Despite World Health Organization recommendations  and strong evidence of effectiveness [2–4], implementation of isoniazid preventive therapy (IPT) to reduce HIV-associated tuberculosis (TB) has been limited. Reasons include concerns about promoting drug resistance, although there is little evidence that this occurs .
TB incidence in South Africa was estimated at 948/100 000 in 2007  and is higher still among miners, partly due to high prevalence of HIV (estimated at 29%)  and silicosis . The prevalence of latent TB infection among gold miners was recently estimated at 89% . We are conducting a cluster-randomized trial (‘Thibela TB’) of community-wide IPT among gold miners in South Africa. The aim of this analysis was to describe characteristics of TB disease in individuals previously exposed to IPT in this trial.
Mining companies provide employees with free, on-site comprehensive healthcare, including TB and HIV services. In the Thibela TB study, clusters (all employees at a single or group of mine shafts) were randomized to control (routine TB control, including annual case finding by chest radiograph and targeted IPT offered to individuals with HIV or silicosis) or intervention (routine TB control as above plus the study intervention, i.e. TB screening and IPT offered to all employees without a specific contraindication, regardless of perceived TB risk or HIV status) arms. In intervention clusters, screening for active TB and other contraindications is by symptom questionnaire and chest radiograph [7,10,11]. Eligible participants are dispensed isoniazid 300 mg and pyridoxine 25 mg daily, self-administered, for 30 days initially and 9 months in total, with monthly review by nurses at workplace-based study clinics for TB symptoms and adverse events.
Study population and case ascertainment
Individuals were included in this case series (‘TB after IPT’ group) if they were dispensed IPT at one of eight intervention clusters (starting July 2006), attended at least one follow-up visit and subsequently started TB treatment (up to 16 February 2009). Cases were identified through surveillance of incident TB as part of Thibela TB, through a concurrent mycobacterial culture substudy and through review of clinical records for individuals stopping IPT early. We included all those treated for TB, who attended at least one follow-up visit (to be reasonably sure that some isoniazid had been taken), unless an outcome of ‘not TB’ was recorded or only nontuberculous mycobacteria were isolated from sputum.
Sites of disease, episode type, treatment regimens and outcomes (standard WHO definitions ) were abstracted from TB treatment records. We estimated the maximum duration of IPT using the number of monthly dispensing visits. Participants stopping IPT before 9 months were assumed to have taken medication for 15/30 days preceding their final visit.
To provide comparisons for the proportion with drug resistance in the TB after IPT group, we used two data sources:
- ‘Control clusters’: TB case ascertainment in the seven control clusters was identical to that in intervention clusters. All TB cases with available drug susceptibility data from two large control clusters were included in this comparison group.
- ‘Laboratory substudy’: sputa from individuals presenting to clinics with symptoms but no prior history of TB underwent culture and drug susceptibility testing . All TB cases in the laboratory substudy at any control cluster were included.
The TB after IPT group included those who had not yet completed treatment in order to avoid excluding those with longer treatment duration; that is retreatment cases or drug-resistant TB. This was not possible in the control cluster comparison group, as laboratory data are not abstracted until the end of treatment. In the laboratory substudy, this was not an issue, as specimens were collected at the time of investigation for TB and thus drug susceptibility data were available in real time.
Data analysis using STATA version 10 (Stata Corporation, College Station, Texas, USA) included 95% confidence intervals (CIs) (binomial exact method) for proportions of isolates with drug resistance.
‘Thibela TB’ was approved by Research Ethics Committees of the University of KwaZulu Natal and the London School of Hygiene and Tropical Medicine.
One hundred and twenty-six individuals fulfilled inclusion criteria for the TB after IPT group, from a total of 23 095 individuals starting IPT in Thibela TB up to 16 February 2009. Median age was 43 [interquartile range (IQR) 38–47] years and 125 (99%) were men; this is consistent with workforce demographics. In control clusters, 11 of 275 (4%) individuals had evidence from medical records of ever having had IPT.
Description of tuberculosis after isoniazid preventive therapy group
Seventy-seven of 126 sputum samples were cultured for Mycobacterium tuberculosis; 42 were culture-negative and seven had no culture results available. The median estimated duration of IPT was 105 days (IQR 45–195), with 28 of 126 (22%) of this group completing all 270 days. The median time from starting IPT to starting TB treatment was 316 days (IQR 174–491) and 53 of 126 (42%) started TB treatment within the planned 270 days IPT.
Eighty-nine of 103 (86%) with known status were HIV positive. Median CD4 cell count was 196 cells/μl (IQR 81–296) (n = 51). Twenty-one of 89 (24%) HIV-positive individuals were known to be taking antiretroviral therapy at the start of TB treatment.
Ninety-four of 126 (74.6%) were first TB episodes and 87 of 126 (69.0%) were pulmonary. Forty-three (34.1%) were smear and culture positive and 11 (8.7%) were smear positive but culture negative.
In 18 of 126, treatment was ongoing at the time of data collection and outcome not yet recorded. Among the remaining 108 of 126, 64 of 108 (59.3%) had documented cure or treatment completion, 33 (30.6%) were transferred out or had unknown outcome, two of 108 had treatment failure or interruption. There were nine deaths (8.3%), four within the first 2 months of TB treatment; all eight with known status were HIV-positive; median CD4 cell count was 124 cells/μl (n = 6). Five of the deaths had culture-positive M. tuberculosis and all four cases with susceptibility results were susceptible to isoniazid and rifampicin.
Drug susceptibility: tuberculosis after isoniazid preventive therapy and comparison groups
Of 77 M. tuberculosis isolates in the TB after IPT group, 71 (92.2%) had susceptibility testing results for isoniazid and rifampicin. None of the five with concurrent nontuberculous mycobacterial isolates had susceptibility data. In control clusters, of 319 M. tuberculosis isolates, 275 (86.2%) had susceptibility test results for isoniazid and rifampicin (Table 1).
Among first TB episodes, seven of 58 (12.1%; 95% CI 5.0–23.3) were resistant to isoniazid in the TB after IPT group, compared with 12 of 200 (6.0%; 95% CI 3.1–10.2) in the control clusters and 32 of 270 (11.8%; 95% CI 8.2–16.3) in the laboratory substudy (Fig. 1). For retreatment episodes, isoniazid resistance occurred in one of 13 (7.7%; 95% CI 0.2–36.0) in the TB after IPT group and 14 of 75 (18.7; 95% CI 10.6–29.3) in control clusters.
Tuberculosis screening failures in the tuberculosis after isoniazid preventive therapy group
Four individuals most likely had TB that was missed at screening before IPT. Two were culture positive with fully susceptible M. tuberculosis from specimens taken due to abnormal chest radiographs, but IPT was dispensed in error: they were started on first-line treatment 9 and 184 days after starting IPT and outcomes were transfer-out and cure, respectively. The individual starting treatment after 184 days had taken an estimated 45 days of IPT and a later sputum specimen grew isoniazid-resistant M. tuberculosis. The other two had negative TB screens at the Thibela TB study site and were smear and culture negative on specimens taken later by mine health services, which initiated treatment based on results of occupational screening radiographs after 11 and 14 days of IPT. Both were documented to have completed treatment.
Concern about generating isoniazid resistance is a major obstacle to wider implementation of IPT. These data do not support this concern. Treatment outcomes were typical of this setting, taking into account the high numbers of transfers to other treatment programmes and unrecorded outcomes. South Africa as a whole has not yet met the WHO targets for treatment outcomes . Drug resistance was not more prevalent than in comparison groups. Among first episodes of TB, the prevalence of isoniazid resistance, at 12.1%, was higher in the TB after IPT group than in control clusters (6.0%), but similar to that in the laboratory substudy (11.8%). In retreatment cases, isoniazid resistance was less common in the TB after IPT group than in control clusters (7.7 vs. 18.7%). The prevalence of isoniazid resistance in the TB after IPT group is also in keeping with a drug-resistance survey among gold miners in this area in the 1990s in which 7.3% (95% CI 6.1–8.7) of first TB episodes and 14.3% (11.3–17.9) of retreatment episodes had isoniazid resistance . In this analysis, we present several comparators, acknowledging that each has limitations. Overall, although numbers of resistant cases were small and CIs accordingly wide, the data do not suggest an increase in proportion of isoniazid-resistant cases among those exposed to TB screening and IPT.
Treatment of latent TB should not, theoretically, promote antituberculous drug resistance, as in latent infection the probability of selecting for a spontaneously-occurring isoniazid-resistant organism is remote, as organism numbers are low and bacterial division slow . In a population of individuals with latent TB exposed to IPT, a higher prevalence of isoniazid resistance among subsequent TB cases would be expected even if IPT use did not itself generate resistance, as IPT is assumed to be more effective in treating isoniazid-susceptible than isoniazid-resistant latent TB.
We reported on TB episodes occurring during or relatively early after completing IPT. Given the high prevalence of HIV and TB in this population, the majority of TB cases are likely to have been due to recent infection . However, this relatively short follow-up inevitably biases toward cases identified during IPT, who are those most likely to have been screening failures, to have received inadvertent monotherapy for active disease, and hence to have acquired isoniazid resistance. With longer follow-up, we would expect an increasing proportion of cases to be due to re-infection after IPT. The control cluster comparison group may be biased toward a lower proportion with resistance, as drug susceptibility data are collected at the end of the treatment episode, thus potentially excluding individuals with longer treatment durations, due to previous treatment exposure or known baseline resistance. Effective screening before administering IPT is essential to avoid inadvertent isoniazid monotherapy for active TB. Overall consensus on the optimal screening method in resource-limited settings is yet to be reached.
Three previous papers specifically describe active TB following IPT [17–19]. Methods differ and numbers are very small, so it is difficult to draw clear conclusions in terms of drug resistance from these studies. Data concerning TB after IPT from clinical trials have been reviewed previously in a meta-analysis . The summary estimate for the risk of isoniazid resistance following IPT compared with those not exposed to IPT was 1.45 (95% CI 0.85–2.47), suggesting no evidence for an increase in resistance.
In conclusion, TB disease among mostly HIV-infected people previously exposed to IPT had treatment outcomes typical of this setting and a similar prevalence of isoniazid resistance to background. Concerns about generating drug resistance should not impede implementation of IPT.
The study was supported by the Consortium to Respond Effectively to the AIDS and TB Epidemics (CREATE; funded by a grant from the Bill and Melinda Gates Foundation); The Colt Foundation, UK; Department of Health, UK; Safety in Mines Research Advisory Committee, South Africa.
Funding: Thibela TB is funded by CREATE, with grants from the Bill and Melinda Gates Foundation, and the Safety in Mines Research Advisory Committee (South Africa).
C.L.vH. was funded by a grant from the Colt Foundation, UK.
A.D.G. was supported by a Public Health Career Scientist Award from the Department of Health, U.K.
K.L.F. is part-funded by the Biostatistics core of the Consortium to Respond Effectively to the AIDS and TB Epidemics (CREATE) with a grant from the Bill and Melinda Gates Foundation.
J.J.C.L. is funded by the Biostatistics Core of CREATE.
Author contributions: C.L.vH.: study design, data collection, paper writing.
K.L.F.: study design, epidemiological and statistical advice, paper writing.
V.N.C.: laboratory work, management of Thibela TB substudies, manuscript review.
E.C.R.: data management and logistical input, manuscript review.
J.J.C.L.: epidemiological input and statistical advice, manuscript review.
G.J.C.: study design, supervision, manuscript review.
A.D.G.: study concept and design, epidemiological input, paper writing.
The authors are grateful to the staff of Thibela TB in South Africa.
There are no conflicts of interest.
1. World Health Organization. World Health Organization HIV/AIDS and TB Department. Report of a joint World Health Organization HIV/AIDS and TB department meeting
. Geneva, World Health Organization, 2008.
2. Grant AD, Charalambous S, Fielding KL, Day JH, Corbett EL, Chaisson RE, et al
. Effect of routine isoniazid preventive therapy on tuberculosis incidence among HIV-infected men in South Africa: a novel randomized incremental recruitment study. JAMA 2005; 293:2719–2725.
3. Smeija MJ, Marchetti CA, Cook DJ, Smaill FM. Isoniazid for preventing tuberculosis in non-HIV infected persons
. Cochrane Database Syst Rev
4. Woldehanna S, Volmink J. Treatment of latent tuberculosis in HIV infected persons
. Cochrane Database Syst Rev
5. Balcells ME, Thomas SL, Godfrey-Faussett P, Grant AD. Isoniazid preventive therapy and risk for resistant tuberculosis. Emerg Infect Dis 2006; 12:744–751.
7. Lewis JJ, Charalambous S, Day JH, Fielding KL, Grant AD, Hayes RJ, et al
. HIV infection does not affect active case finding of tuberculosis in South African gold miners. Am J Respir Crit Care Med 2009; 180:1271–1278.
8. Corbett EL, Churchyard GJ, Clayton TC, Williams BG, Mulder D, Hayes RJ, et al
. HIV infection and silicosis: the impact of two potent risk factors on the incidence of mycobacterial disease in South African miners. AIDS 2000; 14:2759–2768.
9. Hanifa Y, Grant AD, Lewis J, Corbett EL, Fielding K, Churchyard G. Prevalence of latent tuberculosis among gold miners in South Africa. Int J Tuberc Lung Dis 2009; 13:39–46.
10. Day JH, Charalambous S, Fielding KL, Hayes RJ, Churchyard GJ, Grant AD. Screening for tuberculosis prior to isoniazid preventive therapy among HIV-infected gold miners in South Africa. Int J Tuberc Lung Dis 2006; 10:523–529.
11. Grant AD, Fielding KL, Chihota V, Lewis J, Luttig M, Muller D, et al. Performance of TB screening prior to community-wide IPT among gold miners in South Africa
[abstract]. Int J Tuberc Lung Dis
2008; 12 (S2)
12. World Health Organization. Treatment of tuberculosis: guidelines for national programmes
. 3rd ed. Geneva, World Health Organization, 2003. (WHO/CDS/TB/2003.313)
13. Chihota V, Poswa X, Fielding K, Molefe P, Pilane G, Lewis J, et al. A comparison of MGIT versus LJ culture by smear status in pulmonary TB in gold miners in South Africa
[abstract]. 38th World Conference on Lung Health of the International Union Against Tuberculosis and lung Disease
; 2007. Abstract PS-71729-10.
14. Churchyard GJ, Corbett EL, Kleinschmidt I, Mulder D, De Cock KM. Drug-resistant tuberculosis in South African gold miners: incidence and associated factors. Int J Tuberc Lung Dis 2000; 4:433–440.
15. Tufariello JM, Chan J, Flynn JL. Latent tuberculosis: mechanisms of host and bacillus that contribute to persistent infection. Lancet Infect Dis 2003; 3:578–590.
16. Sonnenberg P, Murray J, Glynn JR, Shearer S, Kambashi B, Godfrey-Faussett P. HIV-1 and recurrence, relapse and reinfection of tuberculosis after cure: a cohort study in South African mineworkers. Lancet 2001; 358:1687–1693.
17. Casado JL, Moreno S, Fortun J, Antela A, Quereda C, Navas E, et al
. Risk factors for development of tuberculosis after isoniazid chemoprophylaxis in human immunodeficiency virus-infected patients. Clin Infect Dis 2002; 34:368–369.
18. Fitzgerald DW, Morse MM, Pape JW, Johnson WD. Active tuberculosis in individuals infected with human immunodeficiency virus after isoniazid prophylaxis. Clin Infect Dis 2000; 31:1495–1497.
19. Nolan CM, Aitken ML, Elarth AM, Anderson KM, Miller WT. Active tuberculosis after isoniazid chemoprophylaxis of Southeast Asian refugees. Am Rev Respir Dis 1986; 3:431–436.