A majority of the global burden of HIV-associated tuberculosis (TB) lies in sub-Saharan Africa  and rates in southern Africa have reached almost unprecedented levels . Antiretroviral treatment (ART) is now becoming more widely available in the region. However, the fact that many patients accessing ART are already receiving treatment for TB [3,4] presents a major clinical challenge due to the complexities involved in the concurrent management of these two infections [5–8]. In addition to high pill burden, drug co-toxicity and pharmacokinetic drug interactions, TB-associated immune reconstitution disease (IRD) has been reported in up to 43% of patients in high-income countries receiving concurrent treatment for these infections [9,10].
Since the prevalence of TB among patients entering ART programmes in sub-Saharan Africa is extremely high [3,4], TB-IRD could greatly complicate the delivery of ART in the region, causing substantial in-programme morbidity and mortality and increasing the burden on secondary health-care facilities. Moreover, the risks associated with TB-IRD are important variables in the debate over the optimal time for ART initiation in patients with TB [7,8,10]. There are, however, no existing published reports of the frequency and impact of TB-IRD in ART programmes in sub-Saharan Africa.
Data concerning the burden of TB-IRD in ART programmes in sub-Saharan Africa are clearly needed to inform both clinicians and those involved in the development of treatment guidelines for ART programmes. We have previously described a community-based ART programme in Cape Town, reporting detailed analyses of morbidity, mortality, immunological recovery and overall programme outcomes [3,11–15]. Within this service 25% of patients have prevalent TB at programme entry, the majority of whom subsequently receive overlapping antituberculosis treatment and ART . Collection of detailed prospective data on this large cohort over 3 years provided the opportunity to analyse the burden of TB-IRD in this setting.
The ART service based at the Gugulethu Community Health Centre in Cape Town was started in 2002 and has previously been described in detail [11–13]. The district has a predominantly African population of over 300 000, the vast majority of whom live in conditions of low socioeconomic status. In 2003 the antenatal HIV seroprevalence was 28% and the TB notification rate exceeded 1 000/100 000 . Patients were referred from primary care HIV clinics to the ART programme. National guidelines for use of ART were based on the WHO 2002 recommendations , which advise treatment for those with a prior AIDS diagnosis (WHO stage 4 disease) or a blood CD4 cell count < 200 cells/μl. First-line ART comprised stavudine, lamivudine plus a non-nucleoside reverse transcriptase inhibitor (predominantly efavirenz). Treatment compliance rates were very high as reflected by rates of viral load suppression < 400 copies/ml, which exceeded 90% at all follow-up time-points over the first 3 years of the programme . In addition to the minimum schedule of clinic appointments at 4, 8, and 16 weeks, patients with clinical problems such as concurrent TB were reviewed more frequently. All patients had open access to the clinic for medical problems during weekdays and 24-hour access to care by the nearby secondary hospital. All care and medication was supplied free of charge to the patients.
The high burden and diagnostic criteria for TB in this programme have been described in detail . Available investigations for TB included sputum smear microscopy and culture (MGIT, Becton Dickinson, Sparks, Maryland, USA), chest radiology, abdominal ultrasonography, fine needle aspiration of lymphadenopathy for cytology and culture and drug sensitivity testing of mycobacterial isolates. Nebulized sputum induction was accessible when required. Results of investigations for TB were cross-checked with electronic records held by the National Health Laboratory Service.
Detailed structured clinical and laboratory records were maintained for every patient visit. Data were transferred on a weekly basis to an electronic database. Patients requiring in-patient care were referred to a nearby 200-bed secondary hospital. Information on in-patient care was gained from discharge letters, hospital and laboratory records and post-mortem examinations. Deaths and losses to follow-up were ascertained by active community-based follow-up as described in detail previously [12,13]. Outcomes were determined for all patients. Attributable causes of death were assigned based on all the available information after detailed review by two specialists in infectious diseases and HIV medicine.
Patients with ‘prevalent TB’ included those receiving antituberculosis treatment at entry to the ART programme as well as patients with new TB diagnoses established after enrolment in the programme but prior to ART initiation. TB-IRD was diagnosed among patients with prevalent TB who had an initial symptomatic improvement during antituberculosis treatment followed by deterioration of symptoms after initiation of ART, which was not due to another opportunistic infection, drug adverse effect, ineffective TB treatment or drug-resistant TB .
All patients enrolled into the programme between September 2002 and April 2005 were studied and the period between ART initiation and the 4-month follow-up appointment was studied. In analyses, Fisher's exact and Wilcoxon rank-sum tests were used to compare proportions and medians, respectively. Multiple logistic regression was used to examine the associations between IRD and patient demographic and clinical characteristics. Regression diagnostics followed standard procedures . All P-values reported were two-sided at alpha = 0.05.
Patients and follow-up
A total of 756 ART-naive patients who enrolled during the study period initiated ART. Their median age was 33 years [interquartile range (IQR), 28–37 years] and 73% were female. Antituberculosis treatment was being received at the time of ART initiation by 160 (21%) patients. Those with prevalent TB differed from those who were TB-free in having a lower median CD4 cell count (68 cells/μl; IQR, 29–133 cells/μl) versus 101 cells/μl (IQR, 52–156 cells/μl; P < 0.001), a higher median viral load (4.96 versus 4.84 log RNA copies/ml; P = 0.003) and more advanced WHO stages of disease (stages 3 and 4 among 62% and 38% versus 51% and 26%, respectively). Since many patients had initiated antituberculosis treatment prior to entering the programme whereas others had TB diagnosed after enrolment, the duration of the interval between TB diagnosis and ART initiation was very broad (median 105 days; IQR, 61–164 days).
Of the patients with TB (n = 160), 142 (89%) were retained on ART at the 4-month follow-up appointment. Of the remaining patients, one (0.6%) was transferred-out at week 14, one (0.6%) was lost to follow-up at week 5, and 16 (10%) died. Overall a total of 60.1 person-years of observation accrued during follow-up. Deaths were due to advanced Kaposi's sarcoma (n = 4), acute sepsis (n = 3), cryptococcal immune reconstitution disease (n = 2), renal failure (n = 1), and wasting syndrome with profuse chronic diarrhoea (n = 2). Two patients died of TB without evidence of IRD: one had very advanced hypoxic lung disease and the other had active disease due to TB treatment default. Two patients died with TB-IRD as described below.
Cases of TB immune reconstitution disease
Among patients with prevalent TB, a total of 19 (12%) were diagnosed as having TB-IRD. Symptoms developed a median of 2 weeks (IQR, 1.5–3.5 weeks) after initiation of ART and systemic as well as organ-specific symptoms were present in all. IRD presented as an exacerbation of existing disease manifestations alone in 10 patients with pulmonary disease among the majority (n = 9) (Table 1). Seven further patients with initial diagnoses of pulmonary TB developed new disease manifestations at another anatomic site as well as a concurrent exacerbation of respiratory disease in five (Table 1). Two other patients with disseminated disease developed IRD that culminated in death; both had pulmonary, intra-abdominal and bone marrow involvement. These two cases occurred early in the history of the ART programme; since diagnoses were not established ante-mortem, neither was managed as IRD.
Overall, intra-abdominal manifestations occurred in 7 (37%) IRD cases. Hepatomegally with elevated serum concentrations of alkaline phosphatase and gamma-glutamyl transferase but relatively normal serum concentrations of alanine transaminase were detected in four patients (21%). IRD was self-limiting in the majority of patients. Secondary level out-patient or in-patient care was received by three and four patients, respectively, two received oral corticosteroids and one required a laparotomy.
Risk factors for TB immune reconstitution disease
Comparison of the characteristics of TB patients who developed IRD (cases) with those who did not is shown in Table 2. Cases had significantly lower baseline CD4 cell counts and ART was initiated after a shorter interval from TB diagnosis. There was no significant correlation between the baseline CD4 cell count and the delay between TB diagnosis and ART (r, 0.09; P = 0.259). Cases did not differ from those who remained IRD-free with respect to other baseline characteristics or laboratory results at 4 months' follow-up (Table 2).
Among patients with baseline CD4 cell counts < 50, 50–99, 100–149 and ≥ 150 cells/μl, IRD developed among 21%, 14%, 7% and 0%, respectively. Although only 12% of patients overall developed TB-IRD, the proportion of patients affected who commenced ART within 2 months of TB diagnosis was much higher (32%). Furthermore, the risk of IRD for patients with low baseline CD4 cell counts commencing ART early in the course of antituberculosis treatment was especially high (Fig. 1).
Multivariate analysis that included all baseline characteristics showed that increased odds of developing IRD was significantly associated with lower baseline CD4 cell counts but that the association with shorter delays from TB diagnosis to ART initiation was particularly strong (Table 3).
The proportion of patients with HIV-associated TB reported to develop IRD following initiation of ART in high-income countries ranges between 11% and 43% [9,19–22]. This wide variation is likely to reflect differences in cohort characteristics, case definitions and differences in the mean time interval between TB diagnosis and ART initiation. Data from resource-limited countries on TB-IRD is scarce; a rate of 8% was reported from India  but a study from Tanzania reported no cases at all . This study from South Africa is the largest series patients with overlapping ART and antituberculosis treatment yet reported. We found that overall 12% of TB patients developed IRD. However, the low rate is likely to reflect the fact that many patients were referred to the ART programme having already completed several months of antituberculosis treatment (median, 3.5 months) and so many were likely to be at low risk of IRD. Risk of TB-IRD was much higher among those with early initiation of ART, affecting 32% of patients initiating ART within 2 months of TB diagnosis. The rate among these patients, a more relevant comparison group, is consistent with rates reported from high-income countries [9,20–22].
Most previous reports of TB-IRD are from series of patients attending referral hospitals [9,20,21,23]. In contrast, this is the first report of TB-IRD from a community-based ART programme in a resource-limited setting. An important strength of this analysis is that all patients within the programme receiving TB treatment at the time of ART initiation were included and outcomes for all patients are known except for one who was lost to follow-up. Data collection was detailed and prospective; access to unscheduled clinic visits when needed was good, and all treatment was free of charge. As a result, data completeness is likely to be high. All patients were ART-naive and received a standard first-line ART regimen, excluding these as potential variables affecting risk of IRD. In common with most previous studies [9,19,20,22,23], diagnoses of IRD were made retrospectively but were nevertheless based on systematically collected prospective data. Although some cases of TB-IRD have been reported to occur > 16 weeks of ART , only one case occurred beyond 12 weeks and so we restricted this analysis to the first 16 weeks of ART. The retrospective study design may have missed some very mild manifestations of TB-IRD but, because of excellent cohort retention and close follow-up of cases, all clinically relevant disease is likely to have been detected and thus the impact of TB-IRD on this ART programme should have been accurately assessed. The focus of this study was on IRD presenting as clinical deterioration of prevalent TB rather than ‘unmasking’ of new TB presenting after ART initiation.
As the overall size of the cohort of TB patients was large, we were able to perform multivariate analysis to identify risk factors associated with development of IRD. As reported previously, early initiation of ART was associated with a higher risk of IRD [10,20,22]. However, in view of the broad range of the intervals between TB diagnosis and ART initiation, we were able to define this relationship more clearly than has been done previously (Fig. 1, Table 3). Risk of IRD was significantly increased if ART was commenced within 3 months from TB diagnosis but increased greatly when ART was initiated within 1 month. However, as the numbers of patients and cases within each time stratum are small, these data should not be taken as precise estimates of risk of TB-IRD within each stratum.
In addition to the timing of ART, we also found that increased risk of IRD was also significantly associated with lower baseline CD4 cell counts (Tables 2 and 3). A similar trend has been observed in some [19,21] but not other [9,20,23] studies. Compared to other studies, a higher mean baseline CD4 count may explain the unusually low rate of IRD in a study from India . The association of low CD4 cell count with increased risk of IRD is plausible as those with the most advanced HIV-associated immunodeficiency are likely to have higher mycobacterial antigen load and greater impairment of immune responses that may rapidly reverse during early ART . Some previous studies have also found extrapulmonary TB to be associated with increased risk of IRD [9,19,20]. However, in resource-limited settings, patients with a diagnosis of pulmonary TB are often not investigated further for evidence of asymptomatic extrapulmonary involvement, probably resulting in substantial under-diagnosis in those with advanced immunodeficiency.
Existing studies do not show consensus on whether early immunovirological responses to ART are risk factors for IRD [9,20–22] although the largest of these studies did find that risk was increased among those with a more rapid fall in viral load . In this cohort, rates of virological suppression and immunological recovery were exceptionally high but were not discriminatory as risk factors for IRD. However, these measurements were made at 16 weeks of ART and not at the onset of IRD. In view of the timing of these measurements, however, these data strongly suggest that following development of IRD, treatment compliance and responses to ART thereafter were not undermined.
Manifestations of TB-IRD are diverse, but literature from high-income countries  and from India  reported cervical lymphadenitis much more frequently than was observed in this study. This is likely to reflect true differences in the spectrum of disease since cervical adenitis is clinically readily apparent and is therefore unlikely to have been missed. Respiratory manifestations were most common but a high rate of intra-abdominal disease was also observed, including frequent hepatic involvement with elevation of serum concentrations of bile cannalicular enzymes. Many patients initially treated for pulmonary TB developed additional manifestations of IRD at extrapulmonary sites, again reflecting likely initial under-diagnosis of extrapulmonary involvement in patients with pulmonary TB. In this study, two (1%) TB patients died of IRD; both diagnoses were retrospective and neither was managed with either use of corticosteroids or discontinuation of ART. This is an important finding as a review of previous cases from high-income countries revealed that although life-threatening manifestations were reported there were no deaths . This illustrates the importance of establishing clinical diagnoses of TB-IRD and the need for access to secondary level health care for some patients in this setting. However, overall mortality associated with TB-IRD in this cohort was low and the majority of TB-IRD was self-limiting. Indeed, we have previously reported that cryptococcal IRD is a far more common cause of death in this service [12,15].
IRD could potentially lead to a high rate of secondary health-care utilization. However, just 4% of TB patients were referred for either in-patient or out-patient secondary care for this reason, representing approximately 1% of all patients in the cohort. This represents a very small proportion of total secondary health-care utilization by individuals accessing ART in this service, which was received by approximately 16% of patients during the first 16 weeks of ART (Guy Harling, unpublished data).
The factors affecting the optimal timing for initiation of ART in patients with HIV-associated TB are multiple. It has recently been shown that active TB and antituberculosis treatment during ART do not compromise immunological and virological responses to treatment in this  and other cohorts [26,27], suggesting that pharmacokinetic and compliance issues are not strong arguments for delaying treatment . Risk of IRD has also been raised as a factor favouring delayed treatment initiation [6,8]. However, the present study suggests that the overall mortality risk associated with TB-IRD is small whereas we have previously reported that even short delays in initiation of ART are associated with a high mortality risk in this cohort [12,13]. Taken together, these data would favour a policy of earlier initiation of ART among TB patients. Our data, however, indicate that such a policy would substantially increase the proportion of patients at risk of IRD. Diagnostic and management strategies for TB-IRD in resource-limited settings must be developed, including guidelines for use of corticosteroids. Moreover, the long delays for many TB patients in this study to initiate ART are indicative of the lack of integration of TB and ART services locally. These delays and the need for coordinated management of patients with TB-IRD strongly support moves to integrate such services.
In summary, the risk of TB-associated IRD in this setting is very high for those with low baseline CD4 cell counts and early initiation of ART. Although the contribution of TB-IRD to secondary health-care utilization and mortality risk in the whole cohort was small, any future policy changes recommending early initiation of ART among TB patients would be very likely to lead to an increase in this burden of disease.
The authors are grateful to Sister Marjorie Ntobongwana and the staff at the Hannan Crusaid antiretroviral clinic in Gugulethu and at the Desmond Tutu HIV Centre.
Sponsorship: SDL is funded by the Wellcome Trust, London, UK with grant 074641/Z/04/Z. RW is funded in part by the National Institutes of Health, USA, RO1 grant (A1058736-01A1). LM, LGB and RW are all funded in part by the National Institutes of Health through a CIPRA grant 1U19AI53217-01. Provision of ART at the programme was initially by Crusaid, London, UK and latterly by the Global Fund for Malaria, Tuberculosis and HIV/AIDS administered through the Provincial Government of the Western Cape.
1. Corbett EL, Watt CJ, Walker N, Maher D, Williams BG, Raviglione MC, et al
. The growing burden of tuberculosis: global trends and interactions with the HIV epidemic. Arch Intern Med 2003; 163:1009–1021.
2. Lawn SD, Bekker LG, Middelkoop K, Myer L, Wood R. Impact of HIV infection on the epidemiology of tuberculosis in a peri-urban community in South Africa: the need for age-specific interventions. Clin Infect Dis 2006; 42:1040–1047.
3. 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; 20:1605–1612.
4. Moore D, Ekwaru P, Liechty C, Were W, Mwima G, Solberg P, et al
. Prevalence, incidence and outcomes of tuberculosis treatment in HIV+ individuals initiating home-based antiretroviral therapy in rural Uganda
. Thirteenth Conference on Retroviruses and Opportunistic Infections
. Denver, CO, February 2006 [abstract 794].
5. Dean GL, Edwards SG, Ives NJ, Matthews G, Fox EF, Navaratne L, et al
. Treatment of tuberculosis in HIV-infected persons in the era of highly active antiretroviral therapy. AIDS 2002; 16:75–83.
6. de Jong BC, Israelski DM, Corbett EL, Small PM. Clinical management of tuberculosis in the context of HIV infection. Annu Rev Med 2004; 55:283–301.
7. Havlir DV, Barnes PF. Tuberculosis in patients with human immunodeficiency virus infection. N Engl J Med 1999; 340:367–373.
8. Burman WJ, Jones BE. Treatment of HIV-related tuberculosis in the era of effective antiretroviral therapy. Am J Respir Crit Care Med 2001; 164:7–12.
9. Breton G, Duval X, Estellat C, Poaletti X, Bonnet D, Mvondo MD, et al
. Determinants of immune reconstitution inflammatory syndrome in HIV type 1-infected patients with tuberculosis after initiation of antiretroviral therapy. Clin Infect Dis 2004; 39:1709–1712.
10. Lawn SD, Bekker LG, Miller RF. Immune reconstitution disease associated with mycobacterial infections in HIV-infected individuals receiving antiretrovirals. Lancet Infect Dis 2005; 5:361–373.
11. Bekker LG, Myer L, Orrell C, Lawn S, Wood R. Rapid scale-up of a community-based HIV treatment service: programme performance over 3 consecutive years in Guguletu, South Africa. S Afr Med J 2006; 96:315–320.
12. Lawn SD, Myer L, Orrell C, Bekker LG, Wood R. Early mortality among adults accessing a community-based antiretroviral service in South Africa: implications for programme design. AIDS 2005; 19:2141–2148.
13. Lawn SD, Myer L, Harling G, Orrell C, Bekker LG, Wood R. Determinants of mortality and nondeath losses from an antiretroviral treatment service in South Africa: implications for program evaluation. Clin Infect Dis 2006; 43:770–776.
14. Lawn SD, Myer L, Bekker LG, Wood R. CD4 cell count recovery among HIV-infected patients with very advanced immunodeficiency commencing antiretroviral treatment in sub-Saharan Africa. BMC Infect Dis 2006; 6:59.
15. Lawn SD, Bekker LG, Myer L, Orrell C, Wood R. Cryptococcocal immune reconstitution disease: a major cause of early mortality in a South African antiretroviral programme. AIDS 2005; 19:2050–2052.
16. Cape Town TB Control. Progress report 1997–2003. Health Systems Trust, Cape Town, South Africa, 2004. www.hst.org.za
17. World Health Organization. Scaling up Antiretroviral Therapy in Resource-limited Settings: Guidelines for a Public Health Approach; Executive Summary
. Geneva: World Health Organization; 2002.
18. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: Wiley Interscience; 2004.
19. Wendel KA, Alwood KS, Gachuhi R, Chaisson RE, Bishai WR, Sterling TR. Paradoxical worsening of tuberculosis in HIV-infected persons. Chest 2001; 120:193–197.
20. Breen RA, Smith CJ, Bettinson H, Dart S, Bannister B, Johnson MA, et al
. Paradoxical reactions during tuberculosis treatment in patients with and without HIV co-infection. Thorax 2004; 59:704–707.
21. Narita M, Ashkin D, Hollender ES, Pitchenik AE. Paradoxical worsening of tuberculosis following antiretroviral therapy in patients with AIDS. Am J Respir Crit Care Med 1998; 158:157–161.
22. Shelburne SA, Visnegarwala F, Darcourt J, Graviss EA, Giordano TP, White AC Jr, et al
. Incidence and risk factors for immune reconstitution inflammatory syndrome during highly active antiretroviral therapy. AIDS 2005; 19:399–406.
23. Kumarasamy N, Chaguturu S, Mayer KH, Solomon S, Yepthomi HT, Balakrishnan P, et al
. Incidence of immune reconstitution syndrome in HIV/tuberculosis-coinfected patients after initiation of generic antiretroviral therapy in India. J Acquir Immune Defic Syndr 2004; 37:1574–1576.
24. Shao H, Crump J, Ramadhani H, Uiso L, Sendui-Nguyaine, Kiwera R, et al
. A randomised trial of early versus delayed fixed dose combination zidovudine/lamivudine/abacavir in patients coinfected with HIV and tuberculosis: early findings of the tuberculosis and immune reconstitution syndrome trial
. Thirteenth Conference on Retroviruses and Opportunistic Infections
. Denver, CO, February 2006 [abstract 796].
25. Lawn SD, Bekker LG, Wood R. How effectively does HAART restore immune responses to Mycobacterium tuberculosis? Implications for tuberculosis control. AIDS 2005; 19:1113–1124.
26. Breen RA, Miller RF, Gorsuch T, Smith CJ, Ainsworth J, Ballinger J, et al
. Virological response to highly active antiretroviral therapy is unaffected by antituberculosis therapy. J Infect Dis 2006; 193:1437–1440.
27. Manosuthi W, Kiertiburanakul S, Sungkanuparph S, Ruxrungtham K, Vibhagool A, Rattanasiri S, et al
. Efavirenz 600 mg/day versus efavirenz 800 mg/day in HIV-infected patients with tuberculosis receiving rifampicin: 48 weeks results. AIDS 2006; 20:131–132.
28. Lawn SD, Wood R. Impact of antituberculosis treatment on virological response to highly active antiretroviral therapy: implications for resource-limited settings. J Infect Dis 2006; 194:1467–1468.