Abstract
PURPOSE: Plasmablastic lymphoma (PBL) is a rare and aggressive B-cell malignancy with a heterogenous clinical and prognostic spectrum, determined by multiple factors, including age, HIV- and MYC-status. While there exist several validated scoring systems for diffuse large B-cell lymphoma, which incorporate basic clinical features (age, lactate dehydrogenase, sites of (extranodal) involvement, stage and performance), none of these have been systematically assessed in PBL. METHODS: We determined the (age-adjusted; aa)-International Prognostic Index (IPI), revised IPI (R-IPI), and National Comprehensive Cancer Network IPI (NCCN-IPI) in a comprehensive multi-center cohort (n = 78) of PBL patients. Further, all indices were comparatively investigated for model quality and concordance. RESULTS: Univariate analysis revealed significant prognostic capabilities for all indices, all of which identified a subgroup with favorable outcome. Discriminatory power between patients with less benign prognosis and especially refractory disease exhibited significant variability. Subsequently, stratified models for each risk score were compared employing corrected Akaike's information criterion (cAIC) and Harrel's concordance index (c-index). Here, the NCCN-IPI outperformed both IPI and R-IPI regarding c-index with ambiguous cAIC results, underlining its clinical utility and suggesting it for preferential use in clinical practice. CONCLUSION: Our current observations support the use of the IPI and its enhanced derivatives in PBL patients. There is, however, a distinct requirement for novel prognostic tools to better delineate subgroups at risk for early relapse or refractory disease as well as late relapse. A comprehensive molecular characterization of a clinically annotated cohort of PBL patients is therefore urgently warranted.