Marras, Theodore K. MD*; Daley, Charles L. MD†
After isolating a nontuberculous mycobacterial (NTM) organism from a respiratory specimen, the clinician must determine whether the patient has either clinically significant disease and therefore warrants treatment, or an indolent infection, not requiring treatment at that time. Mycobacterium kansasii is generally considered to be one of the most pathogenic NTMs because it is most frequently associated with clinically important disease at the time of isolation. In a review of studies from around the world, an average of approximately 50% of M. kansasii respiratory cultures represented clinically important disease, compared with 20–25% with most other species. 1 When isolated from respiratory specimens, M. kansasii is far more likely to be associated with clinically important disease than other NTMs and in the setting of HIV infection, M. kansasii usually presents with pulmonary disease.
However, the American Thoracic Society (ATS) guidelines apply the same diagnostic criteria to M. kansasii as to other NTM species and recommend that clinical disease should be diagnosed only in the presence of multiple positive cultures. 2 The situation in patients infected with HIV is more controversial. Some authorities argue that a positive respiratory culture for M. kansasii in an HIV-infected patient always represents clinically relevant disease, 3–7 while the ATS diagnostic criteria again do not distinguish between the different NTM species. Deciding whether a patient with M. kansasii isolated from a respiratory specimen has disease, and therefore requires treatment, is very important because the treatment requires that multiple antimycobacterial drugs be given for a prolonged period. The decision to treat or not treat is particularly salient in the HIV-infected population because there is a high frequency of drug reactions in HIV patients in general and there are important interactions between rifamycins, particularly rifampin, and certain antiretroviral agents. 8,9
The term colonization has been used to describe NTM infections that do not require treatment. This term may be misleading, implying that there is no risk of eventually developing clinically important disease and requiring treatment. Given this uncertainty, we prefer to use the term indolent infection in the context of a respiratory NTM isolate without definite clinical disease. Given the uncertainties surrounding pulmonary M. kansasii infection in HIV-infected patients, we performed a systematic review to address 2 main objectives. First, we assessed whether HIV-infected patients with pulmonary M. kansasii isolates may have indolent infection, not requiring immediate treatment, or if all such patients have clinically significant disease. Second, we attempted to determine the utility of the ATS diagnostic criteria in distinguishing between patients with disease vs. indolent infection among HIV-infected patients with pulmonary M. kansasii isolates.
We identified all studies of HIV-infected adults with positive respiratory cultures for M. kansasii published in the English language literature. Studies were included if they presented clinical data on more than one HIV-infected patient, described case ascertainment methods, distinguished between patients according to the presence of symptoms and source of isolates (pulmonary vs. extrapulmonary), and provided information on clinical follow-up. Studies including patients without HIV infection were included if data were presented on HIV-infected patients separately. Medline (1966–July 2003) and Embase (1980–July 2003) were searched using the terms Mycobacterium kansasii combined with HIV, HIV-1, or HIV infections, searched as medical subject headings and text words as appropriate. The bibliographies of all retrieved studies and several pertinent review articles were also reviewed. The abstracts of all studies were reviewed for eligibility. One author (T.K.M.) reviewed selected studies to collect data on methods, patient characteristics, treatment, and outcomes. We expected that the available literature would be limited to primarily retrospective cohorts and that definitions of colonization (indolent infection) and disease between studies would be variable. In anticipation of these limitations, we decided to include all studies and calculate pooled rates of patient characteristics, treatments, or outcomes using data from only the studies in which the relevant information was presented. In the“Results” section, we use the term colonization, consistent with the majority of the reviewed studies. In the “Discussion” section we revert to the term indolent infection, to stress the potential of important disease in the future. Data were entered into a Microsoft Excel (Office 2000; Microsoft, Redmond, WA) spreadsheet, and all calculations were performed using this software. Due to study heterogeneity, formal meta-analysis techniques were judged inappropriate and therefore not performed.
The search strategy yielded 86 citations whose abstracts were reviewed, of which 22 studies were selected for full review. Based on full review, 6 studies were excluded for the following reasons. Two studies presented inadequate clinical detail, and one each did not: contain any cases of M. kansasii; provide information on M. kansasii–infected patients separately from patients with other NTM infections; specify if cultures were from a pulmonary source; or distinguish patients according to HIV status. Sixteen studies were thus retained for this review, reporting on between 9 and 187 subjects each (median 23). 3–7,10–20 Due to overlap between studies, the total number of patients included could not be determined, but lies between 573–646. All studies were retrospective. Studies were published between 1991 and 2003 and included patients from the United States, 3,4,10–13,16 Europe, 7,17–20 and South Africa. 6,14,15 Table 1 summarizes study location, time period, patient population, and definitions of colonization and disease. Colonization and disease were generally defined using a combination of microbiologic, symptomatic, and radiographic criteria as summarized for each study in Table 1. In studies reporting the following characteristics, the mean age was 44 years (based on 10 studies including 443 patients), 3–5,11–14,17–19 91% were male (based on 13 studies including 504 patients), 3–6,11–15,17–20 64% were men who had sex with men (based on 6 studies including 148 patients), 3–5,17–19 32% used injection drugs (based on 7 studies including 353 patients), 3–5,11,18–20 78% smoked tobacco (based on 5 studies including 113 patients), 3–5,18,19 and 36% were alcohol dependent (based on 3 studies including 50 patients). 3,18,19 The median CD4 lymphocyte count was reported in 6 studies (n = 339 patients) and ranged from 12–381 cells/μL. 3,5,6,10–12 The pooled mean annual infection rate was 532 of 100,000 people (median 115, range 73–2667), based on data from 7 studies, containing 532 patients. 10 Table 2 summarizes data regarding infection severity and frequency of dissemination. Data regarding the frequency of pulmonary colonization were derived from 11 studies with a median colonization rate across studies of 8% (range 0–83%). 3–5,7,10,11,15–19 Information on the number of specimens obtained for culture was generally not provided. The most common presenting symptoms and signs included fever in 79% (based on 9 studies including 199 patients), 3–5,7,10,12,13,16,18 weight loss in 56% (based on 4 studies including 107 patients), 4,5,12,13 cough in 66% (based on 8 studies including 183 patients), 3–5,7,10,12,13,18 dyspnea in 42% (based on 7 studies including 171 patients), 3–5,7,10,12,13 and chest pain in 22% (based on 6 studies including 129 patients). 3,5,7,12,13,16 Chest radiographic abnormalities were varied and are summarized in Table 3. Interstitial (38%) 3–5,12–14,19 and airspace (39%) 5,12,18,19 opacities were most commonly seen, whereas cavitation and nodal enlargement were observed in approximately one-quarter3–6,10,12–4,17,18 and one-fifth 12–14,16–18 of patients, respectively. Overall, among 7 studies that presented coinfection data, 37% of patients were also infected with other pulmonary pathogens. 3,5,11–13,16,18 Co-pathogens most commonly included other NTMs in 14% (28 patients in 6 studies of 201 patients) 3,5,11–13,16 and Pneumocystis jiroveci in 12% (25 patients in 6 studies of 201 patients). 3,5,11–13,16
We assessed for differences between patients with exclusively pulmonary infection vs. patients with both pulmonary and extrapulmonary infection. Of 11 studies presenting data on patients with disseminated infection, 3–5,7,10,12,13,17–20 only 4 studies presented data to suggest differences between patients with and without extrapulmonary infection. 10,12,13,19 Two studies presented data suggesting that patients with disseminated disease had lower CD4 lymphocyte counts, but neither achieved statistical significance (mean 21 vs. 60, P = 0.08, Campo and Campo 12; and median 7 vs. 20, P = 0.18, Pintado et al 19). Bamberger et al 10 observed a lower rate of upper lobe abnormalities on chest radiographs in patients with pulmonary plus extrapulmonary infection (1/4, 25%) vs. infection limited to the lung (8/17, 47%, P = 0.60). Carpenter and Parks 13 observed a lower rate of chest radiographic abnormalities in patients with pulmonary plus extrapulmonary infection (2/4, 50%) vs. infection limited to the lung (4/4, 100%, P = 0.43). The differences in radiographic presentation were not statistically significant in either study. These studies were limited by the lack of systematic extensive microbiologic investigation in all patients and small sample size.
The majority of patients studied were treated with anti-mycobacterial therapy but indications for treatment were generally not described. The mean duration of therapy was 12 months. 5,6,12,17,18 Seventy-two percent of patients received a rifamycin, 64% received ethambutol, and 61% received isoniazid. 3,5–7,10,12,13,17–19 Less commonly used agents included clarithromycin in 7%, a fluoroquinolone in 6%, and clofazamine in 2%. 7,10,12,17,19 The majority of treated patients (65%; 101/155, reported in 9 studies) were thought to have improved. 3,5–7,10,12,13,17,19 Survival data are summarized in Table 4. The mean duration of survival in treated patients was 3-fold that of untreated patients (12 vs. 4 months) 3,5,7,10,12,13,17–19 and there was a lower rate of total and M. kansasii–related deaths in the treated groups. 3,10,12,13,17–19
Regarding whether unfulfilled diagnostic criteria can be used to define the absence of disease (indolent infection), we found that of the 7 studies that employed ATS or similar criteria, 3,6,7,13,17,18,20 only 3 studies included patients with only a single respiratory isolate, 3,7,17 and only 2 of these studies reported on patients who did not fulfill criteria and were left untreated for M. kansasii. 3,17 Levine and Chaisson 3 observed 2 patients, both of who had M. kansasii isolated and P. jiroveci identified and improved with trimethoprim-sulfamethoxazole alone. One patient subsequently died of other causes and the other later deteriorated but was alive at last follow-up. These patients would have been classified as not having disease according to the ATS criteria and their outcomes were, at least initially, favorable. The study of Klein et al 17 included one patient who did not fulfill ATS diagnostic criteria and was not treated. The utility of ATS criteria could not be evaluated in this patient, however, as he rapidly deteriorated and died of disseminated infection with Aspergillus species.
The first objective of this review was to determine whether HIV-infected patients who have M. kansasii isolated from respiratory specimens may have an indolent infection or if all such patients have clinically significant disease. Based on the review presented here, it seems that some HIV-infected patients with respiratory M. kansasii isolates do indeed have an indolent infection. Several studies classified patients as colonized, 10,12,13,16,17,19 but the definitions varied between reports and the long-term significance of this designation is unclear. A reasonable and clinically useful definition of colonization would include an asymptomatic or minimally symptomatic patient (with respect to the presence of M. kansasii) and the absence of severe or irreversible sequela of M. kansasii infection over time. We propose the term indolent infection, as described in the introduction. The above-noted studies include patients who seem to fit into this category, but they were identified retrospectively in very small numbers, making it difficult to have confidence in definitions developed from the patient characteristics. Furthermore, duration of follow-up was limited to <5 years, which may not be adequate in an age in which HIV-infected patients may survive for many years with the aid of highly active antiretroviral therapy (HAART).
The ultimate goal of future investigations should be to identify which patients with respiratory M. kansasii isolates will have good outcomes without treatment, a goal that will require either prospective study on a relatively large scale or a retrospective study with access to detailed follow-up on a relatively large group of patients. In general, most studies reported that a small but definite proportion of patients were “colonized” and thus seemed to have indolent infection. 3–5,10,11,17,19 The rate of indolent infection may be an overestimate if specimen contamination was the cause of some of the positive cultures, presumably more likely in patients who were deemed not to have clinically significant disease. Unfortunately, none of the reviewed studies addressed false positivity due to contamination. Based on our own clinical experience, we speculate that the rate of contamination was likely very low, but we cannot directly address this issue.
Our second objective was to determine the utility of ATS criteria in distinguishing between patients with disease vs. indolent infection. This objective is more difficult to achieve with the available literature. Given that only 3 relevant case descriptions were found in this review, 3,17 there are only very limited data to support the negative predictive value of the ATS diagnostic criteria in predicting the absence of disease in patients with a respiratory M. kansasii isolate who do not fulfill criteria for disease. Additional studies are needed to address this objective.
The positive predictive value of the ATS criteria intuitively does not require rigorous evaluation, given the inclusion of symptoms as an integral component of the criteria. Regardless, we found ample evidence that patients who fulfilled criteria clinically improved with therapy for M. kansasii, 3,7,13,17,18 supporting the somewhat self-evident impression of good positive predictive value. In addition, in the pooled estimate of outcomes, we found that therapy was beneficial in the overall group of patients, with markedly longer survival in treated patients. The data on survival according to treatment status must, however, be taken in context. Patients were treated according to clinician preference, which may have introduced bias in the assessment of treatment effects. If clinicians had tended to preferentially treat the sicker patients, this would have made the effects of treatment look less good. If clinicians had tended to withhold treatment from the sickest patients (due to an impression of futility) or if the diagnosis was made postmortem in some patients, then the apparent beneficial effect of treatment would be inflated. Given the tendency for treating most patients in most studies, and several instances of postmortem diagnoses, we expect bias in favor of the treatment groups, inflating apparent survival benefit of treatment. However, we believe that the treatment of clinically significant pulmonary M. kansasii infection prolongs survival in the setting of HIV infection.
The high rates of infection and clinical disease observed across studies support the notion that, with HIV infection, M. kansasii infection is more common and is more commonly associated with clinical disease. Based on population at-risk denominators provided in the studies, the annual rate of M. kansasii infection (defined by the presence of a respiratory isolate) in HIV-infected patients is 532 in 100,000, with significant geographic variation. Studies in the southern and western United States reported a range of rates of 100–2667 in 100,000 5,10,11,13; in South Africa, one study reported a rate of 320 in 100,000, 15 whereas a study in the United Kingdom and a study in Spain reported rates of 92 in 100,000 17 and 73 in 100,000, 19 respectively. As expected, this is markedly higher than non–HIV-infected patients, reported at 0.06–2.2 in 100,000 in the majority of studies included in a recent review. 1 The rate of infection—which is several orders of magnitude greater—is paralleled by a higher, but less dramatically elevated, rate of significant disease. We found a median rate of significant disease of 92% in HIV-infected patients, compared with the rate of approximately 50% in non–HIV-infected hosts reported elsewhere. 1 The higher rate of significant disease in HIV-infected people could in part reflect more meticulous assessment of HIV-infected patients who are found to have an isolate of M. kansasii. Conversely, clinicians may have a lower threshold to obtain respiratory specimens for mycobacterial culture in HIV-infected patients and thus potentially find more isolates that are not clinically significant. Given the bidirectional possibility of bias, we think that there is a higher rate of significant disease in HIV-infected patients, but the magnitude of the difference cannot be precisely estimated. It seems true that HIV-infected patients are not only far more likely to have M. kansasii recovered from their respiratory secretions, but once infected, they are also approximately twice as likely to develop significant disease.
The limited conclusions that may be drawn from the available literature illustrate the difficulty in studying diagnostic criteria for M. kansasii in HIV-infected patients. HIV-infected patients may simultaneously have evidence of pulmonary infection with multiple organisms. In this context it is often difficult to decide which organisms are responsible for existing signs and symptoms and if an organism like M. kansasii is clinically significant. Similarly, cause of death may be extremely difficult to ascertain in a patient infected with multiple organisms. HIV-infected patients also tend to have limited survival, which may result in duration of follow-up that is inadequate to determine whether a respiratory isolate is clinically significant. These problems are more important in patients with advanced immunosuppression, since antiretroviral medications remain inaccessible to many HIV-infected patients and are not universally tolerated and effective.
Two important limitations of this review deserve mention. First, the studies included were all limited by the challenges in studying HIV-infected patients, outlined above. These challenges were compounded by the fact that all studies except one predated the era of HAART, 20 contributing to the complexity of patients’ presentations and difficulties in determining isolate significance and cause of death. Second, given the heterogeneity of the studies, especially with respect to definitions of M. kansasii lung disease, it was deemed inappropriate to use techniques of meta-analysis to yield additional statistical measures from combined results. The data presented in the original studies were too limited to adequately test uniform diagnostic criteria or develop a model to identify long-term survivors not requiring treatment. The amount of detail that would be required to address these objectives made contacting authors for additional information impractical.
In summary, our review of pulmonary M. kansasii infection in HIV-infected patients suggests that most HIV-infected patients with respiratory cultures positive for M. kansasii have clinically important disease and should be treated. However, a single isolate may not necessarily indicate the presence of disease, and in some cases patients have an indolent infection, not requiring treatment. In the setting of indolent infection, careful clinical and radiologic follow-up is mandatory because we are unable to predict which, if any, patients will never require therapy for the M. kansasii infection. Based on the results of our review, we propose recommendations for the further assessment and management of HIV-infected patients with respiratory M. kansasii (Fig. 1). Specific drug treatment recommendations are provided by the ATS guidelines 2 and in a recent review. 21,22 The clinician must use judgment to decide when to embark on the long and complicated course of treatment of this organism, but a decision must be made more expeditiously in patients with advanced immunosuppression who could develop overwhelming infection in a relatively short period of time. Clinical (severity of immune suppression, presence of symptoms unexplained by other disease), radiographic (severity of abnormalities on chest imaging), and microbiologic criteria (acid-fast smear status and other indices of bacillary burden) should all be considered in making this decision. Although the ATS diagnostic criteria appear to intrinsically have a good positive predictive value in this population, their negative predictive value, the likelihood of disease absence when criteria are unfulfilled, is difficult to study and is supported by very limited data. More information is required to assist the clinician in identifying which patients do not require treatment.
1. Marras TK, Daley CL. Epidemiology of human pulmonary infection with nontuberculous mycobacteria. Clin Chest Med. 2002;23:553–567.
2. American Thoracic Society. Diagnosis and treatment of disease caused by nontuberculous mycobacteria. Am J Respir Crit Care Med 1997;156: S1–S25.
3. Levine B, Chaisson RE. Mycobacterium kansasii: a cause of treatable pulmonary disease associated with advanced human immunodeficiency virus (HIV) infection. Ann Intern Med. 1991;114:861–868.
4. Valainis GT, Cardona LM, Greer DL. The spectrum of Mycobacterium kansasii disease associated with HIV-1 infected patients. J Acquir Immune Defic Syndr. 1991;4:516–520.
5. Witzig RS, Fazal BA, Mera RM, et al. Clinical manifestations and implications of coinfection with Mycobacterium kansasii and human immuno-deficiency virus type 1. Clin Infect Dis. 1995;21:77–85.
6. Corbett EL, Blumberg L, Churchyard GJ, et al. Nontuberculous myco-bacteria: defining disease in a prospective cohort of South African miners. Am J Respir Crit Care Med. 1999;160:15–21.
7. Rooney G, Nelson MR, Gazzard B. Mycobacterium kansasii: its presentation, treatment and outcome in HIV infected patients. J Clin Pathol. 1996;49:821–823.
8. Spradling P, Drociuk D, McLaughlin S, et al. Drug-drug interactions in inmates treated for human immunodeficiency virus and Mycobacterium tuberculosis infection or disease: an institutional tuberculosis outbreak. Clin Infect Dis. 2002;35:1106–1112.
9. Burman WJ, Gallicano K, Peloquin C. Therapeutic implications of drug interactions in the treatment of human immunodeficiency virus-related tuberculosis. Clin Infect Dis. 1999;28:419–429.
10. Bamberger DM, Driks MR, Gupta MR, et al. Mycobacterium kansasii among patients infected with human immunodeficiency virus in Kansas City. Clin Infect Dis. 1994;18:395–400.
11. Bloch KC, Zwerling L, Pletcher MJ, et al. Incidence and clinical implications of isolation of Mycobacterium kansasii: results of a 5-year, population-based study. Ann Intern Med. 1998;129:698–704.
12. Campo RE, Campo CE. Mycobacterium kansasii disease in patients infected with human immunodeficiency virus. Clin Infect Dis. 1997;24: 1233–1238.
13. Carpenter JL, Parks JM. Mycobacterium kansasii infections in patients positive for human immunodeficiency virus. Rev Infect Dis. 1991;13: 789–796.
14. Corbett EL, Churchyard GJ, Hay M, et al. The impact of HIV infection on Mycobacterium kansasii disease in South African gold miners. Am J Respir Crit Care Med. 1999;160:10–14.
15. Corbett EL, Churchyard GJ, Clayton TC, 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.
16. Fishman JE, Schwartz DS, Sais GJ. Mycobacterium kansasii pulmonary infection in patients with AIDS: spectrum of chest radiographic findings. Radiology. 1997;204:171–175.
17. Klein JL, Corbett EL, Slade PM, et al. Mycobacterium kansasii and human immunodeficiency virus co-infection in London. J Infect. 1998;37: 252–259.
18. Lortholary O, Deniel F, Boudon P, et al. Mycobacterium kansasii infection in a Paris suburb: comparison of disease presentation and outcome according to human immunodeficiency virus status. Int J Tuberc Lung Dis. 1999;3:68–73.
19. Pintado V, Gomez-Mampaso E, Martin-Davila P, et al. Mycobacterium kansasii infection in patients infected with the human immunodeficiency virus. Eur J Clin Microbiol Infect Dis. 1999;18:582–586.
20. Santin M, Alcaide F. Mycobacterium kansasii disease among patients infected with human immunodeficiency virus type 1: improved prognosis in the era of highly active antiretroviral therapy. Int J Tuberc Lung Dis. 2003;7:673–677.
21. Griffith DE. Management of disease due to Mycobacterium kansasii. Clin Chest Med. 2002;23:613–622.
22. Jones D, Havlir DV. Nontuberculous mycobacteria in the HIV infected patient. Clin Chest Med. 2002;23:665–674.
© 2004 Lippincott Williams & Wilkins, Inc.