estimates has resulted from increased availability of data from greater coverage ... The initial weeks of immune recover
Tropical Medicine and International Health
doi:10.1111/j.1365-3156.2010.02568.x
volume 15 no 8 pp 981–989 august 2010
Tuberculosis research update W. A. Hanekom1, S. D. Lawn2,3, K. Dheda4,5 and A. Whitelaw6,7 1 2 3 4 5 6 7
South African Tuberculosis Vaccine Initiative, University of Cape Town, Cape Town, South Africa Desmond Tutu HIV Centre, University of Cape Town, Cape Town, South Africa Clinical Research Unit, London School of Hygiene & Tropical Medicine, London, UK Lung Infection and Immunity Unit, University of Cape Town, Cape Town, South Africa Centre for Infectious Diseases and International Health, University College London, London, UK Division of Medical Microbiology, University of Cape Town, Cape Town, South Africa National Health Laboratory Service, Cape Town, South Africa
Summary
Tuberculosis (TB) remains a major challenge to global public health in the 21st century. In 2007, there were an estimated 9.27 million new cases and 1.3 million deaths among HIV-negative patients with TB. The HIV-associated TB epidemic, drug-resistant disease, the need for better diagnostic assays and the limited efficacy of Bacille Calmette–Guerin vaccination are four important obstacles to further progress in global TB control. In this brief review, we provide a focused update on these four key areas of TB research. keywords tuberculosis, vaccine, drug resistance, HIV, diagnosis
HIV-associated TB Global burden of HIV-associated TB HIV-associated TB (HIV-TB) remains a major challenge to global health, and the epidemic is one of the major stumbling blocks to attainment of the 2015 Millennium Development Goals for TB control (United Nations 2008). An estimated 1.37 million of new TB cases that occurred worldwide in 2007 were HIV-associated (14.8% of all TB cases), and resulted in approximately 456 000 deaths (23% of global deaths from HIV ⁄ AIDS) (WHO 2009). Comparison of the 2008 and 2009 annual WHO global TB control reports reveals an apparent twofold increase in estimated disease burden. This major upward revision in estimates has resulted from increased availability of data from greater coverage of HIV-testing among patients with TB, particularly in Africa, rather than any change in disease burden. The epicentre of the HIV-TB co-epidemic lies in subSaharan Africa, which accounts for 79% of the worldwide disease burden. South Africa alone accounts for 25% of cases, despite only having 0.7% of the world’s population (Abdool-Karim et al. 2009, WHO 2009). Here, TB incidence rates continue to rise, although in many other countries in the region, rates have started to decrease between 2003 and 2007 (WHO 2009). These trends are likely to reflect the natural evolution of the HIV epidemic; the contribution, if any, of other potential factors such
ª 2010 Blackwell Publishing Ltd
as scale-up of antiretroviral therapy (ART) is as yet unknown. HIV-associated TB is also an important public health challenge elsewhere. In Eastern Europe, overlapping HIV and TB drug-resistance epidemics have contributed to an approximate doubling of the estimated regional TB incidence rate since 1990 (Lazarus et al. 2008; WHO 2009). Even in England and Wales, HIV prevalence among patients with TB increased from 3.1% in 1999 to 8.3% in 2003 and HIV-TB cases (mostly non-UK born) contributed almost one-third to the increase in overall TB notifications (Ahmed et al. 2007). In contrast, the United States experienced a threefold decrease in the number of cases between 1993 and 2004, coinciding with improvements in TB control and advances in HIV diagnosis and treatment (Albalak et al. 2007).
ART for prevention of HIV-associated TB Antiretroviral therapy results in restoration of immune responses to Mycobacterium tuberculosis, leading to sustained reductions in long-term TB risk (Lawn et al. 2005a). The initial weeks of immune recovery, however, may be associated with a transiently heightened risk of TB because of immune-mediated ‘unmasking’ of sub-clinical TB unrecognized at the time of ART initiation (Breen et al. 2005; Lawn et al. 2008). This phenomenon was estimated to account for approximately 40% of TB cases presenting 981
Tropical Medicine and International Health
volume 15 no 8 pp 981–989 august 2010
W. A. Hanekom et al. Tuberculosis research update
during early ART in a South African cohort (Lawn et al. 2009a). Thereafter, the risk of TB decreases rapidly, particularly in the first 2–3 years of ART (Lawn et al. 2005b). Data from cohorts in diverse settings show that patients receiving ART have a 54–92% reduction in adjusted hazards of TB compared to untreated patients (Lawn et al. 2009b). Despite this, however, rates during long-term ART may remain more than fivefold higher than background rates in HIV-uninfected people from the same communities (Girardi et al. 2005; Lawn et al. 2006). In a South African cohort, the dominant risk factor for TB at any time-point during long-term ART was the current CD4 count (Lawn et al. 2009a). At current CD4 counts of 500 cells ⁄ ll (1.5 cases ⁄ 100 person-years) (Lawn et al. 2009a). Further analyses showed that overall cumulative risk of TB depends on the duration of time that patients accrue with low CD4 cell counts and that this risk would be reduced by initiation of ART at higher CD4 counts (Fitzgerald 2009; Lawn et al. 2009a). Despite the major individual benefit of ART on risk of TB and mortality, it is as yet unclear whether scale-up of ART will contribute to TB control in countries with high HIV prevalence where existing TB control efforts have failed (Lawn et al. 2005a, 2009b). Mathematical modelling and empirical observations suggest that any effect will be small (Williams & Dye 2003). Most patients with HIV present with very low CD4 counts and much HIV-TB occurs prior to ART initiation (Lawn et al. 2006). High rates also persist during long-term ART, with the result that long-term survivors will have very high cumulative life-time TB risk (Lawn et al. 2006, 2009b). Moreover, high community coverage with ART is difficult to achieve (WHO 2008b). However, early data from a study community in South Africa do suggest a modest population level effect on rates and prevalence of HIV-TB (Middelkoop et al. 2009). Immune reconstitution disease associated with TB Development of consensus case definitions for TB immune reconstitution disease (TB-IRD) for use in high-income and resource-limited settings is an important advance, facilitating diagnosis and comparability of research data (Lawn et al. 2005c; Meintjes et al. 2008). Two forms are distinguished: ‘paradoxical’ TB-IRD, in which ART initiation results in clinical deterioration of TB during initial successful treatment, and ‘unmasking’ TB, in which ART initiation triggers the presentation of previously unrecognized TB. 982
Between 8% and 43% of patients with TB initiating ART develop paradoxical TB-IRD (Meintjes et al. 2008). Risk factors include early initiation of ART during TB treatment, low baseline CD4 cell counts, disseminated TB and rapid immune and virological responses to ART (Lawn et al. 2005c; Meintjes et al. 2008). Detectable urinary lipoarabinomannan (LAM) at the time of TB diagnosis may be a novel marker of paradoxical TB-IRD, possibly reflecting high and disseminated mycobacterial load and advanced immunodeficiency (Lawn et al. 2009c). TB multi-drug resistance has more recently been described both as a risk factor for TB-IRD and as a competing alternative diagnosis (Meintjes et al. 2009a). This potentially complicates both diagnosis and management. A randomized placebo-controlled trial of corticosteroids has been evaluated in the management of moderate (nonlife-threatening) TB-IRD in South Africa (Meintjes et al. 2009b). Although no mortality benefit was observed, there was a significant reduction in the need for hospitalization and interventions in the treatment arm. Although deaths associated with TB-IRD have been reported, it is evident that the contribution of such deaths to overall mortality in resource-limited settings is small and TB-IRD is not undermining ART programme outcomes (Lawn et al. 2007; Castelnuovo et al. 2009). MDR ⁄ XDR-TB Mycobacterium tuberculosis is an old pathogen, but recent changes in drug-susceptibility profiles have brought on new challenges for global TB control. The diagnosis and management of multidrug resistant-TB (MDR-TB) and extensively drug resistant-TB (XDR-TB) has recently been covered in detail elsewhere (Schaaf et al. 2009). Here, we focus on the epidemiology in Africa and on the management challenges facing clinicians in resource-poor settings. Multidrug resistant-TB is defined as resistance to rifampicin and isoniazid, whilst XDR-TB is defined as resistance to rifampicin, isoniazid, any fluoroquinolone and one of the second-line injectable agents, i.e. amikacin, kanamycin or capreomycin. Epidemiology In 2007, an estimated 5.3% or approximately 500 000 of all TB cases worldwide were because of MDR-TB strains; approximately 6% or 40 000 of these were estimated to be XDR (WHO 2009). This represents a near-doubling of MDR cases since 2000. About 60% of the MDR-TB case load is from regions of the former Soviet Union, India and China. However, there is a paucity of data from Africa, particularly from longitudinal cohorts (Wright et al. 2009).
ª 2010 Blackwell Publishing Ltd
Tropical Medicine and International Health
volume 15 no 8 pp 981–989 august 2010
W. A. Hanekom et al. Tuberculosis research update
Recent surveys showed an MDR-TB prevalence of 26% in Ethiopia (retreatment cases) (Meskel et al. 2008), 54% in Nigeria (tertiary hospital patients) (Kehinde et al. 2007), 9.5% in Zambia (prisoners) (Umubyeyi et al. 2007) and 9.4% in Rwanda (retreatment cases) (Umubyeyi et al. 2007). Statistical modelling suggests that MDR rates in Africa could be substantially higher than originally estimated (Nunn et al. 2007; Ben Amor et al. 2008). Overall, there appears to have been an alarming increase in the prevalence of MDR-TB in many areas of Africa, and better surveillance data are urgently needed. For example, a relatively small survey from the Cape Town region of South Africa showed that 4.8% of new TB cases and 10.5% of retreatment cases were caused by MDR strains in 2008 (Cox et al. 2009). This contrasted with rates of 1.6% and 4%, respectively, from a larger survey in 2002. Clinical presentation The clinical presentation of MDR-TB and XDR-TB is similar to that of drug-sensitive TB. Previous treatment for TB, a history of poor treatment adherence, a failing treatment regimen in the face of good adherence and contact with known cases of drug-resistant TB should raise suspicion about MDR-TB and XDR-TB. However, in up to 50% of cases, such risk factors may not be present. In HIVinfected patients, the clinical and radiological picture may be atypical, and disease progression may be rapid, particularly in the setting of nosocomial transmission with virulent strains (Gandhi et al. 2006). Advances in the diagnosis of MDR-TB are addressed in the ‘Diagnosis of TB’ section below. Management challenges A detailed description of specific treatment regimens for MDR-TB and XDR-TB has recently been outlined (Schaaf et al. 2009); selected aspects will be covered here. The WHO has outlined various treatment options for TB; e.g. regimen 1 consists of 2 months of rifampicin, isoniazid, pyrazinamide and ethambutol, followed by 4 months of rifampicin and isoniazid. If this regimen fails, it is important to establish the degree of drug adherence, and whether the diagnosis was correct, by considering alternative respiratory disorders, and in patients co-infected with HIV, other opportunistic infections, and IRD. A failing regimen 1 should never be replaced by WHO regimen 2, which consists of 8 months of therapy that includes streptomycin during the intensive phase, as this would constitute adding a single drug to a failing regimen. Selection of an MDR-TB treatment regimen must be based on local drug-susceptibility patterns and the patient’s
ª 2010 Blackwell Publishing Ltd
previous treatment history. All efforts should be made to obtain susceptibility data on the patient’s strain; however, a particular management challenge in resource-poor settings is lack of availability of drug-susceptibility testing. At least two drugs to which susceptibility is shown should be added to the regimen. High-dose isoniazid should be considered, as this is cheap and may overcome resistance at low mean inhibitory concentrations of the drug (Katiyar et al. 2008). Where failure of an MDR-TB treatment regimen occurs, in the face of good adherence, suspicion of XDR-TB should be raised. This presents an even greater challenge because several second-line drugs are often not available to clinicians in resource-limited settings. Although the definition of XDR-TB includes resistance to fluoroquinolones, we recently found that inclusion of moxifloxacin in the treatment regimen was an independent predictor of survival, in 199 patients initiating treatment for XDR-TB (Dheda et al. 2009). Studies are urgently required to determine the potency of specific quinolones against XDR strains of M. tuberculosis; in the meanwhile, we recommend, in the absence of specific susceptibility test results, that all patients with XDR-TB receive a regimen containing moxifloxacin. When localized XDR disease is diagnosed, early surgery should be considered, but this approach is only feasible when cardiopulmonary function permits. In the Western Cape Province of South Africa, we have observed failure using capreomycin-based regimens in a large proportion of patients with XDR-TB. These patients, as well as patients who are recurrently nonadherent to their regimens, pose ethical and medico-legal dilemmas, because of the risk of transmission of XDR strains to others (Bateman 2007; Singh et al. 2007). Many of these patients are now being discharged back into communities, after being hospitalized for prolonged periods. A co-ordinated response to community or hospital-based management of XDR-TB in relatively resource-limited settings is urgently required, to prevent further transmission and amplification of the XDR-TB problem. Infection control and the treatment of contacts of patients with XDR and MDR-TB have recently been reviewed elsewhere (Schaaf et al. 2009). Until newer therapeutic options (Mitnick et al. 2009), including the promising drug TMC 207 (Diacon et al. 2009) and newer diagnostic technologies become widely available, XDR-TB will be challenging to manage and continue to divert much needed resources away from existing treatment programmes. Prevention of XDR-TB is therefore paramount. This may be achieved through programmatic strengthening, improved laboratory capacity, access to newer therapeutic and diagnostic technologies, active case finding, and aggressive treatment of drug-sensitive and MDR-TB. This will only occur through sufficient political 983
Tropical Medicine and International Health
volume 15 no 8 pp 981–989 august 2010
W. A. Hanekom et al. Tuberculosis research update
will, de-escalation of war and conflict and large-scale reduction in levels of poverty within Africa. Outcome Treatment outcomes of MDR-TB in some settings have been encouraging (Mitnick et al. 2003; Yew & Leung 2008). This is not the case in more resource-poor settings, such as those that prevail in Africa. A recent report from the Western Province of South Africa showed that only 49% of 491 patients with MDR-TB were cured or completed treatment (Shean et al. 2008). In a hospital outbreak of XDR-TB, among persons with advanced HIV disease, in the Kwazulu-Natal province of South Africa, the median time to survival was approximately 2 weeks (Gandhi et al. 2006). These data suggested that XDR-TB strains in Africa are almost exclusively associated with HIV infection and that prognosis is invariably poor. Recent findings that overall mortality among 107 HIV-infected patients with XDR-TB was 41% (Dheda et al. 2009) challenge this perception. Furthermore, 53% of patients with XDR-TB from this cohort were HIV negative. Significant survival of HIV-infected patients with XDR-TB has now been confirmed by other investigators in South Africa (Schaaf et al. 2009). Nevertheless, treatment-related outcomes of XDR-TB in resource-limited settings remain poor, regardless of HIV status. The overall mortality in our recent cohort was 42%, and the 1 year mortality 36%, which is comparatively higher than the mortality of most aggressive malignancies (Dheda et al. 2009). Diagnosis of TB To control the worldwide TB epidemic, there is an urgent need for diagnosis of disease to be more rapid and reliable than is currently the case (Abu-Raddad et al. 2009). As discussed above, optimal management of drug-resistant TB requires rapid detection of resistance. We focus on advances in case detection and diagnosis of resistance; we will not discuss interferon-gamma release assays, as these have been reviewed recently (Lange et al. 2009). Microscopy The sensitivity of sputum smear microscopy is at best 70–80% in HIV-uninfected adults, and decreases in the settings of HIV and TB co-infection, in paediatrics and in extrapulmonary TB. However, this technique is the cornerstone of diagnosis in many developing countries and is likely to remain so for some years. Optimizing the sensitivity of microscopy is therefore important. 984
Currently, optimal sensitivity is achieved by liquefying and concentrating sputum samples, followed by fluorescent microscopy, using a stain such as auramine-O (Steingart et al. 2006). Fluorescent microscopes are expensive, their bulbs have a short lifespan and a reliable power supply is required, limiting usefulness in many resource-limited settings. A recent advance has been description of fluorescent microscopes using light-emitting diode (LED) technology (Anthony et al. 2006; Hanscheid 2008). The sensitivity of LED-based techniques has been shown to be similar to that of conventional fluorescent microscopy (Hung et al. 2007; Marais et al. 2008). LED microscopes are more affordable than conventional fluorescent microscopes, the bulb lifespan is longer (up to 50 000 h for LED vs. 200 h for a mercury vapour lamp) and some models may be powered by a battery (Anthony et al. 2006). These microscopes are therefore ideal for application in resource-limited settings. Culture Culture has long been the gold standard for diagnosing TB. Prolonged incubation time and the laboratory infrastructure required are major limitations. Automated broth-based mycobacterial culture systems, such as the mycobacteria growth indicator tube (MGIT) system (Becton Dickinson), reduce the time to detection of a positive culture by 50–60%, compared with culture on Lowenstein Jensen (LJ) slopes (Pfyffer et al. 1997; Somoskovi et al. 2006). Automated systems are expensive and not widely used in resource-limited settings. The microscopic observation drug-susceptibility (MODS) assay provides an attractive alternative, with the benefit of simultaneous susceptibility testing. The assay entails inoculating a sputum sample into wells of a tissue culture plate containing Middlebrook 7H9 broth; in addition, some wells contain set concentrations of isoniazid, rifampicin, ethambutol or streptomycin (Caviedes et al. 2000; Moore et al. 2006). The wells are examined for growth on a daily basis, using an inverted microscope: M. tuberculosis is identified by its typical corded appearance. Drug resistance is indicated by the presence of growth in the drug-containing wells. Time to detection of culture (approximately 8 days) is at least as good as, or better than, automated broth-based (Moore et al. 2006; Arias et al. 2007). An excellent correlation between MODS and reference methods for testing susceptibility of isoniazid and rifampicin has been shown (Albalak et al. 2007). While the MODS assay does not require expensive equipment, it is labour-intensive, and substantial laboratory infrastructure is still needed. Potential cross-contamination has been a concern, although this may be
ª 2010 Blackwell Publishing Ltd
Tropical Medicine and International Health
volume 15 no 8 pp 981–989 august 2010
W. A. Hanekom et al. Tuberculosis research update
minimized by placing each tray in its own plastic bag (Moore et al. 2006). Nucleic acid amplification techniques (NAATs) Many NAATs focus on detection of M. tuberculosis in clinical samples. A recent advance has been the introduction of commercial NAATs to detect drug resistance. Two commercial line-probe assays have been designed to detect either rifampicin resistance mutations in the rpoB gene (Inno-LiPA Rif TB assay; Innogenetics, Ghent, Belgium), or rifampicin and isoniazid resistance in rpoB and katG and inhA genes, respectively (Genotype MTBDRplus; Hain Lifesciences, Nehren, Germany). Line-probe assays were recently endorsed by the WHO for rapid detection of drug resistance (WHO 2008a,b). Using the Genotype MTBDRplus on smear-positive sputum samples (Barnard et al. 2008) could conceivably result in confirmation of presence of M. tuberculosis, and susceptibility to rifampicin and isoniazid, within 24–48 h of receipt of the sample. The Genotype MTBDRsl (Hain Lifesciences) is able to rapidly detect resistance to fluoroquinolones, amikacin, capreomycin and ethambutol (Hillemann et al. 2009). Larger scale evaluations are still needed but this test shows great promise. Drawbacks to many of the commercial NAATs include sometimes laborious sample processing, the risk of crosscontamination and the need for specialized equipment. The GeneXpert MTB test (Cepheid, Sunnyvale, CA, USA) is a nested, real-time PCR assay which detects M. tuberculosis in sputum samples as well the presence of mutations in the rpoB gene by means of molecular beacons. Sample preparation is minimal (approximately 15 min) and involves inoculation into a cartridge, which is inserted into the GeneXpert instrument, for automated DNA extraction, PCR amplification and product detection within the cartridge. This minimizes the chance of crosscontamination. Results may be available within about 2 h of sample receipt. The manufacturers claim a sensitivity in smear-negative cases of >95%, although these studies were conducted in settings with relatively low HIV-TB co-infection rate (