Artificial Intelligence Combined with Digital Pathology Predicts PARPi Treatment Response in Ovarian Cancer
Principal Investigator
Name
Lingying Wu
Degrees
Ph.D.
Institution
Department of Gynecology Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College
Position Title
Professor
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCOI-1823
Initial CDAS Request Approval
Feb 12, 2025
Title
Artificial Intelligence Combined with Digital Pathology Predicts PARPi Treatment Response in Ovarian Cancer
Summary
Ovarian cancer has a relatively poor prognosis among gynecologic malignancies, with a recurrence rate exceeding 70% after traditional first-line treatment involving surgery and platinum-based chemotherapy. In recent years, the introduction of maintenance therapy into standard treatment has significantly improved the overall survival of ovarian cancer patients. Poly (ADP-ribose) polymerase inhibitors (PARPi) are one of the representative drugs in ovarian cancer maintenance therapy, particularly benefiting patients with BRCAm/HRD. However, some patients exhibit resistance to PARPi treatment or achieve prolonged survival without the need for maintenance therapy. Given the significant physical and financial burdens posed by the adverse effects and long-term use of PARPi, identifying the actual beneficiaries of PARPi among ovarian cancer patients is crucial. With advancements in algorithms, the application of artificial intelligence (AI) in medicine has become increasingly widespread. This study aims to develop an AI model capable of predicting PARPi treatment response by integrating deep learning and digital pathology technologies. The model's predictive accuracy will be enhanced by incorporating multimodal data, including multi-tissue source pathological images and comprehensive clinical information of patients.
Aims
(1) Establish an AI (deep learning) prediction model capable of identifying advanced ovarian cancer patients who achieve prolonged prognosis (progression-free survival (PFS) exceeding 2 years after the last chemotherapy) with only first-line surgery and chemotherapy, or different survival benefit categories (whether PFS exceeds 2 years after the last chemotherapy) following PARPi treatment, based on digital pathological features.
(2) Further enhance the predictive accuracy of the AI model by incorporating multimodal data, including multi-tissue source pathological images and comprehensive clinical information.
Collaborators
-