← Back to results

Front Genet · 2026

Chemotherapy-refractory high-grade serous ovarian carcinoma: single-cell findings, TCGA-OV transcriptomic context, and a single-case MRI-transcriptome illustration.

Frontiers in genetics, 17: 1943981 PMID 42973797 DOI 10.3389/fgene.2026.1943981

Open in PubMed ↗ Open via DOI ↗

Abstract

BACKGROUNDPrimary chemotherapy refractoriness remains a clinical challenge in high-grade serous ovarian carcinoma (HGSC) and may reflect pre-existing tumor-cell states and microenvironmental heterogeneity. We examined public single-cell and bulk transcriptomic resources, a barcode-matched MRI case, and an in vitro cisplatin-perturbation RT-qPCR experiment.

METHODSWe reanalyzed GSE300897, comprising 19,156 annotated cells from nine treatment-naïve HGSC biopsy libraries (four chemotherapy-refractory and five chemotherapy-sensitive), with response summaries retained at the library/patient level. The Cancer Genome Atlas Ovarian Cancer (TCGA-OV) provided 421 unique primary tumors for cohort-scale transcriptomic and overall survival context. One TCGA-OV case (TCGA-24-1423) linked a multiparametric MRI examination to a tumor RNA-seq barcode. OVCAR-3 and Caov-3 cells were exposed to cisplatin (5 μM, 24 h), and IFIT1, MX1, SNAI1, and HIF1A expression was measured by RT-qPCR.

RESULTSAcross the nine GSE300897 libraries, no cancer-cell program passed Benjamini-Hochberg (BH) q < 0.05. Refractory libraries showed a lower type-I interferon (IFN-I) score (-0.15 log1p(CP10k)) and epithelial-to-mesenchymal transition (EMT) balance (-0.16), with a small increase in hypoxia (+0.05). In TCGA-OV, none of the six displayed program scores reached BH q < 0.05 in age- and stage-adjusted Cox models; IFN-I tertiles separated nominally by Kaplan-Meier analysis (log-rank p = 0.029), whereas continuous IFN-I was not significant after adjustment (HR = 0.96 per SD, BH q = 0.927). In TCGA-24-1423, the paired transcriptome showed an IFN-I score at the 86.8th TCGA-OV percentile, with IFIT1 and MX1 at the 87.5th and 93.7th percentiles. RT-qPCR showed lower IFIT1/MX1 and higher SNAI1/HIF1A after acute cisplatin exposure in both cell lines (all Holm-adjusted p < 0.05).

CONCLUSIONThese findings support a conservative, hypothesis-generating interpretation of chemotherapy response-associated transcriptional heterogeneity in HGSC. Reduced IFN-I and altered cell-state programs were directional features of the small refractory single-cell cohort, whereas TCGA-OV provided context rather than independent response validation. The MRI-linked case demonstrates imaging-transcriptome pairing but does not establish an imaging biomarker. The RT-qPCR experiment demonstrates cisplatin-responsive changes in selected genes but models acute drug perturbation rather than the pretreatment clinical refractory state.

Cite