Pulmonary disease due to nontuberculous mycobacteria: an epidemiologist’s view

Until recently, the epidemiology of nontuberculous mycobacterial (NTM) disease has been largely a ‘black box’. Collectively, these organisms (e.g., Mycobacterium avium, Mycobacterium kansasii and Mycobacterium abscessus) cause both extrapulmonary and pulmonary disease in susceptible hosts, the latter frequently occurring in seemingly immunocompetent individuals and generally manifesting as a slowly progressive, often debilitating lung disease. In the last several years, new data suggest NTM disease is not uncommon, and while first described in males decades ago, elderly females now appear to bear the brunt of this disease. While these papers argue that NTM disease should be put ‘on the map’ as an environmental health problem worthy of further research, they also remind us that very little is known regarding pulmonary NTM disease epidemiology, its natural history, and how best to treat and prevent the disease. This editorial will summarize the latest developments in pulmonary NTM disease epidemiology, and highlights key areas necessary for future research.

Within North America, pulmonary NTM disease was first described in elderly males with chronic lung disease in southeastern USA. Over the last 20 years, experts have reported a general increase in pulmonary NTM, with a number of treatment centers reporting increasing numbers of female patients in particular [1]. In 1990, and then in 1997, the American Thoracic Society (ATS) issued statements on the diagnosis and treatment of pulmonary NTM disease [2,3]. In 2007, the ATS and the Infectious Diseases Society of America (IDSA) jointly issued an updated statement [1]. It is perhaps surprising that for a disease in which three major statements have been issued, there have been virtually no population-based data made available that firmly document those most basic questions of epidemiology: the ‘who, what, where and how much?’ One major reason is that NTM are environmental organisms ubiquitous in soil and water, and they can be present in respiratory microbiologic samples in the absence of disease. Accordingly, laboratory-based studies tracking the incidence of NTM isolation in the population cannot distinguish between diseased and nondiseased persons without evaluating the clinical and radiographic records of patients with isolates.

However, there have been population-based studies evaluating trends in NTM laboratory isolation. These have consistently reported NTM isolations to be increasing in absolute number and as a proportion of all mycobacterial isolates identified [1,4]. Yet, to date, no investigators have been able to review clinical records and apply the ATS/IDSA NTM case definition (Box 1) to determine the proportion of patients with isolates that actually have NTM disease. O’Brien et al. surveyed state public health laboratories within the USA in the early 1980s and found 90% of all NTM isolates were of respiratory origin. The investigators reported NTM isolation incidence and then estimated a population NTM disease rate of 1.8/100,000 by surveying TB controllers to estimate the proportion of their isolates that represented disease [5]. This survey did not include private laboratories and therefore could have substantially underestimated isolation incidence or prevalence from the general population. Much more recently, Marras et al. found the incidence of NTM isolation in Ontario, Canada, to have significantly increased from 9.1/100,000 in 1997 to 14.1/100,000 in 2007 [4]. The reason for greater numbers of NTM noted in the last several decades is likely to be multifactorial (including more sensitive laboratory isolation techniques); however, at least one potential contributing factor is that more people are infected [4,6].

Notably, one study published in 2009 relied upon the microbiologic criteria of the ATS/IDSA case definition in order to estimate disease prevalence [7]. Rather than reporting isolation prevalence, this study applied the microbiologic component of the ATS/IDSA case definition to estimate disease. This method assumes that for patients meeting the microbiologic criteria (Box 1), they have coexistent radiographic abnormalities and symptoms compatible with NTM disease. Although the accuracy of this approach is unknown, it is likely to be a reasonable assumption, as most patients undergoing bronchoscopic or sputum evaluation are doing so because of radiographic or symptomatic findings. This statewide public health project in Oregon recorded patient and specimen information from 100% of all NTM isolates within the state for a 2-year time period (2005–2006). In applying the microbiologic criteria, Cassidy et al. estimated annual pulmonary NTM disease prevalence to be 5.6/100,000 statewide, but as high as 15.5/100,000 for those over 50 years of age, and greater than 20/100,000 in those over 70 years of age [7].

Contrary to early NTM disease reports where males were predominately affected, Cassidy et al. also documented across a population what many experts and institutional case-series had suggested for the last several decades: this is no longer a male dominated disease. Rather, elderly women are disproportionately affected, making up nearly 60% of prevalent cases [7]. Although this study did not report the clinical characteristics of these patients, from previously published case series it is known that many females with this disease have no history of underlying lung disease or smoking prior to their diagnosis [8]. A second study, conducted on the US west coast within several large health maintenance organizations, employed similar techniques, using the ATS/IDSA microbiologic criteria to estimate disease prevalence. This study is currently in peer review, but their findings were presented in abstract form at the 2009 ATS conference where their data suggested similar yearly prevalence estimates for NTM pulmonary disease and also a female predominance [9]. Billinger et al. recently reviewed state-level hospital discharge data for patients with international classifications of diseases-9 discharge diagnoses of pulmonary NTM. In an 8-year time period from 1998 to 2005, a significant annual increase in such diagnoses was noted in New York and Florida, although rates in California were stable during the time period. Overall, annual prevalence rates were highest in women aged 70 years and older (9.4/100,000) compared with similarly aged men (7.6/100,000) [10]. While this study did not seek to ascertain NTM disease prevalence, and the validity of international classifications of diseases-9 codes for pulmonary NTM is unknown, their data are consistent with the Oregon project in portraying this disease as one that disproportionately affects elderly females and one that surpasses NTM’s more notable ‘cousin’ (tuberculosis) in prevalence.

We presume that patients at risk for pulmonary NTM acquire their disease from environmental sources such as municipal water, soil and other potential water exposures [11]. This had been documented in cases of M. aviumhot tub lung and in a handful of cases of pulmonary NTM where patients had identical pulsed-field gel electrophoresis matching isolates obtained from lung and their home or work water supplies [12–14]. However, aside from these few investigated cases, little scientific work has successfully documented the types of environmental exposures necessary to cause disease and their risks to individuals in the community. Interestingly, the Oregon project found significantly higher disease rates in the urban, wetter portions of the state versus the eastern more arid, and rural portion of the state [7]. Their findings suggest that either climatic factors and/or differences in urban versus rural water supplies might influence the risk of disease. These ideas are ripe for further investigation.

A total of 3 years ago we lacked firm estimates for pulmonary NTM disease incidence or prevalence in North America. We now have prevalence estimates based on the ATS/IDSA microbiologic criteria, and these estimates suggest that pulmonary NTM disease is not uncommon among our increasingly elderly population. It should be noted that few prevalence estimates for this disease are available from Europe or elsewhere in the world, but that the emerging picture from Europe is of a different epidemiology, namely more male cases and more disease with species rarely or less frequently encountered in North America (namely Mycobacterium malmoense and Mycobacterium xenopi) [15]. There are many next steps. In the immediate future, it is necessary to validate the use of the ATS/IDSA microbiologic criteria as a proxy for true disease. If validated, laboratory-based surveillance for pulmonary NTM disease would be possible and would facilitate future surveillance efforts. Further, additional epidemiologic work must be done to ascertain disease risk factors and identify potentially modifiable exposures for those who are at risk. If, in fact, most patients are infected from our public water supplies, then efforts should be made to better understand determinants of mycobacterial concentration in our municipal water supplies, their risk and efforts made to mitigate any risk they might pose. Beyond the epidemiology, the need for host genetic and pathogen virulence studies, molecular epidemiologic analyses and clinical trials examining drug therapy cannot be understated. Much is to be done. The epidemiology is the foundation for these efforts, and the data collected to date suggest that such efforts are becoming increasingly worthwhile.