Frontiers in Oncology (Sep 2021)

Construction and Validation of a Platinum Sensitivity Predictive Model With Multiple Genomic Variations for Epithelial Ovarian Cancer

  • Hong Zheng,
  • Tong Shu,
  • Shan Zhu,
  • Chao Zhang,
  • Min Gao,
  • Nan Zhang,
  • Hongguo Wang,
  • Jie Yuan,
  • Zaixian Tai,
  • Xuefeng Xia,
  • Yuting Yi,
  • Yuting Yi,
  • Jin Li,
  • Yanfang Guan,
  • Yanfang Guan,
  • Yang Xiang,
  • Yunong Gao

DOI
https://doi.org/10.3389/fonc.2021.725264
Journal volume & issue
Vol. 11

Abstract

Read online

Platinum-based chemotherapy is still the standard of care after cytoreductive surgery in the first-line treatment for epithelial ovarian cancer. This study aims to integrate novel biomarkers for predicting platinum sensitivity in EOC after initial cytoreductive surgery precisely. To this end, 60 patients were recruited from September 2014 to October 2019. Based on the duration of progress-free survival, 44 and 16 patients were assigned to platinum-sensitive and platinum-resistant group, respectively. Next generation sequencing was performed to dissect the genomic features of ovarian tumors obtained from surgery. Multiple genomic variations were compared between two groups, including single-nucleotide variant, single base or indel signature, loss of heterozygosity (LOH), whole-genome duplication (WGD), and others. The results demonstrated that patients with characteristics including positive SBS10a signature (p < 0.05), or FAM175A LOH (p < 0.01), or negative WGD (p < 0.01) were significantly enriched in platinum-sensitive group. Consistently, patients with positive SBS10a signature (15.8 vs. 10.1 months, p < 0.05), or FAM175A LOH (16.5 vs. 9.2 months, p < 0.05), or negative WGD (16.5 vs. 9.1 months, p < 0.05) have significantly longer PFS than those without these genetic features. By integrating these three biomarkers, a lasso regression model was employed to train and test for all patients, with the AUC value 0.864 in platinum sensitivity prediction. Notably, 388 ovarian cancer patients from TCGA dataset were leveraged as independent validation cohort with AUC value 0.808, suggesting the favorable performance and reliability of this model.

Keywords