Optimizing the program of sequential chemotherapy for advanced gastric cancer based on gene-level CNVs called from ctDNA.

Abstract

e16068 Background: Tumor evolutionary trade-offs under treatment pressure can be utilized to optimize therapy. We aimed to profile the dynamic pattern of gene-level copy number variations (CNVs) for advance gastric cancer (GCA) patients who underwent sequential chemotherapy and optimized the therapy program. Methods: 27 chemo-naive advanced GCA patients with distant metastasis were enrolled. Peripheral blood samples were collected at different time points, including pre-therapy and disease progression on a specific regimen. Gene-level CNVs were called from ctDNA of blood samples and drug modified score (DMS) was defined as the ratio of CNV value between samples of pre- and post-treatment for each gene. Positive and negative DMS of a gene represented its resistance and sensitivity to a selected regimen, respectively. Then, the best initial regimen was selected and the optimal program of sequential chemotherapy was designed for each patient. Results: DMS of gene was calculated for four chemotherapy regimens, including platinum-based doublet regimen (Pt-2d), oral-fluorouracil prodrug (O-FP), paclitaxel-based regimen (PTX) and irinotecan-based regimen (CPT-11). Based on DMS distribution, resistant and sensitive genes were determined for each regimen through a heuristic method. Functional enrichment analysis showed that resistant genes mainly involved in cellular response to oxygen-containing compound and regulation of cell proliferation, while sensitive genes mainly involved in positive regulation of macromolecule biosynthetic process and regulation of apoptotic process. Then, drug resistant score (DRS, defined as DMS weighted sum of CNV values of resistant genes) was calculated and the regimen with the lowest DRS was selected as the best initial treatment for each patient. The best initial treatment of 16 patients was CPT-11, of 8 patients was O-FP, of 2 patients was Pt-2d and of 1 patient was PTX. Following that, the optimal program of sequential chemotherapy was designed according to the percentage of shared resistant genes between regimens for each patient, where two regimens with the lowest percentage were recommended to use one by one. Finally, prognostic analysis showed that differences between the optimal and practical program of sequential chemotherapy were negatively correlated to overall survival of patients. Conclusions: Based on gene-level CNVs called from ctDNA of blood samples, we determined the best initial treatment and designed the optimal program of sequential chemotherapy for advanced GCA patients, pacing a key step toward precision medicine in GCA field.