Estimating the United States cure-adjusted prevalence of diffuse large B-cell lymphoma (DLBCL): An epidemiological model

Abstract Introduction: Prevalence reflects both how often a disease occurs and how long patients survive. In diffuse large B-cell lymphoma (DLBCL), treatment advances now enable cure for an expanding subset of patients, with front-line treatment curing about 60% of patients and those cancer-free at two years expecting normal life expectancy. However, traditional prevalence estimates may overestimate disease burden by including cured patients, potentially leading to misallocation of healthcare resources and treatment planning. Accurate prevalence estimates that account for cures are essential to reflect the true number of patients currently living with active disease. Using Surveillance, Epidemiology, and End Results (SEER) data, it was estimated that fewer than 200,000 persons in the U.S. were living with DLBCL in 2021 [Chihara et al. 2022]. This epidemiological model provides an updated cure-adjusted prevalence estimate of DLBCL in the U.S. for 2025 based on more recent SEER data (2018-2022). Methods: Using 2000-2022 SEER-21 Research Data (SEER*Stat Version: 9.0.40.1), DLBCL incidence and survival for patients with first primary DLBCL tumors identified using the 2021 Lymphoid neoplasm recode classification system (code 2(a)2.3) were analyzed. Crude incident case counts between 2000 and 2022 were used as observed. Incident cases for 2023-2025 were projected using linear extrapolation of historical incident cases. Parametric Weibull survival models were fitted by diagnosis year in R using the flexsurvreg package, with different parameters for each year. Standard extrapolation equations [Briggs et al. 2006] were used to extrapolate survival to 2025 in a Microsoft Excel model. Prevalence estimates were calculated as the sum of patients alive in 2025 based on overall survival rates of modeled U.S. population-level incident cases conditional on diagnosis year. With cure assumptions, patients assumed to be cured based on cure parameters were removed from the prevalence pool. Cure was defined by three parameters: timepoint at which patients can first be considered cured (1-4 years), proportion of patients who were cured at that time (40-70%), and timepoint at which all survivors are considered cured (5 or 10 years). Results: When incorporating cure assumptions, prevalence estimates decreased substantially. In the least aggressive scenario (no early cure, all survivors considered cured at year 10), prevalence was 150,363. When shortening the cure timepoint to 5 years (no early cure, all survivors cured at year 5), prevalence decreased further to 94,162. When incorporating early cure parameters (40-70% cured at years 1-4), prevalence decreased even more. With all remaining survivors considered cured at year 10, prevalence ranged from 79,199 (70% cured at year 1) to 123,190 (40% cured at year 4). With all remaining survivors considered cured at year 5, prevalence ranged from 62,338 (70% cured at year 1) to 89,469 (40% cured at year 4). Without accounting for cure, the estimated 2025 DLBCL prevalence was 233,904. Conclusions: Adjusting prevalence estimates for cure reflects the advances of DLBCL therapies which have demonstrated curative potential. All cure-adjusted prevalence estimates for DLBCL in the U.S. were substantially below 200,000 cases, independent of cure assumptions. The presented estimates are for any first primary DLBCL case. Future work is needed to differentiate prevalence estimates for both newly diagnosed and relapsed/refractory DLBCL, and should include exploring alternative methods and leveraging U.S. real-world evidence database analyses.