Abstract 529P
Background
The choice of systemic therapies is currently based on probabilistic outcomes. For patients with gastrointestinal (GI) malignancies, a majority will not receive available standard therapies due to early progression and decline in performance status. Even when an actionable mutation exists, many patients do not benefit from it due to the presence of other cancer driver processes. To overcome these limitations, we developed a high-throughput drug screen (HTDS) to improve our ability to prioritize systemic therapy and select novel targeted therapy.
Methods
PDTOs were established from tumor tissues obtained from advanced GI cancer patients at Weill Cornell Medicine. Whole exome sequencing and RNA sequencing confirmed genomic similarities between PDTOs and tumors. PDTOs were screened for 156 drugs using a robotic HTDS system to examine drug sensitivity. PDTOs ex-vivo response to FOLFOX and FOLFIRI were compared with their matching patients’ response. Drug sensitivity was correlated with PDTO genes expression profiles, gene set enrichment scores, transcription factor activities and tumor checkpoint modules and master regulator analysis.
Results
44 PDTOs were established from 35 patients (21 CRC, 14 GEA). Drug response curves (DRC) for all 156 drugs were generated and AUC and IC50s were calculated. Overall concordance between PDTOs and patients’ response to FOLFOX and FOLFIRI was 74%. We observed 280 copy number alterations and 350 mutated genes per PDTO. Several novel therapies were identified via HTDS. Comparisons of resistant and sensitive groups using genomic data identified various genomic signatures, with varied results not necessary consistent with published drug mechanisms. Out of 106 drug targets, RAS-MAPK was upregulated in resistant PDTOs to KRAS and MEK inhibitors. PLK was upregulated for sensitive PDTOs to PLK inhibitors. DDR, HDAC inhibitors didn’t show any discernible genomic signatures.
Conclusions
Robotic HTDS can successfully predict response to standard therapy (74% concordance). This powerful technology can also identify several novel treatment options that may not be identified using genomic and transcriptomic analyses.
Legal entity responsible for the study
The authors.
Funding
Torrey Coast Grant.
Disclosure
All authors have declared no conflicts of interest.