Abstract A056: Development and application of an epithelial ovarian cancer-specific line of therapy algorithm using real-world data

Abstract Introduction. Characterizing lines of therapy (LoTs) helps researchers describe the epithelial ovarian cancer (EOC) patient treatment journey, but they are not routinely captured in real-world data. Algorithms must therefore be developed to transform individual treatment data into summary LoTs. Special considerations must be made for EOC as maintenance (MTx) therapy, which includes poly ADP ribose polymerase inhibitors (PARPi) and/or bevacizumab, may be given following treatment (Tx) lines. We developed an EOC-specific LoT algorithm and evaluated its performance using real-world data. Methods. Patients were sampled from the Syapse Learning Health Network, a database of cancer patients treated in large community health systems across 25 states in the US and were eligible for inclusion if they were diagnosed with EOC between 01/01/2015 and 09/30/2022 (index date) and had manually curated systemic therapy data. All patients were followed through 12/31/2022. Assignment of Tx and MTx lines and regimens and allowable treatment gaps were based on national treatment guidelines and typical disease course. To distinguish between Tx and MTx lines, all eligible patients who received systemic therapy following the initial diagnosis of EOC were assigned first-line (1L) Tx, and each subsequent LoT was classified as a Tx or MTx line depending on the therapies in the present LoT and one immediately preceding it. Therapies could be added or switched within a LoT, and assigned regimens distinguished between therapies given in combination from switch therapies. A shorter treatment gap was used to identify later LoTs to account for more rapid disease progression. Unexpected regimens were flagged for further evaluation. A subset of patients with advanced stage, non-mucinous EOC who received 1L platinum-based chemotherapy were manually reviewed by a medical oncologist to identify patients who received 1L MTx. Results. There were 2, 275 eligible EOC patients. 415 patients with missing, incomplete, or invalid therapy dates were excluded, leaving 1, 860 patients in the LoT analysis. Median follow-up was 23 months (IQR, 13-41 months). Nearly half (48%) of the patients received only one LoT; 15% of patients received 3 or more LoTs. Out of all patients, 15% received 1L MTx without any subsequent treatment. The top 1L Tx regimens were platinum and taxane combination, platinum and taxane plus bevacizumab, and platinum alone (73%, 16%, and 2% of patients who received 1L Tx, respectively). The top 1L MTx regimen was PARPi alone, followed by bevacizumab alone and bevacizumab plus PARPi (57%, 35%, and 8% of patients who received 1L MTx, respectively). 1, 355 (74%) patients were identified for expert review. Among these, concordance in 1L MTx determination between expert review and the algorithm was 98%. Conclusion. The EOC-specific LoT algorithm performed well: line and regimen distributions aligned with real-world practice and expert clinical review. Researchers may wish to utilize this algorithm for real-world studies of ovarian cancer. Citation Format: Connor Sweetnam, Tirza Areli Calderón Boyle, Sara Burns, Jessica Perhanidis, Matthew Rioth, Rayna K Matsuno. Development and application of an epithelial ovarian cancer-specific line of therapy algorithm using real-world data abstract. In: Proceedings of the AACR Special Conference on Ovarian Cancer; 2023 Oct 5-7; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84 (5 Suppl₂): Abstract nr A056.