Treatment Outcomes of AIDS-Associated Kaposi's Sarcoma under a Routine Antiretroviral Therapy Program in Lilongwe, Malawi: Bleomycin/Vincristine Compared to Vincristine Monotherapy

PURPOSE Despite Kaposi's sarcoma (KS) being the most prevalent AIDS-associated cancer in resource limited settings, optimal treatment options remain unknown. We assessed whether bleomycin/vincristine compared to vincristine monotherapy was associated with improved treatment outcomes for AIDS-associated KS among patients initiating combination antiretroviral therapy (cART) in Malawi. METHODS All patients initiating cART and chemotherapy for AIDS-related KS were identified from an electronic data system from the HIV Lighthouse Clinic from 2002 to 2011. Treatment responses were compared between patients receiving vincristine monotherapy and vincristine/bleomycin. Binomial regression models were implemented to assess probability of tumor improvement for patients receiving vincristine/bleomycin compared to vincristine monotherapy after a complete cycle of chemotherapy (9-10 months). A chi-squared test was used to compare changes in CD4 count after six months of chemotherapy. RESULTS Of 449 patients with AIDS-associated KS on chemotherapy, 94% received vincristine monotherapy and 6% received bleomycin/vincristine. Distribution of treatment outcomes was different: 29% of patients on vincristine experienced tumor improvement compared to 53% of patients on bleomycin/vincristine. Patients receiving bleomycin/vincristine were 2.25 (95% CI: 1.47, 3.44) times as likely to experience tumor improvement as to those on vincristine monotherapy. This value changed little after adjustment for age and baseline CD4 count: 2.46 (95% CI: 1.57, 3.86). Change in CD4 count was similar for patients receiving vincristine monotherapy and bleomycin/vincristine (p = 0.6). CONCLUSION Bleomycin/vincristine for the treatment of AIDS-associated KS was associated with better tumor response compared to vincristine monotherapy without impairing CD4 count recovery. Replication in larger datasets and randomized controlled trials is necessary.