Abstract
Background: Diffuse large B-cell lymphoma (DLBCL) is a heterogeneous malignancy with outcomes influenced by clinicopathologic factors and treatment modalities. Public safety-net hospitals often serve patients with higher comorbidities and socioeconomic challenges, necessitating tailored approaches to optimize care. We conducted a 10-year retrospective study at New York City Health + Hospitals/Kings County to evaluate the prognostic impact of clinical features and treatment on survival in DLBCL patients.
Method: We retrospectively analysed the electronic health records of 75 patients diagnosed with DLBCL from 2013 to 2023. Key clinicopathologic variables included age, gender, ECOG performance status, Charlson Comorbidity Index (CCI), International Prognostic Index (IPI), HIV status, cell-of-origin subtype (germinal center B-cell [GCB] vs. non-GCB), and treatment approach. Logistic regression was used to identify predictors of mortality, while Cox proportional hazards models and log-rank tests were applied to evaluate overall survival (OS), adjusting for relevant covariates.
Results: Of the 75 patients, 62 with at least one year of follow-up were included in the final analysis. Older age at diagnosis was a significant predictor of mortality (odds ratio [OR] 1.07 per year; 95% CI: 1.03–1.11; p=0.001). Among 61 patients with IPI data, high-risk IPI (score ≥3; n=44) was associated with significantly worse one-year OS (43.0%) compared to low-risk IPI (score <3; n=17; OS 93.8%; log-rank p=0.0008). Each one-point IPI increase conferred a 55% increased hazard of death (hazard ratio [HR] 1.55; 95% CI: 1.01–2.37; p=0.044). ECOG performance status was an independent predictor of mortality (95% CI: 1.66–15.10; p=0.004).
CCI data were available for 61 patients; 41 (67%) had a score ≥4. Higher CCI significantly correlated with worse survival, with each one-point increase conferring a 21% increased hazard of death (HR 1.21; 95% CI: 1.09–1.35; p=0.0004). HIV-positive status showed a non-significant trend toward worse survival in multivariable Cox analysis (HR 1.67; p=0.20).
Patients with GCB subtype exhibited worse unadjusted survival compared to non-GCB subtype (log-rank p=0.02). However, after adjusting for Ann Arbor stage in a Cox model, this difference was attenuated (HR 0.46 for non-GCB vs. GCB; 95% CI: 0.18–1.15; p=0.097), suggesting advanced-stage disease in GCB patients as a critical factor influencing survival. Insurance status (Medicaid, private, or uninsured) did not significantly impact OS (log-rank p=0.8), with observed deaths proportional to expected deaths across groups.
Treatment regimens included DA-EPOCH-R (n=16) for patients with high-risk features, R-CHOP (n=27), supportive care only (n=13), and other reduced-intensity regimens (n=6). One-year OS was 68.8% for DA-EPOCH-R (95% CI: 49.4–95.7), and 73.6% for R-CHOP (95% CI: 58.5–92.5), indicating comparable short-term outcomes. Patients receiving supportive care or single-agent therapy had significantly worse outcomes (one-year OS 44.4%; 95% CI: 16.7–100), reflecting their palliative treatment intent due to poor performance status.
Conclusion: This 10-year retrospective study at a public safety-net hospital identifies older age, high-risk IPI, and elevated CCI as significant predictors of poorer survival in DLBCL patients. Comorbidities frequently precluded more intensive therapies such as DA-EPOCH-R or R-CHOP, underscoring the need for personalized treatment strategies. Compared to SEER data, our cohort exhibits a higher prevalence of comorbidities (e.g., 67% with CCI ≥4 vs. ~40% in SEER DLBCL cohorts), likely due to the socioeconomic and health disparities in our safety-net population. Insurance status did not influence survival outcomes, likely due to uniform access and financial support systems within our safety-net health system.
