HIV viral load as an independent risk factor for tuberculosis in South Africa: collaborative analysis of cohort studies

Abstract

Introduction: Chronic immune activation due to ongoing HIV replication may lead to impaired immune responses against opportunistic infections such as tuberculosis (TB). We studied the role of HIV replication as a risk factor for incident TB after starting antiretroviral therapy (ART).

Methods: We included all HIV-positive adult patients (≥16 years) in care between 2000 and 2014 at three ART programmes in South Africa. Patients with previous TB were excluded. Missing CD4 cell counts and HIV-RNA viral loads at ART start (baseline) and during follow-up were imputed. We used parametric survival models to assess TB incidence (pulmonary and extrapulmonary) by CD4 cell and HIV-RNA levels, and estimated the rate ratios for TB by including age, sex, baseline viral loads, CD4 cell counts, and WHO clinical stage in the model. We also used Poisson general additive regression models with time-updated CD4 and HIV-RNA values, adjusting for age and sex.

Results: We included 44,260 patients with a median follow-up time of 2.7 years (interquartile range [IQR] 1.0-5.0); 3,819 incident TB cases were recorded (8.6%). At baseline, the median age was 34 years (IQR 28-41); 30,675 patients (69.3%) were female. The median CD4 cell count was 156 cells/µL (IQR 79-229) and the median HIV-RNA viral load 58,000 copies/mL (IQR 6,000-240,000). Overall TB incidence was 26.2/1,000 person-years (95% confidence interval [CI] 25.3-27.0). Compared to the lowest viral load category (0-999 copies/mL), the adjusted rate ratio for TB was 1.41 (95% CI 1.15-1.75, p < 0.001) in the highest group (>10,000 copies/mL). Time-updated analyses for CD4/HIV-RNA confirmed the association of viral load with the risk for TB.

Conclusions: Our results indicate that ongoing HIV replication is an important risk factor for TB, regardless of CD4 cell counts, and underline the importance of early ART start and retention on ART.

Keywords: CD4 cell count; HIV; antiretroviral treatment; incidence; opportunistic infection; prediction; time-updated; tuberculosis; viral load.

Figure 1.

Selection of the study population. ART: antiretroviral treatment; TB: tuberculosis.

Figure 2.

Crude incidence rates (per 1,000 person-years) of tuberculosis after starting antiretroviral therapy, by baseline levels of CD4 cell count (cells/μl) and HIV-RNA viral load (copies/mL). Based on 44,260 HIV-positive patients. Bars correspond to 95% confidence intervals. Unknown HIV-RNA viral load categories are shown in blue (unknown CD4 cell count category not shown).

Figure 3.

Influence of current CD4 cell count, and current HIV-RNA viral load on tuberculosis (TB) incidence. Models of TB incidence after starting antiretroviral therapy (ART) per 1,000 person-years, CD4 cell count (cells/µl), and HIV-RNA viral load (copies/mL) after imputation of missing CD4 cell counts and viral loads at start of ART and during follow-up, based on 44,260 HIV-positive patients. Curves represent patients with different HIV-RNA viral loads. (a) Viral load 100 vs. 10,000 copies/mL; (b) 100 vs. 100,000 copies/mL.

Figure 4.

Three-dimensional model of current CD4 cell count (cells/µl), and HIV-RNA viral load (copies/mL) as predictors of tuberculosis (TB) incidence per 1,000 person-years after starting antiretroviral therapy (ART). Based on 44,260 HIV-positive patients after imputation of missing CD4 cell counts and HIV-RNA viral loads at start of ART and during follow-up. TB:  tuberculosis.