Universal Time: 20:24  |  Local Time: 20:24 (0h GMT)
Select your timezone:

Clinical classifiers, trials and the patient voice

Thursday May 13, 2021 - 20:20 to 21:00

Room: Breakout

P-4.4 Clinical utility for molecular subtypes of high-grade serous ovarian cancer: Improving response to bevacizumab

Evan S Cairns, Canada

Graduate student
Department of Obstetrics and Gynaecology
University of British Columbia

Abstract

Clinical utility for molecular subtypes of high-grade serous ovarian cancer: Improving response to bevacizumab

Evan Cairns1, Pauline Krämer2, Lauren Tindale1, Boris Winterhoff3, Gottfried Konecny4, David Hunstman1,5, Andreas du Bois6, Friedrich Kommoss7, Jacobus Pfisterer6, Stefan Kommoss2, Michael Anglesio1, Aline Talhouk1.

1Department of Obstetrics and Gynaecology, University of British Columbia, Vancouver, BC, Canada; 2Department of Women's Health, Tübingen University Hospital, Tübingen, Germany; 3Department of Obstetrics, Gynecology and Women's Health, University of Minnesota, Twin Cities, MN, United States; 4Department of Medicine, University of California, Los Angeles, Los Angeles, CA, United States; 5Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, BC, Canada; 6AGO Study Group, Wiesbaden, Germany; 7Institute of Pathology, Im Medizin Campus Bodensee, Friedrichshafen, Germany

Introduction: High-grade serous ovarian cancer (HGSOC) can be divided into four prognostic subtypes based on gene expression. Bevacizumab is an antiangiogenic drug recently introduced to treat recurrent ovarian cancer. ICON7 was a phase 3 trial of bevacizumab, and samples from this study have been used to demonstrate that molecular subtypes of HGSOC are predictive of treatment response. Here, we re-examined ICON7 samples to investigate whether molecular subtypes of HGOC inferred through the novel NanoString-based PrOTYPE signature can reproduce the previously reported treatment response signature. 

Methods: To test the reproducibility of existing subtyping methods, we applied the PrOTYPE classifier to 408 ICON7 patient tumours and compared our results with those previously published with subtype labels produced from Illumina DASL expression array data. Concordance and outcomes were compared across the different subtype label sets.

Results: DASL-derived labels suggested modest improvement for bevacizumab treatment in the proliferative and mesenchymal subtypes, while PrOTYPE classification showed significant improvements in overall and progression-free survival of the immunoreactive subtype. Repeated clustering of DASL data using the same cases and methods described by the original study produced subtype labels agreed in only 65% of cases. The original and re-derived DASL labels also showed poor agreement with PrOTYPE labels. The DASL-generated gene expression signature also showed poor concordance with signatures from previous studies, particularly for the mesenchymal subtype.

Conclusions: DASL-based subtype signatures and labels were challenging to reproduced, while PrOTYPE offers a more reproducible method of determining HGSOC subtypes. Findings from our initial cohort suggest that PrOTYPE-defined subtypes could be used to predict patient response to bevacizumab. Future steps will include the validation of these findings in an independent cohort and search for a bevacizumab-specific novel response signature.

© 2022 Virtual CCOCR 2021