Abstract Body

Despite increasing numbers of HIV-infected South Africans receiving antiretroviral therapy, tuberculosis remains the leading cause of mortality. Approximately 25% of suspected tuberculosis is not confirmed microbiologically. We assessed whether co-infection with Kaposi Sarcoma-associated Herpes Virus (KSHV), and particularly the newly described KSHV inflammatory cytokine syndrome (KICS) contributes to mortality in HIV-infected patients with suspected tuberculosis.

We conducted a cohort study of HIV-infected patients presenting to Khayelitsha Hospital, Cape Town, South Africa, with suspected tuberculosis and followed for 12 weeks. KSHV serostatus was evaluated using ELISA for KSHV K8.1 and LANA. KSHV DNA was measured in blood cells. Clinical correlatives of KSHV lytic activation were evaluated. KSHV viral load and clinical criteria for KICS were evaluated for prognostic value.

682 participants were accrued, 47% men, 53% women. Median age 36 years (range: 18-80), median CD4 count 62 cells/μl (range: 0-526). 30.7% (95% CI: 27-34%) were KSHV seropositive. 10 (1.5%) had clinical Kaposi sarcoma, 5% had elevated cell-associated KSHV DNA (>100 copies/106 cells). Anemia was associated with antibodies against KSHV lytic antigen K8.1 (p=0.003). Overall, 22% died. In patients without tuberculosis or other microbiologically confirmed co-infections, elevated KSHV viral load was associated with mortality after adjusting for age, sex, CD4 count and ART (p=0.023, OR=6.467 [95% CI: 1.290, 32.406]). Applying KICS working case criteria, we identified six ‘possible KICS’ patients. Five died, a significantly worse survival (p<0.0001) than other patients without microbiologically confirmed co-infections.

Due to high mortality associated with KICS, contribution of lytically active KSHV in critically ill patients presenting with tuberculosis should be considered to avoid misdiagnosis and to determine appropriate treatment strategies. KSHV viral load testing and application of KICS criteria is warranted to identify HIV-infected South African patients with suspected tuberculosis but no microbiologically identified infection at high risk of death.