Genomic Instability Is Defined by Specific Tumor Microenvironment in Ovarian Cancer: A Subgroup Analysis of AGO OVAR 12 Trial

Abstract

Background: Following disappointing results with PD-1/PD-L1 inhibitors in ovarian cancer, it is essential to explore other immune targets. The aim of this study is to describe the tumor immune microenvironment (TME) according to genomic instability in high grade serous ovarian carcinoma (HGSOC) patients receiving primary debulking surgery followed by carboplatin-paclitaxel chemotherapy +/- nintedanib.

Methods: 103 HGSOC patients' tumor samples from phase III AGO-OVAR-12 were analyzed. A comprehensive analysis of the TME was performed by immunohistochemistry on tissue microarray. Comparative genomic hybridization was carried out to evaluate genomic instability signatures through homologous recombination deficiency (HRD) score, genomic index, and somatic copy number alterations. The relationship between genomic instability and TME was explored.

Results: Patients with high intratumoral CD3⁺ T lymphocytes had longer progression-free survival (32 vs. 19.6 months, p = 0.009) and overall survival (OS) (median not reached). High HLA-E expression on tumor cells was associated with a longer OS (median OS not reached vs. 52.9 months, p = 0.002). HRD profile was associated with high HLA-E expression on tumor cells and an improved OS. In the multivariate analysis, residual tumor, intratumoral CD3, and HLA-E on tumor cells were more predictive than other parameters.

Conclusions: Our results suggest HLA-E/CD94-NKG2A/2C is a potential immune target particularly in the HRD positive ovarian carcinoma subgroup.

Keywords: HLA-E; HRD; copy number alterations; homologous recombination deficiency; ovarian cancer; tumor immune microenvironment.

Figure 1

Intratumoral CD3 confirmed to be a major prognostic biomarker in HGSOC. Intratumoral CD3 (“CD3 tumor”) is the main prognostic biomarker of HGSOC patients’ survival. (A–C) Forest Plots of the univariate analysis showing the hazard ratio for (A) progression-free survival and (C) overall survival for each immune parameter evaluated by IHC. (B–D) Kaplan–Meier estimates for (B) progression free survival and (D) overall survival according to the intratumoral CD3 expression using the best cutoff.

Figure 2

HLA-E on tumor cells is an emergent prognostic biomarker in HGSOC. HLA-E predicts HGSOC patients’ survival. (A) Representative snapshots of HLA-E staining assessed by IHC in percentage of positive tumor cells where 0 is lower than 1%, 1 is between 1% and 5%, 2 between 5% and 50%, and 3 is higher than 50%. 0 and 1 scoring corresponded to HLA-E^low expression. 2–3 corresponded to HLA-E^high expression. (Original magnification ×10). (B) Kaplan–Meier estimates for overall survival according to the HLA-E expression using the best cutoff. (C) Bar plots showing the proportion of HLA-E expression stratified according to the platinum sensitivity (platinum sensitive vs. platinum resistant). (D) Bar plots showing the repartition of immune populations (intratumoral CD3, Foxp3, IgG, and ICOS) according to the expression of HLA-E. (E) Kaplan–Meier curves for PFS (upper panels) and OS (lower panels) according to the HLA-E and intratumoral CD3 expression. * p < 0.05, ** p < 0.005.

Figure 3

Prognostic and predictive value of genomic instability signatures. (A) Repartition of patients according to genomic scores: histograms for Focal and Arm/chromosomal SCNA scores. Donut charts showing percentages of HRD (n = 39), HRP (n = 28) or unknown (n = 36) patients and percentages of GI^high or GI^low patients that were stratified by best cutoff of 88. (B) Forest Plots showing the hazard ratio for progression free survival for each genomic signature. (C) Forest Plots showing the hazard ratio for overall survival for each genomic signature. * p < 0.05.

Figure 4

Genomic instability and tumor immune microenvironment in HGSOC. (A) Bar plots showing the proportions of HLA-E^high versus HLA-E^low patients according to HRD status (HRD versus HRP). (B) Boxplots showing the MXA score according to HRD status (HRD versus HRP). (C) Bar plots showing the proportions of patients with CD39 vessels^pos versus CD39 vessels^neg according to GI status (GI^high versus GI^low). (D,E) Boxplots showing the Focal SCNA score according to (D) intratumoral CD3 expression and (E) CD20 expression. (F–H) Boxplots showing the Arm SCNA score according to (F) CD20 expression, (G) CD163 expression, and (H) FOXP3 expression. * p < 0.05.