Causes and Predictors of Early Mortality in HIV-Positive and HIV-Negative Patients with Diffuse Large B-Cell Lymphoma

INTRODUCTION: There is a lack of data on early mortality in aggressive lymphomas, particularly HIV-related lymphomas. Patients who present with advanced disease and poor performance status are at a higher risk of adverse outcomes and are not offered clinical trial enrollment. Factors that may contribute to early mortality have not been delineated. In this retrospective analysis, we sought to identify causes and predictors of early mortality, occurring within 180 days of index diagnosis, in patients with HIV-positive and negative DLBCL, at an academic medical center. METHODS: We reviewed an electronic database to identify patients with a diagnosis of HIV-related DLBCL, treated at our center between January 1, 2005 and December 31, 2017. We further identified patients with non-HIV positive DLBCL diagnosed from January 1, 2015 to December 31, 2017. Data on patient demographics, clinical outcomes and treatment was collected. We defined early mortality as those patients who died within 180 days of diagnosis. RESULTS: HIV-NEGATIVE DLBCL: There were 103 HIV-negative patients with DLBCL, with a mean age of 65 years. N=17 (16.5%) patients expired within 180 days of diagnosis. In univariate analysis, factors predicting increased risk of mortality included poor ECOG performance status (PS) 3-4, presence of B-symptoms and higher international prognostic index (IPI) scores (Table 1). In multivariate analysis, poor PS (HR 4.86, p HIV-POSITIVE DLBCL: We identified 82 patients with HIV-positive DLBCL and divided patients into Cohort A (N=25) patients who expired within 180 days and Cohort B (N=57), patients who were alive beyond 180 days. 9 patients died within the first 30 days, 20 patients within 90 days and 25 patients within 180 days of index diagnosis. Both groups predominantly consisted of African-Americans with advanced stage DLBCL with a mean age of 53 and 48 years respectively (p=0.02). In univariate analysis, more patients in Cohort A had Stage 3-4 DLBCL (100% vs. 86%), poorer ECOG PS of 3-4 (52% vs. 7%), higher IPI scores (56% vs. 11%), lower CD4 counts (median of 70 cells/uL vs. 140 cells/uL) and higher LDH (544 U/L vs. 270 U/L). In multivariate analysis, we found that poor ECOG PS (HR 2.4, p We wanted to examine whether the time period of diagnosis affected mortality from HIV-related DLBCL in patients. Group 1 patients were diagnosed from January 2005 to December 2010 (N=20) and Group 2 from January 2011 to December 2017 (N=29). We found that in Group 1, 50% of patients died from progressive disease and 15% of patients died from an acute event. Meanwhile, in Group 2, 38% of patients died from progressive disease, and 52% died from an acute event. Sepsis was the most common acute event in both groups. CONCLUSION: Preventing early mortality in aggressive lymphomas is crucial to improving overall survival. For patients with non-HIV related DLBCL, we found poor ECOG PS and presence of B-symptoms to be predictors of early mortality. In patients with HIV-positive DLBCL, poor ECOG PS, higher LDH and older age were contributing factors. The leading cause of death in both HIV-positive and HIV-negative patients (2011- 2017) was sepsis followed by progressive disease at 180 days after diagnosis. Although sepsis is a major factor contributing to early mortality and morbidity, it is preventable with prophylaxis against opportunistic infections. In order to decrease sepsis related mortality, we propose early initiation of antimicrobial therapy in patients at high risk based on independent clinical variables identified in this study. A prospective study preventing sepsis in high risk patients is warranted to decrease early mortality in patients with aggressive lymphoma. Disclosures Janakiram:Seattle Genetics: Membership on an entity9s Board of Directors or advisory committees.