Abstract 4089: Time to treatment discontinuation and time to next treatment as proxies of real-world progression-free survival in breast cancer patients

Background: There is no standardized real-world proxy for progression-free survival (PFS), although it is a key treatment response metric in clinical trials. Real-world PFS (rwPFS) has been curated from clinician-documented progression in unstructured electronic health records (EHRs), a time and resource intensive approach, that is not readily accessible in claims or structured EHRs. Alternatively, time to treatment discontinuation (rwTTD) and time to next treatment (rwTTNT) are used as treatment-based scalable proxies of disease progression. We compared clinician-documented rwPFS to rwTTD and rwTTNT. Methods: The study included patients with hormone receptor-positive/human epidermal growth factor receptor 2-negative advanced or metastatic breast cancer (A/MBC) who initiated first line treatment with palbociclib plus an aromatase inhibitor (1L) between 2015 and 2019 in US community health systems. Demographics, clinical characteristics, and outcomes were extracted from EHRs. rwPFS was captured through EHR-documented confirmation by two clinicians of disease progression based on imaging and/or pathology results. rwPFS, rwTTD and rwTTNT were defined as the time from 1L initiation until the date of clinician-confirmed progression, 1L discontinuation, or 2L initiation, respectively. Patients were censored at the first of: (1) death, (2) date of last contact, (3) study end date, (4) or at 2L for rwPFS. Results: The study included 241 patients with median age 66 at A/MBC diagnosis, among whom 85% were postmenopausal and 57% were de novo A/MBC. During study follow-up (median 22.4 months), 86 (36%) patients had a clinician-confirmed progression (rwPFS events); 130 (54%) discontinued 1L (rwTTD events), and 96 (40%) initiated 2L (rwTTNT events). With clinician-confirmed progression as the “gold standard”, rwTTD had higher sensitivity (90% vs 71%) but lower specificity (66% vs 77%) than rwTTNT. Patient’s choice, hospice referral, and toxicity were common reasons for treatment discontinuation apart from disease progression. The 12, 24, and 48-month survival estimates were closer between rwPFS (0.78, 0.64, 0.51) and rwTTNT (0.80, 0.63, 0.47), than rwTTD (0.69, 0.51, 0.37). Median rwPFS (39 months, 95% confidence interval [95% CI]: 30-NA) and rwTTNT (32 months, 95% CI: 29-46) were also more alike (log-rank p = 0.90) than median rwPFS and rwTTD (24 months, 95% CI: 18-30; log-rank p=0.02). The correlation between rwPFS and rwTTNT was 0.72 (95% CI: 0.56-0.83), compared to