We show that HIV-infected adults who report prior active tuberculosis exhibit a markedly enhanced hazard of subsequent active tuberculosis, suggesting that intensified preventive measures should be directed toward these extremely high-risk individuals.
Keywords: tuberculosis, HIV, recurrent
Abstract
Background. Active tuberculosis is common among human immunodeficiency virus (HIV)–infected persons living in tuberculosis-endemic areas, but the hazard of subsequent tuberculosis disease has not been quantified in a single prospective cohort.
Methods. Among HIV-infected, BCG-immunized adults with CD4 counts ≥200 cells/μL who received placebo in the DarDar tuberculosis vaccine trial in Tanzania, we compared the prospective risk of active tuberculosis between subjects who did and who did not report prior active tuberculosis. All subjects with a positive tuberculin skin test without prior active tuberculosis were offered isoniazid preventive treatment. Definite or probable tuberculosis was diagnosed during active follow-up using rigorous published criteria.
Results. We diagnosed 52 cases of definite and 92 cases of definite/probable tuberculosis among 979 subjects during a median follow-up of 3.2 years. Among the 80 subjects who reported prior active tuberculosis, 11 (13.8%) subsequently developed definite tuberculosis and 17 (21.3%) developed definite/probable tuberculosis, compared with 41 (4.6%) and 75 (8.3%), respectively, of 899 subjects without prior active tuberculosis (definite tuberculosis risk ratio [RR], 3.01; 95% confidence interval [CI], 1.61–5.63, P < .001; definite/probable tuberculosis RR, 2.55; 95% CI, 1.59–4.09, P < .001). In a Cox regression model adjusting for age, CD4 count, and isoniazid receipt, subjects with prior active tuberculosis had substantially greater hazard of subsequent definite tuberculosis (hazard radio [HR], 3.69; 95% CI, 1.79–7.63, P < .001) and definite/probable tuberculosis (HR, 2.78; 95% CI, 1.58–4.87, P < .001).
Conclusions. Compared to subjects without prior tuberculosis, the hazard of active tuberculosis is increased 3-fold among HIV-infected adults with prior active tuberculosis.
Clinical Trials Registration. NCT0052195.
Heimbeck reported in 1930 that student nurses in Oslo with latent tuberculosis infection before starting work in a tuberculosis hospital had a decreased risk of subsequent active tuberculosis compared with classmates with a negative baseline skin test [1]. More recently, Andrews and colleagues estimated that individuals with latent tuberculosis who were newly tuberculosis exposed had a 79% lower risk of active tuberculosis compared to subjects without prior latent tuberculosis [2]. These studies indicate that among immunocompetent persons, latent tuberculosis infection without progression to active tuberculosis is associated with protection from subsequent active tuberculosis on reexposure. In contrast, subjects who have developed active tuberculosis disease once and then have been successfully treated exhibit a 4-fold increased risk of subsequent tuberculosis disease [3]. Taken together, these data in human immunodeficiency virus (HIV)–negative subjects show that although immunologic protection against tuberculosis is generated by infection that does not progress to tuberculosis, persons who fail to contain tuberculosis infection once are at risk to develop active tuberculosis.
HIV infection increases the overall risk of active tuberculosis disease developing after latent tuberculosis infection with a risk of active tuberculosis disease ranging from 3.7 to 16.2 cases per 100 years [4–9]. The risk of active tuberculosis also appears to be increased by a prior episode of active tuberculosis: a study in HIV-infected South African miners with at least 1 episode of active tuberculosis showed that the risk of subsequent active tuberculosis disease was 19.7 cases per 100 years compared with 3.7 cases per 100 years among a similar but distinct cohort of HIV-infected miners without known prior active tuberculosis [9]. However, to our knowledge, no studies among HIV-infected adults have compared the incidence of subsequent active tuberculosis between subjects with and without prior active tuberculosis in a single cohort to calculate of the impact of prior active tuberculosis on the hazard of subsequent tuberculosis.
The DarDar Trial was a 7-year, randomized placebo-controlled phase 3 efficacy trial of a novel tuberculosis booster vaccine in HIV-infected adults [10]. At baseline, 80 (8.2%) subjects reported treatment for prior active tuberculosis. Here we evaluate the hazard of subsequent active tuberculosis among DarDar Trial placebo recipients with and without prior active tuberculosis.
MATERIALS AND METHODS
Study Subjects
Subject enrollment occurred from 2001 to 2005, and follow-up continued through January 2008. Subjects who gave informed consent were eligible for enrollment if they were aged 18 years or older and had 2 positive enzyme-linked immunosorbent assay (ELISA) tests for HIV, a CD4 count ≥200/μL, and a BCG scar. At baseline, all subjects were evaluated with a clinical history (including of prior active tuberculosis treatment), physical examination, tuberculin skin test (TST), single-view chest radiograph, 3 sputum samples for acid-fast bacilli (AFB) smear, and sputum mycobacterial culture. Subjects with active tuberculosis at baseline were excluded from enrollment. TST-positive subjects without a history of treatment for prior active tuberculosis were administered isoniazid for 6 months. Subjects were randomized 1:1 to receive 5 intradermal doses of active vaccine or borate-buffered saline placebo [10, 11]. For the present study, we confined our analyses to subjects who received placebo.
Human Research Conduct
Human experimentation guidelines of the US Department of Health and Human Services, as well as that of the Committee for the Protection of Human Subjects at Dartmouth College and the Research Ethics Committee of the Muhimbili University of Health and Allied Sciences, were followed in the conduct of this research.
Tuberculin Skin Test
Tuberculin (0.1 mL, RT-23, Statens Serum Institut, Copenhagen, Denmark) was injected intradermally on the forearm, and trained personnel measured any resultant induration at 48–72 hours. Subjects with positive reactions of ≥5 mm were offered isoniazid (INH) therapy unless the subject had been treated for prior active tuberculosis. Among subjects with positive reactions of ≥5 mm, 88% completed 6 months of INH [12]. In inadvertent deviation from the study protocol, 2 subjects completed 6 months of INH therapy despite a history of prior active tuberculosis.
Surveillance for Tuberculosis Disease
We evaluated subjects for active tuberculosis disease via physical examination and history at months 2, 4, and 6 months and every 3 months thereafter. In addition, subjects with ≥2 weeks of fever, cough, or weight loss underwent evaluation for active tuberculosis via a single-view chest radiograph, 3 sputum collections for AFB smear and mycobacterial culture, phlebotomy for mycobacterial blood culture, and additional studies as clinically indicated (eg, cultures of other body fluids or tissue biopsies).
Definitions of Tuberculosis
A 3-person, blinded adjudication panel reviewed all episodes of illness evaluated for active tuberculosis and designated subjects as having definite or probable tuberculosis according to rigorous study definitions (Table 1) [10]. Information about subject prior active tuberculosis status was not provided to the adjudication panel.
Table 1.
DarDar Tuberculosis Definitions
| Definite tuberculosis | 1. One or more sputum cultures positive for MTB with ≥10 CFU; or |
| 2. Two or more sputum cultures with 1–9 CFU of MTB (indeterminate MTB culture); or | |
| 3. Two or more positive sputum smears for AFB; or | |
| 4. One or more cultures for MTB from the blood or other sterile body site. | |
| Probable tuberculosis | 1. Positive chest radiographa plus either a. 1 positive sputum AFB smear or b. 1 indeterminate MTB culture result; or 2. Clinical symptoms/signs plus either a. 1 positive sputum AFB smear or b. an indeterminate MTB culture result; or 3. Clinical symptoms/signs and a positive radiograph plus a response to antituberculosis therapy; or 4. One positive sputum AFB smear from a sterile site plus clinical symptoms/signs of tuberculosis; or 5. Caseous necrosis on a tissue biopsy. |
Abbreviations: AFB, acid-fast bacilli; CFU, colony-forming units; MTB, Mycobacterium tuberculosis.
a A positive chest radiograph was defined as one which exhibited, compared with a baseline film, new pulmonary infiltrates, cavitation, pleural effusion, pulmonary fibrosis, or intrathoracic lymphadenopathy.
Microbiology
Subjects provided sputum samples via spontaneous expectoration or induction for AFB stain and mycobacterial culture. We employed standard culture techniques including culture on Lowenstein-Jensen medium and an automated mycobacterial blood culture (MB BacT, bioMérieux, Durham, North Carolina). Putative Mycobacterium tuberculosis isolates were inoculated in duplicate into 7H9 liquid medium supplemented with oleic acid, albumin, dextrose, and catalase and 15% glycerol, frozen at –70°C and shipped to the United States where M. tuberculosis complex was confirmed by DNA probes (Accuprobe, Gen-Probe, San Diego, California). Isolates not DNA probe-positive for M. tuberculosis complex were not included in these analysis.
Immunological Assays
We measured immune responses to mycobacterial antigens using 3 assays: we assessed interferon gamma (IFN-γ) responses to mycobacterial antigens using a standard ELISA, we assessed lymphocyte proliferation to mycobacterial antigens via a standard tritiated thymidine lymphocyte proliferation assay (LPA), and we measured antibody responses to the mycobacterial glycolipid lipoarabinomannan (LAM) by ELISA. Assay conduct is described in detail elsewhere [13, 14]. In brief, we incubated freshly isolated and ficolled peripheral blood mononuclear cells (PBMCs) with study antigens for 5 days in parallel IFN-γ and LPA assays. At the end of 5 days, centrifuged cell supernatants were frozen and sent to the United States for later IFN-γ level measurement using a standard IFN-γ ELISA (R&D Systems, Minneapolis, Minnesota). We assessed IFN-γ responses to medium alone (negative control), phytohemagglutinin (positive control), 1 μg/mL M. tuberculosis ESAT-6, 0.5 μg/mL M. tuberculosis Ag85, or 0.5 μg/mL M. tuberculosis whole cell lysate, with all antigens acquired through the National Institutes of Health (see Acknowledgments). LPA assays were conducted on the same PBMCs with the same antigens used in the IFN-γ assay, with cells harvested onto filter paper and sent to the National Public Health Institute in Helsinki, Finland, for data acquisition on a scintillation counter. To measure antibody responses, serum diluted 1:100 was assayed via ELISA for immunoglobulin G antibodies to LAM using an in-house assay.
HIV RNA Testing
A preplanned subset of 341 placebo subjects underwent testing for HIV RNA after each dose of vaccine. Plasma was collected, frozen, and shipped to the United States, and HIV RNA was measured using the Bayer Versant HIV-1 RNA 3.0 (bDNA) assay.
Statistical Analysis
We compared demographic data and immune response data between subjects with and without prior active tuberculosis using the 2-tailed Student t test, Mann-Whitney U test, and Fisher exact test as appropriate with a P value threshold for significance of <.05. We then constructed a multivariable Cox proportional hazards regression model of the hazard of tuberculosis among subjects with and without prior active tuberculosis during prospective follow-up, adjusting for cofactors such as age, baseline CD4 count, INH completion, TST status, HIV RNA, and antiretroviral treatment status as detailed in the multivariate model results. We confirmed that the proportional hazards assumption was not violated using log-log plots and Schoenfeld residuals. We analyzed the data using Stata software version 9 (StataCorp, College Station, Texas).
RESULTS
Subject Characteristics
Among the 979 enrolled subjects assigned to the placebo arm of the trial, 80 (8.2%) reported prior active tuberculosis. At baseline, subjects with and without such a history had similar clinical and demographic characteristics (Table 2), including the frequency of positive TST. Although subjects with prior active tuberculosis were more likely to be on antiretroviral therapy (ART), the proportion of subjects on ART was small and no cases of active tuberculosis developed in subjects on ART.
Table 2.
Baseline Demographic and Clinical Characteristics of Subjects According to Prior Active Tuberculosis History and According to the Subsequent Development of Definite Tuberculosis, Definite/Probable Tuberculosis, or No Tuberculosis During Prospective Follow-up
| Prior Tuberculosis History |
Study Tuberculosis Diagnosis |
||||||
|---|---|---|---|---|---|---|---|
| Prior Active Tuberculosis (n = 80) | No Prior Active Tuberculosis (n = 899) | P Value* | Definite Tuberculosis (n = 52) | Definite/Probable Tuberculosis (n = 92) | No Tuberculosis (n = 887) | P Value** | |
| Age, median, years | 36.4 | 32.0 | <.001 | 33.5 | 33 | 32 | .145 |
| Male, No. (%) | 20 (25.0) | 217 (24.1) | .863 | 13 (25.0) | 27 (29.3) | 210 (23.7) | .260 |
| CD4 count, median (25%–75%) cells/µL | 347 (262–549) | 407 (302–583) | .021 | 349 (271–451) | 321 (255–426) | 415 (306–591) | <.001 |
| HIV RNA, mean log10 copies/mL (n = 341) | 4.1 | 3.9 | .416 | 4.3 | 4.4 | 3.9 | .001 |
| ART, No. (%) | 30 (39.0) | 282 (31.9) | .204 | 0 (0) | 0 (0) | 33 (3.7) | <.001 |
| TST positive, No. (%) | 6 (7.5) | 27 (3.0) | .033 | 29 (56.9) | 44 (48.9) | 268 (30.8) | .170 |
| Completed 180 days of INH, No. (%) | 2 (2.5) | 243 (27.0) | .614 | 20 (38.5) | 29 (31.5) | 216 (24.4) | .072 |
| Prior active tuberculosis, No. (%) | 80 (100%) | 0 (0%) | … | 11 (21.2) | 17 (18.5) | 63 (7.1) | <.001 |
Abbreviations: ART, antiretroviral therapy; HIV, human immunodeficiency virus; INH, isoniazid; TST, tuberculin skin test.
*P values derived via Fisher exact test, 2-tailed Student t test, or Mann-Whitney U test as appropriate.
**P values derived via χ2 test or Kruskal-Wallis test as appropriate.
Incidence of Tuberculosis
During prospective follow-up, we diagnosed definite tuberculosis in 52 (5.3%) subjects and definite/probable tuberculosis in 92 (9.4%) subjects. The median time from study enrollment to a study tuberculosis diagnosis was similar in subjects with and without a history of prior active tuberculosis (definite tuberculosis, 23.4 vs 28.6 months, P = .335; definite/probable tuberculosis, 21.8 vs 26.3 months, P = .376). Among subjects with prior active tuberculosis, the median time between prior active tuberculosis and study tuberculosis diagnosis was 108.3 months (95% confidence interval [CI], 25.9–534.9 months) for definite tuberculosis and 68.4 months (95% CI, 23.0–451.3 months) for definite/probable tuberculosis. At baseline, subjects who subsequently developed active tuberculosis had lower CD4 counts, higher HIV RNA, a higher positive TST prevalence, and a higher rate of prior active tuberculosis (Table 2). Definite tuberculosis was diagnosed in 11 of 80 subjects (13.8%) with prior active tuberculosis (4.57 cases per 100 years) compared with 41 of 899 subjects (4.6%) without prior active tuberculosis (1.43 cases per 100 years; risk ratio [RR], 3.01; 95% CI, 1.61–5.63, P < .001). Definite/probable tuberculosis was diagnosed in 17 of 80 patients (21.3%) with prior active tuberculosis (7.42 cases per 100 years) and 75 of 899 patients (8.3%) without prior active tuberculosis (2.14 cases per 100 years; RR, 2.55; 95% CI, 1.59–4.09, P < .001). The significant association between prior active tuberculosis and the risk of subsequent definite and definite/probable tuberculosis persisted among subjects who did or did not receive INH and among subjects with positive and negative baseline TST status, and prior active tuberculosis was also associated with increased risk of probable tuberculosis alone (data not shown).
Immune Responses in Subjects With or Without Prior Active Tuberculosis
There were no differences in baseline IFN-γ or LPA responses to the mycobacterial antigens Ag85 and ESAT-6 and whole cell lysate between subjects who did or did not report prior active tuberculosis, nor were there differences in antibody responses to LAM between the 2 groups (data not shown).
Multivariable Model
We conducted a multivariable Cox regression analysis of the relation of prior active tuberculosis and other covariates with the hazard of subsequent definite and definite/probable tuberculosis (Table 3). Compared to subjects without prior active tuberculosis, subjects with prior active tuberculosis experienced significant elevated hazard of both definite tuberculosis (hazard ratio [HR], 2.99–3.93) and definite/probable tuberculosis (HR, 2.54–3.31). Adjustment in the multivariable model for INH receipt (or TST status) had little impact on the elevated hazard of subsequent tuberculosis among subjects with prior active tuberculosis.
Table 3.
Multivariable Cox Regression Model of the Hazard of Subsequent Tuberculosis Among Subjects With Prior Active Tuberculosis
| Variable | No. | Hazard of Definite Tuberculosis | 95% CI for HR | P Value | Hazard of Definite/Probable Tuberculosis | 95% CI for HR | P Value |
|---|---|---|---|---|---|---|---|
| Univariate analyses | |||||||
| Age, each 10-year increment | 979 | 1.24 | .88–1.76 | .213 | 1.19 | .91–1.55 | .200 |
| Baseline CD4 count, each 100 cells/µL increment | 977 | 0.76 | .64–.90 | .002 | 0.68 | .59–.79 | <.001 |
| Baseline HIV RNA | 341 | 2.21 | 1.31–3.71 | .003 | 2.45 | 1.64–3.66 | <.001 |
| Baseline ART | 979 | 0 | a | 1.0 | 0 | a | 1.0 |
| TST positive | 961 | 2.44 | 1.40–4.26 | .002 | 1.78 | 1.17–2.69 | .007 |
| Prior active tuberculosis, unadjusted | 979 | 3.27 | 1.68–6.36 | <.001 | 2.82 | 1.66–4.77 | <.001 |
| Multivariable analyses | |||||||
| Prior active tuberculosis, adjusted for age, CD4 count, completed 180 days of INH | 977 | 3.69 | 1.79–7.63 | <.001 | 2.78 | 1.58–4.87 | <.001 |
| Prior active tuberculosis, adjusted for age, CD4 count, completed 180 days of INH, ART | 977 | 3.76 | 1.81–7.77 | <.001 | 2.85 | 1.62–5.01 | <.001 |
| Prior active tuberculosis, adjusted for age, CD4 count, completed 180 days of INH, HIV RNA | 341 | 3.93 | 1.47–10.50 | .006 | 3.31 | 1.55–7.08 | .002 |
| Prior active tuberculosis, adjusted for age, CD4 count, TST status | 959 | 2.99 | 1.50–5.95 | .002 | 2.54 | 1.48–4.36 | .001 |
| Prior active tuberculosis, adjusted for age, CD4 count, TST status, ART | 959 | 3.01 | 1.51–6.02 | .002 | 2.59 | 1.51–4.45 | .001 |
| Prior active tuberculosis, adjusted for age, CD4 count, TST status, ART, HIV RNA | 334 | 3.32 | 1.29–8.56 | .013 | 2.98 | 1.44–6.14 | .003 |
Abbreviations: ART, antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio; INH, isoniazid; TST, tuberculin skin test.
a No subjects on ART developed definite or definite/probable tuberculosis, such that 95% CIs cannot be calculated.
Because subjects with a positive TST did not receive 6 months of INH prophylaxis if they had prior active tuberculosis (with 2 exceptions), we conducted supplemental analyses among the 703 subjects who did not receive INH to determine if lack of INH receipt was the basis for the increased risk of a second episode of tuberculosis among subjects with prior active tuberculosis (Table 4). In these supplemental analyses, subjects with prior active tuberculosis exhibited significantly elevated hazard of subsequent definite tuberculosis and definite/probable tuberculosis. Last, in additional analyses among the lower-risk cohort of 649 subjects with a negative TST, we confirmed the association between prior active tuberculosis and an increased hazard of subsequent definite and definite/probable tuberculosis (Table 5).
Table 4.
Multivariable Cox Regression Model of the Hazard of Subsequent Tuberculosis Among Subjects With Prior Active Tuberculosis Among the 703 Subjects Who Did Not Receive Isoniazid Preventive Therapy (78 With Prior Active Tuberculosis, 625 With No Prior Active Tuberculosis)
| Variable | No. | Hazard of Definite Tuberculosis | 95% CI for HR | P Value | Hazard of Definite/Probable Tuberculosis | 95% CI for HR | P Value |
|---|---|---|---|---|---|---|---|
| Univariate analyses | |||||||
| Age, each 10-year increment | 703 | 1.10 | .68–1.76 | .703 | 1.02 | .72–1.44 | .912 |
| Baseline CD4 count, each 100 cells/µL increment | 701 | 0.76 | .60–.96 | .020 | 0.66 | .54–.81 | <.001 |
| Baseline HIV RNA | 230 | 2.19 | 1.08–4.45 | .030 | 2.24 | 1.34–3.75 | .002 |
| Baseline ART | 703 | 0 | a | 1.0 | 0 | a | 1.0 |
| TST positive | 685 | 5.80 | 2.47–13.59 | <.001 | 3.53 | 1.72–7.22 | .001 |
| Prior active tuberculosis, unadjusted | 703 | 4.93 | 2.35–10.37 | <.001 | 3.31 | 1.86–5.91 | <.001 |
| Multivariable analyses | |||||||
| Prior active tuberculosis, adjusted for age, CD4 count | 701 | 4.79 | 2.23–10.28 | <.001 | 3.07 | 1.69–5.56 | <.001 |
| Prior active tuberculosis, adjusted for age, CD4 count, ART | 701 | 4.91 | 2.29–10.56 | <.001 | 3.15 | 1.74–5.73 | <.001 |
| Prior active tuberculosis, adjusted for age, CD4 count, ART, HIV RNA | 230 | 6.38 | 2.14–19.02 | .001 | 3.91 | 1.71–8.94 | .001 |
Abbreviations: ART, antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio; TST, tuberculin skin test.
a No subjects on ART developed definite or definite/probable tuberculosis, such that 95% CIs cannot be calculated.
Table 5.
Multivariable Cox Regression Model of the Hazard of Subsequent Tuberculosis Among 649 Subjects With a Negative Baseline Tuberculin Skin Test (47 With Prior Active Tuberculosis, 602 With No Prior Active Tuberculosis)
| Variable | No. | Hazard of Definite Tuberculosis | 95% CI for HR | P Value | Hazard of Definite/Probable Tuberculosis | 95% CI for HR | P Value |
|---|---|---|---|---|---|---|---|
| Univariate analyses | |||||||
| Age, each 10-year increment | 649 | 1.08 | .61–1.90 | .792 | 1.04 | .70–1.54 | .842 |
| Baseline CD4 count, each 100 cells/µL increment | 647 | 0.78 | .60–1.03 | .078 | 0.64 | .51–.81 | <.001 |
| Baseline HIV RNA | 207 | 1.94 | .86–4.36 | .109 | 2.29 | 1.26–4.17 | .006 |
| Baseline ART | 649 | 0 | a | 1.0 | 0 | a | 1.0 |
| Prior active tuberculosis, unadjusted | 649 | 3.26 | 1.10–9.66 | .033 | 3.07 | 1.43–6.60 | .004 |
| Multivariable analyses | |||||||
| Prior active tuberculosis, adjusted for age, CD4 count | 647 | 2.69 | .89–8.18 | .080 | 2.22 | 1.02–4.84 | .046 |
| Prior active tuberculosis, adjusted for age, CD4 count, ART | 647 | 2.85 | .94–8.66 | .064 | 2.34 | 1.07–5.12 | .033 |
| Prior active tuberculosis, adjusted for age, CD4 count, ART, HIV RNA | 207 | 2.84 | .58–13.90 | .198 | 2.88 | .93–8.93 | .067 |
Abbreviations: ART, antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; HR, hazard ratio.
a No subjects on ART developed definite or definite/probable tuberculosis, such that 95% CIs cannot be calculated.
DISCUSSION
Among BCG-immunized, HIV-infected adults living in a high-tuberculosis-burden country, we found that the hazard of future tuberculosis was increased 3-fold among those with a history of prior active tuberculosis. This enhanced hazard of future tuberculosis among subjects with prior active tuberculosis was robust and consistent across multivariable analyses adjusted for several key clinical factors.
Classically, Heimbeck showed that prior immune containment of latent tuberculosis among immunocompetent adults is associated with a decreased risk of incident active tuberculosis in a high-risk exposure setting [1]. In contrast, among the HIV-infected subjects in our study, subjects with latent tuberculosis had an elevated hazard of developing tuberculosis. This hazard was even greater, independently, among subjects with prior active tuberculosis [10]. These findings suggest that among immunocompromised patients living where tuberculosis is highly endemic, failure to contain tuberculosis infection once is associated with substantial risk of future active tuberculosis disease.
Cost and infrastructure requirements are major impediments to the implementation of novel laboratory-based measures such as IFN-γ release assays to assess tuberculosis risk and thus prioritize preventive interventions [15]. Here we show that an inexpensive simple measure—a clinical history of prior active tuberculosis—identified HIV-infected adults at markedly increased risk of developing active tuberculosis in a fashion that was independent of known measures like the TST and at least as predictive. This finding should trigger prioritized delivery of interventions designed to prevent the development of tuberculosis, such as isoniazid preventive therapy (IPT) and ART, to HIV-infected adults with a clinical history of prior active tuberculosis, an expansion of existing recommendations to prevent tuberculosis via the provision of ART to HIV-infected adults during active tuberculosis [16–20].
In previous analyses of the placebo subjects from the DarDar trial, we observed that IFN-γ responses to mycobacterial antigens at baseline were associated with a 60% reduction in the hazard of subsequent definite/probable tuberculosis [13]. We did not observe, however, differences in IFN-γ or other immune responses between subjects with or without prior active tuberculosis, suggesting that either we did not measure the relevant immunological parameter or that nonimmunological factors underlie the observed association.
Our study is the first to formally quantify the hazard of active tuberculosis between subjects with and without a history of prior active tuberculosis in a single well-characterized cohort. We assessed baseline demographic and immunological characteristics in a phase 3 clinical trial setting, and assessed tuberculosis outcomes in active prospective follow-up using M. tuberculosis culture in our trial diagnostic protocol. However, our study has important limitations. First, we defined prior active tuberculosis status solely on the basis of subject history and not via direct microbiological confirmation of the original tuberculosis diagnosis. Patient history is, however, by far the most commonly available clinical marker of prior active tuberculosis, allowing better generalization to modern-day clinical practice in the developing world. Second, although there was a clear relationship between the risk of subsequent tuberculosis and the baseline HIV load, we assessed the HIV load only in a preplanned subset of subjects and so could not include this key covariate in the full multivariable model. In the 334 subjects in whom HIV RNA data were available, multivariable analyses adjusting for HIV RNA showed that the risk of subsequent HIV-associated tuberculosis remained substantially elevated among subjects with prior active tuberculosis. Last, all subjects with prior active tuberculosis were given ART, and no tuberculosis developed among subjects on ART, an imbalance in ART administration between study groups that would bias against the finding of an association between prior active tuberculosis and an elevated hazard of future tuberculosis.
Our protocol for the administration of IPT is a potential confounder of this analysis. In contrast to 2011 WHO guidelines advocating the provision of IPT to all HIV-infected adults regardless of TST status [21], in our 2001–2008 study we administered INH only to TST-positive subjects who had not been treated for prior active tuberculosis [12]. It is conceivable that this contributed to the higher subsequent incidence of tuberculosis among subjects with prior active tuberculosis. However, incorporation of INH receipt or TST status into our main multivariable model had only a trivial effect on the increased hazard of tuberculosis among subjects with prior active tuberculosis, and the association between prior active tuberculosis and increased hazard of subsequent tuberculosis persisted in supplemental analyses conducted solely in subjects who did not receive INH or who had a negative TST. Notably, the latter analyses had smaller HRs and borderline P values, likely owing to the lower number of tuberculosis cases in this lower-risk subset of subjects. Taken together, our data suggest that lack of INH receipt and TST status were not major drivers of the strong and consistent association between prior active tuberculosis and the risk of subsequent HIV-associated tuberculosis.
Although subjects with prior active tuberculosis exhibited dramatically elevated susceptibility to recurrent tuberculosis, we believe this was not a consequence of progression of undetected subclinical tuberculosis at baseline or of relapse of previously treated disease. Subjects were screened comprehensively at study enrollment for the presence of active tuberculosis including via sputum and blood cultures even in the absence of suggestive symptomatology, an approach that allowed us to first identify the entity of subclinical tuberculosis in HIV [22]. Moreover, the median time between the prior episode of active tuberculosis and study diagnosis of tuberculosis was >5 years, and IS6110 analyses from our cohort of tuberculosis cases suggests that the majority were from new infection, not reactivation of tuberculosis [23]. These findings support our hypothesis that the major driver of high incidence of recurrent tuberculosis among subjects with prior active tuberculosis is failure to contain tuberculosis to which they have been newly exposed.
The novel confirmation of the independent and substantial association between prior active tuberculosis and subsequent tuberculosis among HIV-infected adults in a tuberculosis endemic area is clinically important: it identifies a specific subset of patients for whom intensified interventions to prevent tuberculosis should be considered. Given the clear preventive benefit of ART either alone or in combination with INH [16–20], we recommend that both IPT and ART be provided to HIV-infected adults on the basis of a clinical history of prior active tuberculosis. Since HIV-infected persons may have subclinical active tuberculosis with minimal or absent clinical and radiologic findings [22], we recommend a comprehensive tuberculosis diagnostic workup at the onset of ART and ongoing surveillance for tuberculosis with regular clinical examinations and laboratory studies for this group of high-risk patients with prior active tuberculosis.
Notes
Acknowledgments. We thank Wendy Wieland-Alter, Outi Rautio, and Betty Mchaki for their skillful conduct of the immunological assays in this study, and Sue Tvaroha for excellent database management. All antigens were acquired through the National Institutes of Health (NIH), National Institute of Allergy and Infectious Diseases (contract number HHSN266200400091C, “Tuberculosis Vaccine Testing and Research Materials,” awarded to Colorado State University.
Financial support. This work was supported by the NIH, Division of AIDS (grant number AI 45407 to C. F. vR.), the Fogarty International Center (D43-TW006807 to C. F. vR. and C. R. H.); and the NIH, National Center for Research Resources, Centers for Biomedical Research Excellence (5P20RR016437-08 to T. L.).
Potential conflicts of interest. All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1.Heimbeck J. The infection of tuberculosis. Acta Medica Scandinavica. 1930;74:143–55. [Google Scholar]
- 2.Andrews JR, Noubary F, Walensky RP, Cerda R, Losina E, Horsburgh CR. Risk of progression to active tuberculosis following reinfection with Mycobacterium tuberculosis. Clin Infect Dis. 2012;54:784–91. doi: 10.1093/cid/cir951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Verver S, Warren RM, Beyers N, et al. Rate of reinfection tuberculosis after successful treatment is higher than rate of new tuberculosis. Am J Resp Crit Care Med. 2005;171:1430–5. doi: 10.1164/rccm.200409-1200OC. [DOI] [PubMed] [Google Scholar]
- 4.Guelar A, Gatell JM, Verdejo J, et al. A prospective study of the risk of tuberculosis among HIV-infected patients. AIDS. 1993;7:1345–9. doi: 10.1097/00002030-199310000-00007. [DOI] [PubMed] [Google Scholar]
- 5.Moreno S, Baraia-Etxaburu J, Bouza E, et al. Risk for developing tuberculosis among anergic patients infected with HIV. Ann Intern Med. 1993;119:194–8. doi: 10.7326/0003-4819-119-3-199308010-00003. [DOI] [PubMed] [Google Scholar]
- 6.Antonucci G, Girardi E, Raviglione MC, Ippolito G. Risk factors for tuberculosis in HIV-infected persons: a prospective cohort study. The Gruppo Italiano di Studio Tubercolosi e AIDS (GISTA) JAMA. 1995;274:143–8. doi: 10.1001/jama.274.2.143. [DOI] [PubMed] [Google Scholar]
- 7.Allen S, Batungwanayo J, Kerlikowske K, et al. Two-year incidence of tuberculosis in cohorts of HIV-infected and uninfected urban Rwandan women. Am Rev Resp Dis. 1992;146:1439–44. doi: 10.1164/ajrccm/146.6.1439. [DOI] [PubMed] [Google Scholar]
- 8.Selwyn PA, Hartel D, Lewis VA, et al. A prospective study of the risk of tuberculosis among intravenous drug users with human immunodeficiency virus infection. New Engl J Med. 1989;320:545–50. doi: 10.1056/NEJM198903023200901. [DOI] [PubMed] [Google Scholar]
- 9.Glynn JR, Murray J, Bester A, Nelson G, Shearer S, Sonnenberg P. High rates of recurrence in HIV-infected and HIV-uninfected patients with tuberculosis. J Infect Dis. 2010;201:704–11. doi: 10.1086/650529. [DOI] [PubMed] [Google Scholar]
- 10.von Reyn CF, Mtei L, Arbeit RD, et al. Prevention of tuberculosis in bacille Calmette-Guerin-primed, HIV-infected adults boosted with an inactivated whole-cell mycobacterial vaccine. AIDS. 2010;24:675–85. doi: 10.1097/QAD.0b013e3283350f1b. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Vuola JM, Ristola MA, Cole B, et al. Immunogenicity of an inactivated mycobacterial vaccine for the prevention of HIV-associated tuberculosis: a randomized, controlled trial. AIDS. 2003;17:2351–5. doi: 10.1097/00002030-200311070-00010. [DOI] [PubMed] [Google Scholar]
- 12.Munseri PJ, Talbot EA, Mtei L, von Reyn CF. Completion of isoniazid preventive therapy among HIV-infected patients in Tanzania. Int J Tuberc Lung Dis. 2008;12:1037–41. [PubMed] [Google Scholar]
- 13.Lahey T, Sheth S, Matee M, et al. Interferon gamma responses to mycobacterial antigens protect against subsequent HIV-associated tuberculosis. J Infect Dis. 2010;202:1265–72. doi: 10.1086/656332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lahey T, Arbeit RD, Bakari M, et al. Immunogenicity of a protective whole cell mycobacterial vaccine in HIV-infected adults: a phase III study in Tanzania. Vaccine. 2010;28:7652–8. doi: 10.1016/j.vaccine.2010.09.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Stop TB Partnership. New technologies for tuberculosis control: a framework for their adoption, introduction and implementation. Geneva, Switzerland: World Health Organization; 2007. [Google Scholar]
- 16.Lawn SD, Myer L, Bekker LG, Wood R. Burden of tuberculosis in an antiretroviral treatment programme in sub-Saharan Africa: impact on treatment outcomes and implications for tuberculosis control. AIDS. 2006;20:1605–12. doi: 10.1097/01.aids.0000238406.93249.cd. [DOI] [PubMed] [Google Scholar]
- 17.Tabarsi P, Saber-Tehrani AS, Baghaei P, et al. Early initiation of antiretroviral therapy results in decreased morbidity and mortality among patients with TB and HIV. J Int AIDS Soc. 2009;12:14. doi: 10.1186/1758-2652-12-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Golub JE, Durovni B, King BS, et al. Recurrent tuberculosis in HIV-infected patients in Rio de Janeiro, Brazil. AIDS. 2008;22:2527–33. doi: 10.1097/QAD.0b013e328311ac4e. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Lawn SD, Wood R, De Cock KM, Kranzer K, Lewis JJ, Churchyard GJ. Antiretrovirals and isoniazid preventive therapy in the prevention of HIV-associated tuberculosis in settings with limited health-care resources. Lancet Infect Dis. 2010;10:489–98. doi: 10.1016/S1473-3099(10)70078-5. [DOI] [PubMed] [Google Scholar]
- 20.Golub JE, Saraceni V, Cavalcante SC, et al. The impact of antiretroviral therapy and isoniazid preventive therapy on tuberculosis incidence in HIV-infected patients in Rio de Janeiro, Brazil. AIDS. 2007;21:1441–8. doi: 10.1097/QAD.0b013e328216f441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.World Health Organization. Guidelines for intensified tuberculosis case-finding and isoniazid preventive therapy for people living with HIV in resource-constrained settings. Geneva, Switzerland: Division of HIV/AIDS, Stop TB Department, WHO; 2011. [Google Scholar]
- 22.Mtei L, Matee M, Herfort O, et al. High rates of clinical and subclinical tuberculosis among HIV-infected ambulatory subjects in Tanzania. Clin Infect Dis. 2005;40:1500–7. doi: 10.1086/429825. [DOI] [PubMed] [Google Scholar]
- 23.Adams LV, Kreiswirth BN, Arbeit RD, et al. Molecular epidemiology of HIV-associated tuberculosis in Dar es Salaam, Tanzania: strain predominance, clustering, and polyclonal disease. J Clin Microbiol. 2012;50:2645–50. doi: 10.1128/JCM.00624-12. [DOI] [PMC free article] [PubMed] [Google Scholar]