© 2004 American Thoracic Society
Tuberculosis due to Multiple StrainsA Concern for the Patient? A Concern for Tuberculosis Control?McGill University Health Centre Montreal, Quebec, Canada The past decade has witnessed a renaissance in tuberculosis (TB) epidemiology, largely driven by the now-established discipline of molecular epidemiology (1). Using bacterial genetics to track organisms in the community, investigators have challenged many of the prevailing dogmas about TB epidemiology to suggest enhanced strategies for TB control (2–4). Central to these studies has been the premise that TB generally results from a single strain of Mycobacterium tuberculosis, and therefore, that epidemiologic inferences about transmission can be derived from genetic study of these single patient isolates. In the current issue of the Journal (pp. 610–614), Warren and colleagues now question this central tenet and propose that an important minority of patients with TB in South Africa simultaneously have multiple strains of M. tuberculosis (5). Their results beg a number of questions, from technical issues regarding their methodology to the resulting clinical, epidemiologic, and immunologic inferences. Based on methodologic issues, arguments can be presented to suggest how results of Warren and coworkers underestimate or overestimate the true frequency of multiple infections. The technique developed essentially dichotomized all M. tuberculosis strains into Beijing-type strains or non-Beijing strains, and based on this simplistic division, it is remarkable that mixed infection could be documented at all. It therefore stands to reason that by using targets specific to more strains, a considerably higher frequency of mixed infections should be detectable. Conversely, it is concerning that the same extracted DNA yielded 35 episodes of mixed infection using the polymerase-chain reaction (PCR)–based technique (19%) and just 9 such instances using the PCR-based fingerprinting method called spoligotyping (5%) (5). This difference suggests that the results may benefit from independent corroboration. Finally, epidemiologic factors could influence the observed results. The current study was restricted to patients with microscopically acid-fast positive disease in an extremely high-risk environment. One suspects a greater likelihood of detecting mixed infections where bacterial numbers are high, and that the probability that patients with TB will be super-infected increases with the local rate of infection. These factors suggest that studies done on all patients with TB in settings with lower TB rates would yield reduced estimates of mixed infection. Put together, it is possible that some of these considerations cancel out, so perhaps 5–20% of patients with TB still have mixed infections. Does this matter? For the clinician, the answer is pending. The current report unfortunately provides no information on whether patients with mixed infections differed in clinical presentation or response to treatment (5). Because antimicrobial susceptibility testing is based on phenotypic resistance, one hopes that a drug-resistant strain would be detectable by its growth in the presence of antibiotics, thereby providing an accurate report of resistance. The most concerning scenario would involve the emergence of the undetected drug-resistant strain under the pressure of antibiotic treatment, presenting as clinical relapse with a new strain. While anecdotal reports attest to this possibility (6), treatment failure with a new strain is infrequent even in a high-risk setting (7), and only a subset of these could be due to dual infection. Therefore, while the clinician might query mixed infection when faced with a complicated treatment course, the demonstration of multiple infections is unlikely to impact more generally on standard therapeutic regimens. What are the epidemiologic issues? If mixed populations can coexist in the same host, it is conceivable that patients who apparently relapse with a new strain of TB were dually infected from the start, and do not represent exogenous reinfection (8, 9). It is also conceivable that patients with TB in the same household have different strains because of transmission of the undetected strain, instead of a different source of infection (10). While conceivable, these scenarios would appear uncommon. If 5–20% of patients have mixed infection, it follows that at least 80% of patients with TB have just one strain of M. tuberculosis. Moreover, to bias epidemiologic studies, patients with mixed infection would have to be more likely to fail therapy or spread their undetected strain. Therefore, aside from a small risk of misclassification, inferences from molecular epidemiologic studies appear safe, unless the frequency of multiple infections is shown to be significantly greater than reported by Warren and colleagues (5). The major lesson of multiple infections appears to relate to the natural immunity of TB in humans. While disease from the first strain may render the host more susceptible to super-infection, as proposed by Warren and colleagues (5), it is also possible that disease is merely a marker of an inadequate immune response, with the first infection neither predisposing to nor protecting against super-infection. The maximum level of protective efficacy registered in clinical trials of bacillus of Calmette and Guerin vaccines was near 80% (11), mirroring cohort studies of nurses and medical students in which prior evidence of tuberculous infection was associated with approximately 80% less incident TB (12). These results imply that a subset of people cannot mount a protective immune response against M. tuberculosis, whether via immunization or natural infection, and are presumably prone to super-infection on repeat exposure. Another possibility is strain-specific immunity, i.e., some degree of protection against the same strain, without protection against others. This consideration may be generic, pertaining to any combination of two strains, or particular to certain strains, such as an unusual capacity for super-infection by the highly prevalent Beijing strain. This latter possibility can be addressed with a refined version of the strain-typing tool that permits testing for multiple strain families. From this will also emerge more information on the types of strains circulating and their relative capacity to super-infect, thereby confirming or refuting the notion of strain-specific immunity in TB. So, do these results challenge TB control? As current TB control relies on treatment, the impact of multiple-strain disease on clinical management of TB remains to be determined, but is hopefully minimal as noted above. As for the future of TB control, an effective vaccine is desperately needed, and efforts to develop and test new vaccines are gaining momentum (13). The report of Warren and colleagues provides a timely reminder of the challenges ahead. If TB itself does not provide complete protection against M. tuberculosis, the designers of a new vaccine must either aim to do better than natural disease or accept this inherent imperfection. Compared with the uncertainty surrounding bacillus of Calmette and Guerin vaccination, an imperfect vaccine would be an important step forward. FOOTNOTES Conflict of Interest Statement: M.A.B. has no declared conflict of interest. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
