Facility-level differences and outcomes in elderly primary CNS lymphoma in the NCDB.

Abstract

e13762 Background: PCNSL is a rare, aggressive subtype of non-Hodgkin lymphoma with increasing incidence in the U.S (ACSPMID:19273630). Outcomes in older adults remain poor; prior cohorts of patients aged > 70 years reported a 2-year overall survival of 50% (NCBIPMID:32241841). Facility-level differences in care have been reported, but their impact on older adults is unclear. We compared academic vs community cancer programs in elderly PCNSL to evaluate demographic, socioeconomic, and survival differences. Methods: A retrospective cohort study of patients diagnosed with PCNSL between 2004-2022 using the National Cancer Database (NCDB) for patients ≥75 years old. Patients were classified by treatment facility: ACP (academic/research programs including NCI-designated centers) vs. CCP (community, comprehensive community, and integrated network cancer programs). Demographic, clinical, and treatment characteristics were compared using chi-square and Wilcoxon tests. Kaplan-Meier and Cox proportional hazards models were used to compare overall survival (OS), adjusting for age, race/ethnicity, insurance status, Charlson-Deyo comorbidity index, and distance traveled for care. Results: 5,013 patients aged ≥75 with PCNSL, 3,204 were treated at ACP and 1,809 at CCP. Median age, sex, comorbidity burden, and stage were similar across facility types. Ethnicity did not differ (p = 0.054). CCP cared for a higher proportion of lower-income patients (2016–2020: 15% vs 10%; p < 0.001) and were more frequently located in non–Medicaid expansion states (39% vs 31%; p < 0.001), while ACP patients traveled farther (median 14.9 vs 9.2 miles; p < 0.001). Treatment at ACP were less likely to receive no treatment (6% vs 12%; p < 0.001), more likely to receive treatment (70% vs 63%; p < 0.001) and initiated systemic therapy sooner (median 15 vs 18 days; p = 0.001). Radiation was more frequently administered in CCP (24% vs 20%; p = 0.002). Early mortality was higher in CCP (30-day: 4% vs 3%, p = 0.013; 90-day: 13% vs 10%, p = 0.011). OS favored ACP, with higher 2-year (31% vs 23%), 5-year (19% vs 13%) and 10-year (9% vs 5%) survival and adjusted median survival (0.48 vs 0.32 years). Conclusions: Among patients aged ≥75 years with PCNS, OS differed significantly by treatment facility type, with a statistically and clinically meaningful survival advantage observed among patients treated at ACP compared with CCP. OS difference was consistent across short- and long-term follow-up with lower early mortality, higher rates of treatment delivery, and shorter time to initiation of therapy at ACP. These results point to system-level gaps in care for older adults with PCNSL and underscore the need to strengthen referral pathways to ACP and build structured academic–community partnerships. Doing so could expand timely access to specialized multidisciplinary care and clinical trials for vulnerable patients more often treated in lower-resource settings.