Published ahead of print on September 18, 2003, doi:10.1164/rccm.200307-910OC
© 2003 American Thoracic Society Tuberculosis in the Inuit Community of Quebec, CanadaDepartment of Medicine and Research Institute, McGill University Health Centre, Montreal; Direction de Santé Publique du Nunavik, Kuujjuaq; and Laboratoire de Santé Publique du Québec, Sainte-Anne-de-Bellevue, Quebec, Canada Correspondence and requests for reprints should be addressed to Marcel Behr, M.D., Division of Infectious Diseases and Microbiology, A5.156, Montreal General Hospital, 1650 Cedar Avenue, Montreal, PQ, H3G 1A4 Canada. E-mail: marcel.behr{at}mcgill.ca
In low-incidence countries targeting tuberculosis (TB) elimination, TB remains a problem of a few high-risk groups. In Canada, Aboriginals, and particularly the Arctic Inuit communities, have witnessed dramatic decreases in TB during the 1960s to 1970s, but rates remain at least 10 to 20 times higher than the national average. We are describing the results of an integrated traditional and molecular epidemiology study of all culture-positive Mycobacterium tuberculosis cases in the Arctic Inuit communities of Quebec from 1990 until 2000. The demographic characteristics of the 46 TB cases included in the study were most notable for a bimodal age distribution (48% under 25 years). Genotyping analysis using multiple modalities (IS6110 restriction fragment length polymorphism, spoligotype, mycobacterial interspersed repetitive units–variable number tandem repeats) showed that 76% (35/46) of TB cases were clustered (six clusters, median size four cases) and estimated that at least 62.5% of TB cases were due to ongoing transmission. By integrating the epidemiologic and genotyping data, we observed that the genotyping clustering results were concordant with recognized epidemiologic links but most notably identified previously unrecognized intervillage transmission. This study demonstrates significant ongoing transmission in a geographically isolated, low-density population. In a resource-rich country such as Canada, these communities illustrate some of the persistent challenges of TB control and elimination.
Key Words: tuberculosis • epidemiology • molecular epidemiology • cluster analysis • IS6110 restriction fragment length polymorphism The resurgence of tuberculosis (TB) in several industrialized countries during the late 1980s has been well documented, and efforts to stem the tide in these countries have been largely rewarded with declining rates (1, 2). The use of molecular typing, in conjunction with epidemiologic investigations, has provided evidence for ongoing transmission within indigenous populations and suggested avenues for improved TB control efforts (3–5). With targeted interventions, TB rates have declined in these subgroups, setting the stage to once again contemplate TB elimination (6, 7). Canada has among the world's lowest incidence of TB; the nationwide TB incidence has steadily decreased since the mid-1940s and is currently at 5.5 cases per 100,000 (8). Similar to many other industrialized Western countries, cases of TB in Canada are now primarily restricted to a few high-risk groups (9–11). Foreign-born individuals account for an ever-increasing proportion of TB cases in many provinces of Canada (12). Among the Canadian-born, the Aboriginal population, and in particular the Inuit, continue to have an incidence of TB 20 to 50 times higher (mean incidence, 77 cases per 100,000) than the non-Aboriginal Canadian-born population (13). After rampant TB epidemics in the Inuit communities during the first half of the 20th century, dramatic declines in TB rates coincided with the implementation of aggressive TB control efforts and the introduction of chemotherapy in the 1950s and 1960s (14–21). However, the incidence rate has leveled since the mid-1980s, and in 2000, the Inuit population still has the highest incidence rate of any group in Canada, including foreign-born sub-populations (13). The reasons for this persistently elevated rate remain largely speculative. This well-defined community provides an opportunity to closely examine the ability of traditional TB control measures to detect and interrupt ongoing transmission in resource-rich countries. To better define the epidemiology of TB in the Inuit communities of Northern Quebec, we have studied culture-positive cases from the 14 villages comprising the region of Nunavik during the years 1990 to 2000. Traditional epidemiologic analysis from public health data and molecular genotyping of the bacterial isolates were integrated to provide a more refined assessment of TB transmission.
Study Setting The Inuit population of Canada is primarily found in Nunavut/Northwest Territories, Labrador, and approximately one fifth of the population is in Nunavik (Quebec). Nunavik is Quebec's arctic region, over 500,000 km2 in land area (approximately three-quarter the size of Texas), bordered by the Hudson Bay (on the West), the Hudson Strait and the Ungava Bay (on the North), and the Labrador border (on the East). The Inuit communities (7,765 inhabitants in the 1996 Canada census) constitute nearly 90% of the population of Nunavik and are located in 14 coastal villages.
Epidemiologic Data
Genotyping
Cluster Analysis of Genotyping Data
Statistical Analysis
Descriptive Epidemiology During the years 1990 to 2000, there were 51 culture-positive cases of M. tuberculosis in the Inuit population of Nunavik, and isolates were available for 48 cases. Two cases were subsequently excluded from the study because the subjects were not living in Nunavik at the time of diagnosis, leaving 46 cases in the study. The incidence rate of reported cases of TB in Nunavik over the 11-year study period is shown in Figure 1 , contrasted with the incidence of TB in the rest of the province of Quebec and the Inuit in the rest of Canada. Even when combining the foreign-born with the non-Nunavik Canadian-born, the incidence in the rest of Quebec showed a steady decline with an all-time low incidence of 3.7 per 100,000 in 2000. The year-to-year variability in the incidence rate of the Inuit (Nunavik and Canada) was much greater due to the small population count. We can nonetheless observe that the incidence of the Inuit in both Nunavik and the rest of Canada was approximately 10-fold that of the rest of Quebec, with the suggestion of a decline at the end of the decade in Nunavik.
The demographic characteristics of culture-positive cases are presented in Table 1 . The age distribution of culture-positive cases is shown in Figure 2 and suggested a bimodal distribution; 48% of cases were under 25 years, and 24% were 55 years or more. The majority of the cases (38/46 = 83%) occurred in four villages that comprise only 38% of the region's population (total of 3,320 inhabitants). None of the subjects were known to be HIV seropositive. All isolates were drug susceptible.
Genotyping and Clustering Analysis Analysis of these isolates by IS6110 fingerprints alone showed highly conserved patterns and suggested that 42 out of 46 isolates likely were derived from the same original epidemic clone (see Figure E1 in the online supplement). The lack of background diversity in IS6110 patterns suggested a potentially reduced utility of IS6110 RFLP alone in tracking ongoing transmission. Because of this, we performed spoligotyping and mycobacterial interspersed repetitive units–variable number tandem repeats on these isolates to provide a more conservative estimate of clustering. By requiring an identical pattern across all 3 genotyping modalities within a 5-year interval, 76% (35/46) of cases fell into 6 clusters ranging from 2 to 14 in size (median cluster size 4). When using a 2-year interval, the proportion clustered was not significantly different (70%, 32/46 still in 6 clusters). Assuming one reactivation case in each of these 6 clusters, 29 of the 46 clustered cases (63%) likely represented ongoing transmission.
Integrated Analysis of Epidemiology and Molecular Epidemiology To determine the epidemiologic profile of individuals who shared genotypically identical isolates, we compared these individuals with patients who had unique M. tuberculosis genotypes (Table 1). TB cases with clustered isolates were younger than those with unique genotypes (median age 23 compared with 51, p = 0.03) and were more likely to have sputum smear positive for acid-fast bacilli (57% compared with 18%, p = 0.02). These characteristics support the interpretation that genotypically clustered cases were due to recent transmission. Of greatest interest to TB control is the examination of the largest cluster of 14 cases, spanning 4 villages and nearly 9 years (Table 2) . Contact investigations had detected an outbreak of four cases in a school in 1995, but failed to link it to subsequent cases in neighboring villages, during the following years. The young age of the subjects in the cluster clearly supported ongoing transmission.
In a low-incidence country such as Canada, it has become increasingly evident that TB represents a problem within a few groups. Aboriginals, including the Inuit, are among the highest risk groups in both urban and rural regions. The northern regions across Canada have been identified by previous studies as having persistently high incidences of TB, with rates 10 to 20 times the national rate (9, 30). A resurgence of TB has also been observed in remote Native Alaskan villages since 1994 (31). This reality reflects the difficulty in further reducing the incidence of TB after the epidemics have subsided. Our study attempts to better analyze the persistent transmission of TB within the Inuit community in the setting of a fully implemented traditional TB control program. Unlike urban hard-to-reach groups identified in previous molecular epidemiologic studies, such as the homeless or injection drug users (32, 33), the Inuit community is well defined and amenable to intervention. Therefore, compared with marginalized populations where multiple difficulties are encountered, from identifying social networks to enforcing compliance to treatment, these challenges should be minimal in this community. The results of our study likely highlight not the challenges in the implementation of known standard TB control principles but the limitations inherent to the structure of TB control. Molecular epidemiology has been used in many settings to better understand the transmission dynamics of TB within a population, in the hope of identifying high-risk groups that may benefit from targeted public health interventions (6). Clustering studies have inferred that matching fingerprints represent recent transmission in the face of a genetically diverse background of M. tuberculosis strains (3). Most TB molecular epidemiology studies have been performed in diversified populations, such as urban centers and countries with large numbers of immigrants (3, 32, 34). The empiric demonstration of background genetic variability among epidemiologically unrelated M. tuberculosis strains permits one to draw epidemiologic inferences about patients with shared M. tuberculosis genotypes. In contrast, we have observed that in a stable population such as the non-Aboriginal Canadian-born in Quebec, the presence of highly related IS6110 RFLP can instead represent a historic endemic strain without evidence of ongoing transmission (35). These isolated Aboriginal communities may pose comparable challenges to the interpretation of IS6110 RFLP–based clustering studies. Therefore, a high degree of IS6110-based clustering may well indicate ongoing transmission, but alternatively, the occurrence of shared strains today may merely reflect reactivation of past epidemic strains (10, 36). To address this potential limitation, we have studied isolates using multiple genotyping methods and tested the validity of our inferences by comparing our clustering results with epidemiologic data. Despite the appearance of a predominant strain type by IS6110 RFLP alone, the combined genotyping was able to distinguish six different clusters and four "unique" genotypes. Two results suggest that clusters defined by this method are compatible with recent transmission of TB. First, the median age of members of these clusters was 23, where a younger age is likely to be associated with a more recent acquisition of the infecting organism. Second, of the 17 suspected epidemiologic links, 14 were confirmed by perfectly matching genotypes across all 3 modalities. It has been recognized that subtle changes in genotypes can occur during documented chains of transmission (37, 38), and our multiple genotyping approach may hence underestimate the degree of ongoing transmission. However, the enhanced resolution afforded by multiple genotypes improves our confidence that isolates matching across the three markers truly indicate ongoing TB transmission. Unfortunately our small study size does not allow a meaningful comparison of the predictive value of IS6110 RFLP alone with the multiple genotyping method. The demonstration of clusters that span several villages is of particular concern. Although the region of Nunavik is extremely vast and sparsely populated (the median distance to the closest village is 126 km with no road connections), individuals regularly travel from one village to another for social or family gatherings. The segregation of the two coasts by genotypic clusters appears to reflect these movements; travel between distant villages is common, but the two coasts share relatively fewer social networks. In contrast to the known movements of the people living in Nunavik, contact tracing is performed at the nursing stations of individual villages. This notion is confirmed by the observation that just 2 of the 17 epidemiologically suspected transmission links involved individuals in different villages. As a result, when a series of TB cases occur within a village, household transmission and outbreaks may be effectively recognized or averted by conventional contact tracing. Conversely, transmission chains across villages are more likely to go unnoticed between villages when contact tracing is performed at the village level. This highlights the limitations of TB control in the "watershed" areas, that is, between public health jurisdictions or units. Furthermore, the social networks through which TB transmission occurs often expand well beyond the reaches of a public health unit, and a thorough understanding of the people's behavior is key to identifying settings in which transmission occurs. There has been and continues to be a tremendous effort to control TB in this community. All villages were subject to widespread tuberculin skin test surveys and isoniozid therapy for latent TB infection as recently as the early 1980s (39). The Bacillus Calmette–Guérin continues to be systematically administered to all newborns in Nunavik with the goal of preventing infantile forms of invasive TB (23). Despite these efforts, TB rates remain markedly elevated compared with the rest of the province, and these results confirm that there is still considerable ongoing transmission. This study highlights some of the challenges faced in the TB elimination phase, after the success of epidemic control (14, 40). As noted in the recent report from the Institute of Medicine, as cases become sparser in time, resources are often scaled back, and the control of TB becomes more arduous (41). Not only do we need to continue aggressive TB control interventions such as contact tracing and latent TB infection treatment in high-risk groups, but the organization of TB control needs to be optimized. The results of this study support the regionalization of TB services (41), whereby multiple public health units are brought together. Furthermore, integrating "real-time" genotyping of M. tuberculosis isolates at the time of contact investigations may allow for better interjurisdictional TB control.
The authors thank Dr. Dick Menzies for his critical review of the manuscript and Michael Purdy for his technical assistance in genotyping.
Supported by the Sequella Global Tuberculosis Foundation. D.N. is a recipient of a FRSQ Bourse de Formation and M.B. is a recipient of a New Investigator Award from the CIHR. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: D.N. has no declared conflict of interest; J-F.P. has no declared conflict of interest; J.W. has no declared conflict of interest; L.T. has no declared conflict of interest; S.D. has no declared conflict of interest; M.A.B. has no declared conflict of interest. Received in original form July 7, 2003; accepted in final form September 10, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
