npj Precision Oncology (Nov 2023)

In situ profiling reveals metabolic alterations in the tumor microenvironment of ovarian cancer after chemotherapy

  • Sara Corvigno,
  • Sunil Badal,
  • Meredith L. Spradlin,
  • Michael Keating,
  • Igor Pereira,
  • Elaine Stur,
  • Emine Bayraktar,
  • Katherine I. Foster,
  • Nicholas W. Bateman,
  • Waleed Barakat,
  • Kathleen M. Darcy,
  • Thomas P. Conrads,
  • G. Larry Maxwell,
  • Philip L. Lorenzi,
  • Susan K. Lutgendorf,
  • Yunfei Wen,
  • Li Zhao,
  • Premal H. Thaker,
  • Michael J. Goodheart,
  • Jinsong Liu,
  • Nicole Fleming,
  • Sanghoon Lee,
  • Livia S. Eberlin,
  • Anil K. Sood

DOI
https://doi.org/10.1038/s41698-023-00454-0
Journal volume & issue
Vol. 7, no. 1
pp. 1 – 26

Abstract

Read online

Abstract In this study, we investigated the metabolic alterations associated with clinical response to chemotherapy in patients with ovarian cancer. Pre- and post-neoadjuvant chemotherapy (NACT) tissues from patients with high-grade serous ovarian cancer (HGSC) who had poor response (PR) or excellent response (ER) to NACT were examined. Desorption electrospray ionization mass spectrometry (DESI-MS) was performed on sections of HGSC tissues collected according to a rigorous laparoscopic triage algorithm. Quantitative MS-based proteomics and phosphoproteomics were performed on a subgroup of pre-NACT samples. Highly abundant metabolites in the pre-NACT PR tumors were related to pyrimidine metabolism in the epithelial regions and oxygen-dependent proline hydroxylation of hypoxia-inducible factor alpha in the stromal regions. Metabolites more abundant in the epithelial regions of post-NACT PR tumors were involved in the metabolism of nucleotides, and metabolites more abundant in the stromal regions of post-NACT PR tumors were related to aspartate and asparagine metabolism, phenylalanine and tyrosine metabolism, nucleotide biosynthesis, and the urea cycle. A predictive model built on ions with differential abundances allowed the classification of patients’ tumor responses as ER or PR with 75% accuracy (10-fold cross-validation ridge regression model). These findings offer new insights related to differential responses to chemotherapy and could lead to novel actionable targets.