Incorporating molecular markers in standard prognostic models for DLBCL patients using real-world data.

Abstract

e19251 Background: NCCN-IPI is a prognostic scoring system that outperforms other risk classification mechanisms in diffuse large B-cell lymphoma (DLBCL) but does not consider the molecular profile of patients. We evaluated the predictive value of NCCN-IPI and clinically relevant molecular markers on the overall survival (OS) of patients with diagnosed DLBCL in real-world data (RWD). Methods: Patients diagnosed with DLBCL were identified in the COTA RWD population, and then subset to those with sufficient attributes to calculate NCCN-IPI at diagnosis (age, stage, LDH ratio, performance status, extranodal disease) and those who received a monoclonal antibody targeting CD20 (n = 383). This population was further filtered to patients tested for BCL-2, BCL-6, and C-MYC (n = 176). Disease characteristics were summarized using descriptive statistics and chi-square tests of independence were performed to assess the relationship between NCCN-IPI Risk-Group and molecular marker results. A Cox proportional hazard model was used to identify prognostic features of OS. Results: There were statistically significant relationships between NCCN-IPI Risk-Group and both BCL-2 (p = 0.007) and C-MYC (p < 0.001) after Bonferroni correction for the number of molecular markers tested. A Cox proportional hazard model with the three molecular markers as covariates revealed a statistically significant correlation between the presence of C-MYC alteration and decreased OS (HR = 2.02, CI: 1.24-3.32, p = 0.005). However, when NCCN-IPI Risk-Group was added as a covariate, the relationship between C-MYC and OS was no longer significant. The hazard ratios associated with high-intermediate and high risk groups were larger than that of the low-intermediate group. All three of these risk groups were statistically significant in the model (p < 0.038). Conclusions: The inclusion of molecular markers to the NCCN-IPI prognostic model did not increase predictive power in this RWD cohort. We validated the NCCN-IPI model and found it to be a robust tool for classifying risk and estimating OS in patients who have received a CD20 monoclonal antibody. A larger sample size would increase power to further explore the impact of molecular markers on overall survival. [Table: see text]