Impact of the Diagnosis-to-Treatment Interval (DTI) on Outcomes in Older Adults with Diffuse Large B-Cell Lymphoma (DLBCL) Treated in the Real-World Setting

Introduction: DLBCL is the most common non-Hodgkin lymphoma and is potentially curable in 60-70% of patients (pts) with anthracycline containing chemoimmunotherapy. Given a predilection for presentation in older adults (OA), the treating oncologist is often confronted with the intersection of age-related comorbidity, more adverse disease biology (non-germinal center B-cell (GCB) like phenotype, higher disease stage) and disease related impairments when deciding on initial therapy. Prior reports have detailed that diagnosis-to-treatment interval (DTI) is inversely related to survival outcomes (Maurer et al, JCO 2018), and that this relationship introduces time-dependent selection bias into clinical trial reports. This occurs because pts with the most aggressive disease and associated comorbidities require more immediate therapy and are inadvertently excluded from clinical trials due to restrictive eligibility, resulting in improved outcome for both control and experimental cohorts. We hypothesized that the impact of shorter DTI on survival in a real-world (rw) database analysis specifically focusing on OA would be pronounced and could inform future trial design.

Methods: This retrospective, observational study utilized the COTA rw database derived from the electronic health records of partner healthcare centers in the United States. Eligible pts were aged ≥ 70 years (yrs) at diagnosis with DLBCL between 1/1/2017 and 12/31/2021 and initiated first-line (1L) treatment. Pts were excluded from the study if they were missing or had imprecise key study dates (date of diagnosis, treatment initiation, or death). Patient characteristics and treatment patterns were summarized descriptively by subgroups. DTI were calculated as the duration of time from DLBCL diagnosis to initiation of the first LOT. Rw time to next treatment (rwTTNT) and overall survival (rwOS) were evaluated using the Kaplan-Meier method. A Cox-proportional hazards model was used to calculate hazard ratios (HRs) for overall survival by DTI group.

Results: A total of 1,766 pts met the study criteria. Median age at diagnosis was 77 (range: 70-90+) yrs; 53.8% pts were male, 69.6% white, 56.3% had stage 3/4 disease, 38.5% B symptoms, 68.5% extranodal disease, 25.0% bulky disease, 13.5% ECOG ≥ 2. Age groupings included 70-79 yr (n=1110, 62.3%), 80-89 yrs (n=593, 33.6%) and 90+ yrs (n=73, 4.1%). The practice setting distribution was 86.6% community and 13.4% academic. A total of 664 pts had GCB subtype (37.6%), 556 (31.5%) had activated B-cell-like (ABC) subtype, and subtype was unknown for 546 pts (30.9%). Mean (SD) CCI score at diagnosis for the study population was 1.2 (1.5), with 288 (16.3%) pts having a CCI score of ≥ 3.

Median follow up from diagnosis was 28.0 (range: 0.4, 86.9) months (mos). Median rwTTNT and rwOS (95% CI) from 1L treatment was 43.7 (35.6, 49.8) and 57.1 (52.2, 59.9) mos, respectively. Pts who initiated treatment within 2 weeks of diagnosis (N=446) had lower median rwTTNT (19.4 mos) and rwOS (38.4 mos) compared to pts who initiated therapy in 2-4 weeks (N=508, rwTTNT: 47.3, rwOS: 61.9), 4-6 weeks (N=398, rwTTNT: 49.6, rwOS: 57.1), or 6+ weeks (N=414, rwTTNT: 49.7, rwOS: 60.1). In the overall population, survival probability at the median follow up time of 28 months was 64% (95% CI: 62%, 67%). As compared to pts who initiated treatment within 0-2 weeks of diagnosis (ref), pts initiating treatment later (> 2 weeks) were at a significantly lower risk of death (HR: 0.72, 95% CI: 0.62, 0.85, p<0.001). Survival probability at 28 months for all pts with a DTI > 2 weeks was 67%, as compared to 57% for patients initiating therapy in 0-2 weeks from diagnosis.

When comparing pts treated within 2 weeks of diagnosis vs. later, pts who received immediate treatment were more likely to be treated in the academic setting, have ECOG score ≥ 2 at diagnosis, have presence of bulky disease, and have ABC subtype.

Conclusions: Older adult pts who initiated treatment rapidly (0-2 weeks from diagnosis) had significantly worse rwTTNT and rwOS, underscoring the need for representativeness and generalizability of clinical trial results given the potential for selection bias if these pts are unable to enroll in clinical trials. Additionally, data such as these can help anchor expected survival data for trial control groups more realistically by utilizing estimates from the cohorts with DTI > 2 weeks.