of a KRAS-targeted treatment approach. Future personalized treatment approaches may be informed by a better understanding of tumor evolution in response to treatment.