University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Short-term and long-term risk of tuberculosis associated with CD4 cell recovery during antiretroviral therapy in South Africa. AIDS
Global Health Sciences Literature Digest
Published February 3, 2010
Journal Article

Lawn SD, Myer L, Edwards D, et al. Short-term and long-term risk of tuberculosis associated with CD4 cell recovery during antiretroviral therapy in South Africa. AIDS. 2009 Aug 24;23(13):1717-25.

In Context

The concurrent treatment of tuberculosis (TB) and HIV is problematic. Rates of TB appear to be particularly high during early initiation of antiretroviral therapy (ART), likely due to the unmasking of subclinical infections. The CD4 cell count above which TB rates are minimized during ART-induced immune recovery has not been well-quantified.


To determine the short- and long-term risks of developing TB during ART and the relation of these risks to CD4 cell count


Gugulethu township, Cape Town, South Africa

Study Design

Observational longitudinal cohort study


Treatment-naïve, HIV-positive patients ≥15 years of age who are eligible for and placed on ART and who have a CD4 count of <200 cells or World Health Organization (WHO) clinical stage 3 or 4.


Incidence rates of TB during early and long-term ART and estimated according to follow-up time spent within different CD4 cell-count strata


Eligible patients were consecutively recruited into the cohort over a 4-year period (2002-2006) and started on first-line ART, which consisted mostly of lamivudine/stavudine/efavirenz. Viral load tests, CD4 cell counts, and a TB screening questionnaire were done at baseline and every 4 months. Patients symptomatic for TB were evaluated with sputum smear fluoroscopy, automated liquid culture, chest radiograph, abdominal ultrasound, or lymph node aspiration, or all of these, as appropriate. The duration of ART treatment was classified as "short-term" (<4 months) or "long-term" (≥4 months). Patients had not previously received, nor were provided, isoniazid preventive therapy (IPT). Person-time of observation accrued during treatment of TB present at baseline (prevalent TB) or during treatment of incident cases was excluded in the analysis. Kaplan-Meier estimates of TB-free survival were calculated; Poisson regression was used to compare incidence between groups during long-term ART. Incidence rates of TB were stratified according to updated CD4 cell counts (i.e., over each 4-month period).


Patients eligible for inclusion numbered 1480; of these, 11% died during ART treatment, 11% were lost to follow-up, and 6% transferred out. Median follow-up was 2.1 years (range, 1.0-4.5 years). There were a total of 203 incident TB infections (7.3 cases/per 100 person years [p100py]), of which 64% were microbiologically confirmed. At baseline, 448 (30%) persons were considered to have prevalent TB. The number of new TB cases was the highest during the first 4 months of ART at 18.8 cases p100py (95% confidence interval [CI]: 15.2-23.3). In multivariate analysis, the incidence rate during the early ART period remained significantly higher than in the late ART period (adjusted incident rate ratio [IRR]=1.7; 95% CI: 1.1-2.6; P=0.03), even adjusting for a history of TB, baseline CD4 cell count, and updated viral load. During long-term ART, TB incidence was highest during person-time accrued within the CD4 cell stratum of <100 cells/mL (16.7 cases p100py; 95% CI: 12.8-21.6). At CD4 counts of 200-500 cells/mL, TB incidence was lower, but still high (from 5.5-4.2 cases p100py). Rates were not markedly reduced until CD4 cell counts reached >500 cells/mL (1.5 p100py; 95% CI: 0.5-3.5). Patients with the lowest baseline CD4 counts (<100 cells/mL) accrued more person-time within the lower CD4 count strata during follow-up and therefore remained at higher risk for a longer period of time.


This study examined changing CD4 counts during ART therapy and its relationship to TB incidence. Low baseline CD4 counts and "unmasking" of subclinical TB are likely to explain the high burden of TB during the first 4 months of ART. During long-term ART, however, the time spent with low CD4 counts, even 200-500 cells/mL, are associated with significant rates of new TB cases. These rates could be reduced by initiating ART at higher baseline CD4 counts and with more effective identification and treatment of subclinical TB.

Quality Rating

The implementation of this study was of high quality. The authors used a somewhat novel analytic approach by using person-time spent at various CD4 cell-count levels in evaluating incidence of TB. The presentation and interpretation of the results could have been improved by including information on "prevalent" TB cases at baseline, TB treatment success, and deaths due to TB. It was not clear whether baseline "prevalent" cases were those that were identified during screening and subsequently initiated on treatment, and whether ART treatment was delayed as a consequence, or were cases that were already undergoing treatment at the time of cohort enrollment.

Programmatic Implications

These data suggest the potential need for routine microbiological screening for TB prior to starting ART, even among asymptomatic patients. Intensified case finding during the first year of ART and among all those with low CD4 counts are also needed. Current guidelines of CD4 cell counts of <200 or 250 cells/mL for ART initiation in most African countries are not optimal for TB prevention because risk of TB is not substantially reduced until CD4 counts exceed 500 cells/mL. As a result, ongoing TB transmission is likely to occur at the population level, to both HIV-infected and -uninfected persons. Reducing TB burden would require earlier initiation of ART at higher CD4 counts. Use of IPT remains problematic because administration may result in inadvertent monotherapy if subclinical TB is unmasked.