Racial Disparity In Immunotherapy for Diffuse Large B-Cell Lymphoma Explained by HIV Status In Veterans Administration Patients

Previous studies have demonstrated that African-American (AA) patients with diffuse large B-cell lymphoma (DLBCL) are less likely to receive rituximab (R) compared to Caucasian patients, despite the incorporation of R into the standard of care in 2002. Reasons for this difference are unclear, though insurance coverage differences have been suggested as one potential cause. We evaluated DLBCL patients treated at Veterans Health Administration (VHA) hospitals in an effort to identify other potential causes of racial differences in R use. Since VHA patients receive treatment regardless of insurance coverage, insurance status should not influence treatment decisions in this population.

Electronic records of patients diagnosed with DLBCL between October 2002 and September 2008 were identified in the Veterans Administration Central Cancer Registry. Data was obtained on patient demographics and stage at diagnosis. Each record was then linked to pharmacy records and ICD-9 diagnostic codes, to identify drugs used in treatment of DLBCL and concurrent co-morbidities respectively. Of 2,164 newly diagnosed DLBCL patients identified in the VACCR, drug records were obtained for 1,942 (90%), and CHOP or R-CHOP was used as first treatment in 1446 (67%). Only patients who received CHOP or R-CHOP were included in the analysis. Demographic and stage variables were reported in approximately 95% of patients.

For comparison between AA and Caucasian patients, see Table. Univariate analyses demonstrated significant differences in R use based on age (p =0.002), race (p= 0.006), and HIV status (p<0.0001), while stage and sex were not significant. A logistic regression model was constructed using statistically significant univariate predictor variables and a categorical variable to differentiate year of treatment (2003 vs. ≥ 2004). In the final logistic model, only HIV + status (OR = 0.096, p <0.0001) and year of treatment ≥ 2004 (OR = 4.311, p<0.0001) remained significant predictors of R use. Linear regression analysis of variables predicting of age at diagnosis demonstrated independent effect of both AA race (β= - 5.13, p <0.0001) and HIV + (β= -13.97, p<0.0001), suggesting that HIV status accounts for only part of the difference in age at diagnosis noted between AA and Caucasians.

After controlling for HIV status, race no longer predicted use of R in patients treated within the VHA system. The decreased use of R in HIV+ patients is likely a related in part to a 2005 publication which demonstrated increased infectious toxicity in HIV associated lymphoma patients treated with R. Given the higher prevalence of HIV in the AA population compared to the Caucasian population in the United States (546.7/100,000 vs. 80.4/100,000 in 2007), differential rates of HIV may be responsible for some of the previously observed disparity in R use outside the VHA. On the other hand, race based differences in age at diagnosis are only partially explained by HIV status. In conclusion, HIV status is an important co-morbid factor that should be considered in future studies of racial disparities in lymphoma patients.

Carson:Genentech, Inc.: Consultancy, Speakers Bureau; American Cancer Society, IRG 58-010-52: Research Funding.