Recurrent Tuberculosis – Relapse, Re-infection and HIV

The catastrophic collision of tuberculosis and human immunodeficiency virus (HIV) has yielded an extraordinary burden of suffering and death, at both the individual and population levels. The convergence of these twin epidemics resulted in 1.8 million deaths due to tuberculosis in 2007, 24% of which were among people with HIV infection, while among the 2 million deaths of people with acquired immunodeficiency syndrome (AIDS), 22% were caused by tuberculosis. (1) Tuberculosis is, thus, the leading cause of mortality in people with HIV/AIDS, and HIV contributes to a substantial proportion of tuberculosis deaths. Interventions to control tuberculosis and HIV, therefore, need to take into account the unique and deadly synergy between these two infections.

There are many challenges that clinicians and public health programs must confront when trying to manage and prevent HIV-related tuberculosis, including prompt diagnosis, effective treatment and successful prevention strategies. One problem that receives little attention is recurrence of tuberculosis following completion of treatment. Historically, patients properly treated with a four-drug regimen have a very high treatment success rate and very low incidence (2–3%) incidence of recurrence. In HIV-infected patients, the risk of recurrent disease is higher, and previous work has demonstrated that this can be due to treatment failure, emergence of drug resistance during therapy or reinfection with a new strain of Mycobacterium tuberculosis (2, 3, 4). Recurrence of tuberculosis after treatment among both HIV-infected and –uninfected individuals is not a trivial problem. On a patient level, recurrent tuberculosis requires another round of treatment, in many parts of the world with a so-called “retreatment” regimen that is more toxic, takes longer to complete and which may, ironically, amplify drug resistance. On a public health level, recurrent tuberculosis may account for 10–30% of all cases within some weaker tuberculosis control programs, particularly those that do not use at least six months of rifampicin treatment (5), and contributes to ongoing transmission of infection to contacts of cases in the home, community and health facilities, particularly from HIV-uninfected patients (6).

In this issue of The Journal, two articles appear that examine the epidemiology of recurrent tuberculosis in populations with a high prevalence of HIV infection. Narayanan and colleagues from India report on two cohorts of patients with tuberculosis who were followed for recurrence, and then underwent DNA fingerprinting of initial and recurrence M. tuberculosis isolates to distinguish exogenous reinfection from relapse. Among HIV-infected patients whose initial episode of tuberculosis was deemed cured, 14% experienced a recurrence of tuberculosis, of which just over half were fingerprinted. Among these cases, 22/25 (88%) were due to reinfection with a different strain of M. tuberculosis. In a sample of HIV-uninfected patients with recurrent tuberculosis (rates of recurrence are not reported for these patients), only 9% of second episodes were due to reinfection with the remainder being relapses of the initial infection. One possible explanation for the difference in levels of relapse in HIV-uninfected patients is that many were treated with non-standard, abbreviated regimens that included ofloxacin in a clinical trial that failed to include an appropriate control regimen; these relapses may simply have been the result of inadequate therapy. Importantly, one-quarter of HIV-related recurrences were with multidrug resistant strains of M. tuberculosis, with a smaller number of resistant recurrences found in HIV-uninfected patients.

While rates of recurrent tuberculosis cannot be compared by HIV status in this report, it is clear that recurrent disease is more common than would be expected in HIV-infected tuberculosis patients in this Indian setting, but that almost all recurrences are due to exogenous reinfection. Narayanan and colleagues used rigorous typing methods, with three independent techniques, and took great care to exclude the possibility of laboratory contamination. While the study breaks no new ground with respect to the existence of exogenous M. tuberculosis reinfection causing disease (4), particularly in HIV-infected patients, it is nonetheless noteworthy for documenting the potential frequency of this phenomenon in developing countries, where recurrences are generally attributed to treatment failure due to either non-compliance or drug resistance. In this study, all of the multidrug resistant recurrences among HIV-infected patients were reinfections, whereas both multidrug resistant cases in HIV-uninfected patients were due to acquired or amplified resistance during treatment. These data confirm what has been inferred from outbreaks of multidrug and extensively drug resistant tuberculosis among people with HIV infection in other settings: that much drug resistant tuberculosis in HIV-infected patients is transmitted from others, rather than acquired by ineffective or insufficient therapy. (7)

A second article in this issue of the Journal, by Glynn and coworkers, reports on rates of recurrent tuberculosis in mine workers in South Africa. These authors followed two cohorts of tuberculosis patients – one with and one without HIV infection – whose illness was initially diagnosed and who underwent HIV testing in the 1990s, and determined the rates of recurrent disease that occurred at least two years following cure of the initial episode. The authors assumed that these late recurrences were due to reinfection rather than relapse, based on the previous observations, supported by the data from Narayanan et al., that almost all relapses occur in the first two years following treatment. Glynn and associates found recurrence rates of 24.4 per 100 person years for initially HIV-infected miners following their index episode and 4.7 per 100 person years among those who were HIV negative at the index episode. Strikingly, rates of recurrent tuberculosis were dramatically higher than incident tuberculosis rates in both HIV-infected and – uninfected miners. These results are consistent with a study from another high tuberculosis transmission setting where reinfection rates following treatment exceeded rates of incident disease. (8)

The work of Glynn and coworkers requires more assumptions about the source of recurrence, and has less direct evidence than the study from Narayanan and associates. No DNA fingerprinting was performed to confirm that recurrences were actually reinfections and not relapses; repeat HIV testing was not performed among the initially HIV seronegative men, despite known high rates of HIV acquisition during the time of the study; and considerable attrition from the cohorts over time occurred, which could affect incidence estimates. Despite these limitations, the study demonstrates convincingly that recurrent tuberculosis is exceedingly common in miners in South Africa, and is seen in both HIV infected and uninfected men. Work by the same group and others (9, 10) previously documented high rates of reinfection in this setting, so it is quite credible that a large proportion of these recurrent cases were, indeed, new infections.

What can be done to reduce the frequency of recurrent tuberculosis, whether due to relapse or reinfection, in patients with and without HIV infection? In fact, quite a lot can be done, and the evidence base for a number of interventions is robust. In the first place, ensuring completion of appropriate therapy for all tuberculosis patients is extremely important. While rates of treatment completion worldwide are now close to 85% for patients cared for in DOTS programs, the real completion rate when all patients are accounted for is substantially lower, particularly in Africa. (1) Failure to complete tuberculosis treatment is associated with very high rates of recurrent disease, especially in HIV-infected patients. (11, 12) In addition, while 5.5% of all new tuberculosis cases worldwide have multidrug or extensively drug resistant disease, only a tiny fraction of these are actually detected and receive appropriate treatment. (1) Improving case detection, easier access to care, improving cure rates with community-based interventions, developing new effective shorter regimens (13) and conducting baseline drug susceptibility testing will all reduce relapses and the evolution of drug resistance. Earlier detection of multidrug resistant TB, through use of rapid rifampin resistance assays such as line probe assays, and earlier effective treatment should also reduce recurrence due to reinfection with multidrug resistant TB in HIV-infected patients. (14)

Because reinfection is common in HIV-infected patients in high burden settings, secondary preventive therapy with isoniazid is another strategy for reducing recurrences. One small controlled trial and one observational study have documented the effectiveness of secondary isoniazid for reducing recurrences in patients with HIV-related tuberculosis. (15, 16) In a non endemic setting, the risk of recurrence among HIV-infected persons was reduced with longer duration of TB treatment. (17)

Low CD4 cell counts are a major predictor of recurrent tuberculosis due to both relapse and reinfection in HIV-infected patients, and treatment with antiretroviral therapy decreases the likelihood of recurrence by at least 50%. (10, 11) Widespread use of antiretroviral therapy for all HIV-infected tuberculosis patients is now supported by evidence from a clinical trial and from observational studies. (18, 19). Scaling up antiretrovirals for tuberculosis patients will also reduce recurrent disease, as suggested by study in Rio de Janeiro, Brazil. (11)

A critically needed intervention to reduce reinfection tuberculosis is the implementation of infection control measures in clinical and community settings. Because patients with tuberculosis and with HIV infection congregate in the same clinic waiting rooms, offices, laboratories and hospital wards in much of the world, ongoing transmission is a significant cause of new disease, especially drug resistant disease, as Narayanan and colleagues demonstrate. HIV and antiretroviral clinics in resource poor settings have unwittingly become cauldrons of tuberculosis transmission, seriously undermining the impact of HIV therapies. The first step in reducing institutional and community transmission of M. tuberculosis infection is the detection of prevalent, undiagnosed cases and the initiation of appropriate therapy. Administrative measures to separate infectious patients from others, engineering and air flow management ranging from opening windows and doors to more sophisticated air handling techniques, and the use of personal protective equipment in more exposure-intense situations are all approaches that need to be ramped up worldwide, given the clear evidence of nosocomial transmission of tuberculosis infection among those with HIV infection. (20)

Finally, what about vaccination as a strategy for preventing recurrent tuberculosis? Both Narayanan et al. and Glynn and coworkers suggest that development of a therapeutic vaccine or immunotherapy as an adjunct treatment for tuberculosis might help reduce relapse and reinfection episodes in those with exquisite susceptibility to the disease. While this is an attractive concept in theory, the feasibility of this approach is currently quite limited. What is eminently sensible, however, is a primary vaccine for tuberculosis that prevents the development of disease in the first place. The current tuberculosis vaccine, BCG, appears to protect infants from severe forms of disease, but does not prevent pulmonary tuberculosis in adults. We know from previous experience that a preventive vaccine can work, but in the case of BCG the protective effects of the vaccine most likely have become attenuated because of continuing evolution during laboratory passage. (21) Thus, prospects for new pre- and post-exposure preventive vaccines are bright. (22) The potential for an effective primary vaccine to control initial and, thus, recurrent tuberculosis, particularly among HIV-infected persons, is enormous. After all, the best way to ensure that patients avoid a second episode of tuberculosis is to make certain that the first episode never occurs.