The role of different TGFβ signatures in predicting outcome in high grade serous ovarian carcinoma.

Abstract

Listen

Translate article

e14262 Background: Ovarian cancer (OV) is the most lethal gynecologic cancer in the US. Despite initial responsiveness to chemotherapy, > 50% experience recurrence. Recurrent disease is characterized by low response to chemotherapy and poor prognosis, especially platinum-resistant disease. Thus, there is an urgent need to identify new targeted therapy to improve the outcome of this aggressive cancer. The Transforming Growth Factor (TGFβ) family controls different cellular responses in development and cell homeostasis. Disruption of TGFβ signaling has been implicated in many cancers, including OV, but with paradox effects. It is unclear if TGFβ family genes can serve as a predictive marker and be selected for future therapy target in OV. Methods: We used quantified RNAseq data from 374 OV samples in The Cancer Genomic Atlas (TCGA) to extract 4 different sets of TGFβ pathway genes including: 3-gene ( TGFB1, TGFB2, TGFB3), 6-gene ( TGFB1, TGFB2, TGFB3, TGFBR1, TGFBR2, TGFBR3), 43-genes from cBioPortal, or 173-genes from Calon et al, 2012 to predict colorectal cancer relapse. Expression counts for these genes were normalized and log transformed to generate a TGFβ score based on the quartiles of the average. Samples were categorized in to low (lowest quartile), medium (middle 2 quartiles), and high (highest quartile) expression groups. Overall (OS) and progression-free (PFS) survivals were compared among groups using Kaplan-Meier method. Results: Interestingly, the minimum TGFβ 6-gene set can predict high grade serous ovarian cancer patient survival among all 4 gene panels using TCGA data, with average age of 60 years (range: 38 –81). Majority of the patients were stage IIIIV. OV patients with highest quartile, TGFβ signature expression have in general less somatic mutations and copy number alterations (average mutation count = 41, average CNA fraction = 0.5) compared to lowest quartile TGFβ expression group (average mutation count = 52, average CNA fraction = 0.6). Patients with TGFB 6-gene expression scored high showed worse OS (36.2 months vs 63.5 months, log-rank p = 0.001) and worse PFS (18.6 months vs 31.4 months, log-rank p = 0.09). Details of outcome by each specified gene signature will be outlined later. Conclusions: Our analysis suggested that the narrowed 6-gene TGFβ family expression profile could potential serve as a prognosis biomarker for OV. With proper validation and mechanistic investigation, existing TGFβ inhibitors may hold promise for targeted or combination therapies for OV. Targeting both TGFβ and PD1 might be of therapeutic value.