Risk factors for drug-resistant tuberculosis at a referral centre in Toronto, Ontario, Canada: 2010–2016

Background: Drug-resistant tuberculosis (TB) poses a major public health concern worldwide. However, no studies have addressed risk factors for drug resistance in Ontario, which has its own unique profile of immigrants. We evaluated demographic and clinical risk factors for drug-resistant TB among patients treated at West Park Healthcare Centre, located in Toronto, Ontario (Canada). Methods: All patients who were diagnosed with TB and treated at West Park Healthcare Centre between January 2010 and December 2016 were included in this retrospective cohort study. Characteristics of patients with isoniazid mono-resistant (INH-R) TB and multidrug resistant (MDR) TB were compared to patients with drug-susceptible TB with bivariate and multivariable logistic regression. Results: Risk factors for INH-R TB included younger age (younger than 35 years), prior TB treatment, non-diabetic and birth in a non-South-East Asian country, but only the latter two factors were significant in multivariable analysis. On the other hand, we found younger generation (younger than 65 years), birth in European region, recent arrival to Canada (fewer than 120 months), prior treatment and human immunodeficiency virus (HIV) infection were associated with MDR-TB, among which younger age (younger than 35 years), more recent immigration (fewer than 24 months), prior treatment and HIV infection were significant in multivariable analysis. Conclusion: These findings may be of use to TB clinicians in the province by informing the initial empiric antibiotic regimen prescribed while awaiting phenotypic drug susceptibility testing and assisting in decisions regarding whether to request rapid molecular drug susceptibility testing. Affiliations 1 Division of Respirology, Department of Medicine, University of Toronto, Toronto, ON 2 Department of Respiratory Medicine, West Park Healthcare Centre, Toronto, ON 3 Division of Respirology, Department of Medicine, Toronto Western Hospital, Toronto, ON 4 Department of Thoracic Surgery, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Miyagi, Japan *Correspondence: thirama@icloud.com Suggested citation: Hirama T, Sabur NF, Derkach P, McNamee J, Song H, Marras TK, Brode SK. Risk factors for drug-resistant tuberculosis at a referral centre in Toronto, Ontario, Canada: 2010–2016. Can Commun Dis Rep 2020;46(4):84–92. https://doi.org/10.14745/ccdr.v46i04a05


Introduction
Drug-resistant tuberculosis (TB) poses a major public health concern worldwide. The two most common, clinically important forms of drug-resistant TB include isoniazid (INH) mono-resistant (INH-R) (resistant to INH) and multidrug-resistant (MDR) TB (resistant to at least INH and rifampin, RMP) (1,2). Drug resistance is identified either by genotypic methods or phenotypic culture-based drug susceptibility testing (DST), the latter being considered the gold standard (3). Identification of drug resistance is critical to guide appropriate selection of anti-mycobacterial drugs and to prevent further drug resistance.
However, phenotypic DST can take weeks to report, and not all clinical settings perform rapid molecular DST routinely. Therefore, clinicians often start empiric TB treatment prior to the availability of phenotypic DST results, and may expand the initial empiric regimen, or may request rapid molecular DST, based upon an individual patient's risk factors for drug resistance.
Few studies have described risk factors for drug resistance in Canada. In British Columbia, age, foreign-born status, ethnicity, prior treatment, diagnosis outside of Canada and certain birth country regions were associated with drug resistance from 1990-2001 (4). In Alberta from 1982-2011, age (younger than 65 years), prior treatment, arrival to Canada from 2002-2011, and recent emigration from Philippines and Vietnam were risk factors for MDR-TB in foreign-born persons (5). A national surveillance study found that age, foreign-born status, prior treatment, and certain World Health Organization epidemiological regions-of-birth were associated with drug resistance on a national level from 1997 to 2008 (1). However, no studies have addressed risk factors for drug resistance in Ontario, which has its own unique profile of immigrants (6) and has the highest burden of drug-resistant TB cases in Canada (7,8). There is also a need for more contemporary data, because risk factors may differ as immigration patterns change and as rates of drug-resistant TB, including primary MDR-TB, change world-wide.
The principal objective of this study was to evaluate possible demographic and clinical risk factors for drug-resistant TB among patients treated at West Park Healthcare Centre (WPHC) and to compare risk factors for INH-R TB against risk factors for MDR-TB. Additionally, the enrolled TB patients were reviewed according to the recent American Thoracic Society/Infectious Diseases Society of America/Centers for Disease Control and Prevention (ATS/IDSA/CDC) statement (3), recommending rapid molecular testing for RMP +/-INH resistance be performed in the following patient sub-groups: 1) previous treatment; 2) born or lived for one or more years in a country with TB incidence of greater or equal to 20/100,000 or primary MDR prevalence of greater or equal to 2%; 3) contact with MDR; and 4) human immunodeficiency virus (HIV) infection.

Methods
The TB program at WPHC, located in Toronto, Ontario, is recognized as a referral centre for drug-resistant TB, and sees the majority of MDR-TB cases in the province (84% between 2000 and 2011) (9). All patients who were diagnosed with TB and treated at WPHC between January 2010 and December 2016 were included in this retrospective cohort study. Chart review was used to identify patients for inclusion and to extract demographic and clinical characteristics. The study protocol was approved by the Joint Bridgepoint/West Park Healthcare Research Ethics Board. In light of the retrospective design, the requirement of informed consent was waived.
Throughout the study period, all drug susceptibility testing was consistently performed at the Public Health Ontario TB and Mycobacteria Laboratory (Toronto, Ontario). The DST was performed according to Clinical Laboratory Standards Institute testing standards recommended methods (as available), using radiometric broth [BACTEC 460; Becton, Dickinson and Co., Franklin Lakes, New Jersey, United States (US)] until October 1, 2010, and nonradiometric broth (MGIT 960; Becton, Dickinson and Co.) thereafter (10,11). The first culture of Mycobacterium tuberculosis complex isolated from a patient was routinely tested for susceptibility to the four first-line drugs: INH; RMP; ethambutol; and pyrazinamide. Isolates resistant to INH at 0.1 mg/L were considered "resistant" herein, but were also tested at 0.4 mg/L and underwent moxifloxacin testing. Any isolate found resistant to RMP or any two of the first line drugs underwent DST to second line drugs. Second line susceptibility testing for the following drugs was performed during the study time period: rifabutin; amikacin; streptomycin; kanamycin; capreomycin; ofloxacin; ethionamide; and p-aminosalicylic acid. The DST for clofazimine was performed until October 1, 2010 and DST for moxifloxacin and linezolid started on October 1, 2010.
Characteristics of patients with INH-R TB and MDR-TB were compared with those patients with drug-susceptible (DS) TB (i.e. susceptible to the four first line drugs) using bivariate and multivariable logistic regression models. Statistical analyses were performed using GraphPad Prism 6.0 (GraphPad Software; La Jolla, California, US), StatPlus:macLE (AnalystSoft; Walnut, California, US) and Jamovi (Version 0.9, retrieved from https://www.jamovi.org). In bivariate analyses, many demographic characteristics (age, sex, birth country region and time from arrival in Canada) and clinical characteristics (known TB risk factors, location of TB and microbiologic results) were analysed for their possible association with drug-resistant TB, to be thorough and exploratory. Variables with a p-value less than 0.05 in bivariate analysis, and variables considered a priori to be clinically important (age, sex, birth country region, time from arrival in Canada and history of TB treatment) were selected for inclusion in the multivariable models. The multivariable models were restricted to foreign-born patients so that the association between time in Canada and drug resistance would be accurately studied. Patients were also divided into slightly different groups by DST (drug susceptible, non-MDR drug resistance and MDR/RMP resistance) to evaluate the recent recommendations for rapid molecular DST for rifampin put forth by the ATS/IDSA/CDC (3).

Results
Between 2010 and 2016, 485 patients with active TB were seen at WPHC, representing 11.1% of the total of 4,384 seen in Ontario (12). Among these WPHC patients, DST results were available in 82.9% (n=402/485) ( Table 1). The other 83 patients (17.1%) did not have a phenotypic DST performed in Ontario (due to lack of culture confirmation or to a diagnosis made outside of Ontario), and were excluded from further risk factor analyses. The TB strains susceptible to the four first-line drugs accounted for 76.1% (n=306/402), strains INH-R accounted for 10.9% (n=44/402) and strains resistant to both INH and RMP +/other drugs (MDR) accounted for 11.4% (n=46/402). Only four patients had mono-resistance to drugs other than INH (one to RMP and three to pyrazinamide), and two had poly-resistance to the first line drugs (but not MDR); six patients were excluded from risk factor analyses. Extensively drug resistant TB (MDR with additional resistance to a fluoroquinolone and a second line injectable) was also rare at 1.0% (n=4/402).
The TB patients were divided into three groups based on DST: DS-TB (n=306); INH-R (n=44); and MDR (n=46) and their demographic characteristics are shown in Table 2 and clinical characteristics in Table 3. Compared with patients with DS-TB (Table 2), in unadjusted analyses, patients with INH-R TB were significantly younger; odds ratio (OR) for age younger than 35 years=2.58, 95% CI 1.06-6.30, with reference age older than 65 years, and less likely to have been born in South-East Asia (OR 0.157, 95% CI 0.03-0.91). Patients with INH-R TB (Table 3) were also more likely to have been previously treated (OR 2.39, 95% CI 1.01-5.68), and less likely to have diabetes (OR 0.26, 95% CI 0.08-0.87) in unadjusted analyses. Compared with DS-TB patients, in unadjusted analyses, patients with MDR-TB (Table 2) were also significantly younger (younger than 35 years, OR 15.2, 95% CI 3.49-66.1) and more likely to have been born in Europe (OR 15.6, 95% CI 1.66-146.4) and had a significantly shorter time from arrival to Canada to TB diagnosis. These patients (Table 3) were more likely to have been previously treated (OR 5.74, 95% CI 2.77-11.9), more likely to have HIV infection (OR 4.76, 95% CI 1.29-17.5) and more likely to have only pulmonary TB and less likely to have pulmonary and extrapulmonary TB.
In our evaluation of the 2017 ATS/IDSA/CDC recommendations for rapid molecular DST for rifampin, we found that patients with MDR/RMP resistance were significantly more likely to have had previous TB treatment (OR 5.39, 95% CI 2.57-11.3) and HIV infection (OR 4.26, 95% CI 1.06-17.0) than DS-TB patients in multivariable analysis ( Table 5).

Discussion
We identified several risk factors for drug resistance among TB patients seen at our institution in Toronto, Ontario. Regarding INH-R TB, we found that young age (younger than 35 years), prior TB treatment, lack of diabetes and birth in a non-South-East Asian country were risk factors in bivariate (unadjusted) analysis, but only the latter two were significant in multivariable analysis. Prior TB treatment has been previously reported as a risk factor for INH-R TB, even after adjustment for possible confounders (13). Somewhat surprisingly, we found a significant association between diabetes and INH-R TB in both bivariate and multivariable analysis, with the former appearing "protective" against the latter. Most previous studies have not included diabetes in their assessment of risk factors for drug resistance; and a study from British Columbia did not find an association between the two variables (4). Additionally, some reports, including a recent meta-analysis, have described a positive association between diabetes and MDR-TB (14). Given that no prior studies reported a negative association between INH-R TB and diabetes, and the lack of a plausible biologic explanation for this finding, we suspect that this association might be spurious. While we did control for age in our multivariable model, there could be residual confounding by age, as INH-R TB was more common in younger patients, who generally have a lower prevalence of diabetes. This association could also have been found by chance, and may be related to multiple testing; future study on this association is needed. Interestingly, in our population, foreign birth (OR 0.93 95% CI 0.31-2.80) was not associated with INH-R TB. Although other North American studies have found foreign birth to be a risk factor for INH-R or mono-resistant TB (1,4,13), only one of these studies adjusted for potential confounders and no association was found (13).
We found several risk factors for MDR-TB in our population that have been described previously in North America, including younger generation, prior treatment (5), more recent arrival to Canada (1) and HIV infection. HIV infection is controversial; a meta-analysis found that most North American studies reported an association. There was no significant association for MDR-TB overall when studies from all world regions were included; yet there was an association with primary MDR-TB (15). In the bivariate analysis, we also found that patients with only pulmonary TB were more likely to have MDR-TB, but patients with pulmonary and extra-pulmonary were less likely. It is possible that the distribution of TB was confounded by the time when they were diagnosed, as patients with MDR-TB were more likely to have recently arrived to Canada and, therefore, may have had less advanced disease. In fact, the overall fraction of pulmonary involvement (pulmonary plus pulmonary and extra-pulmonary) was similar among DST categories (DS 82.0% (n=251/306), INH-R 88.6% (n=39/44) and MDR 84.8% (n=39/46)).
Regarding regions of birth, there was no significant association in multivariable analysis, but birth in Europe (OR 15.6, 95% CI 1.66-146.4) was a risk factor for MDR-TB in bivariate analysis, and the lack of significance in multivariable analysis in our study could have been due to the small numbers of cases analysed.
Given the growing number of immigrants in Canada (5.5 million in 2000 to 7.9 million in 2017) (16,17) and the worldwide epidemic of DR-TB, the prevalence of DR-TB in Canada has the potential to increase (8). One of the many challenges posed by DR-TB in low burden countries is the delay between TB diagnosis and culture-based DST, which can prolong the time to appropriate treatment initiation, increase morbidity and prolong infectiousness. However, in such regions, universal rapid molecular testing may not be cost-effective, and may lead to high numbers of false positives (18). Therefore, targeted testing, based on risk factors, is often used. The most recent ATS/IDSA/ CDC guidelines for TB diagnosis (3) suggest that rapid molecular testing for RMP +/-INH resistance be performed in the following patient sub-groups: 1) previously treated; 2) born or lived for at least one year in a country with TB incidence of greater or equal to 20/100,000 or primary MDR prevalence of at least 2%; 3) contact with MDR; and 4) HIV infection. Our results support the application of these guidelines in Ontario regarding patient sub-groups (patient subgroups 1 and 4).
While we did not find a significant association between RMP-resistance and a history of contact with MDR-TB (patient sub-group 3), the OR was high and our numbers were small, and it seems logical that these patients may be at risk and should be tested. However, our data raises questions about the potential benefits and costs of "targeted" testing for patients in sub-group 2 in a geographical region such as ours (Toronto, Ontario) where the majority of TB patients are immigrants. Perhaps in Ontario and similar regions, this criterion could be modified such that only patients from a higher risk country, who are also of younger age and/or have recently immigrated, would be tested. Targeted testing for patients from very high-risk countries (i.e. European Region) may also be considered.

Strengths and limitations
Given WPHC's status as a referral center for complicated and drug-resistant TB cases in Ontario, it is not surprising that our proportions of drug-resistant cases (10.9% INH-R and 11.4% MDR) were higher than provincial (8.5%/1.4%) (7) and national (6.2%/1.2%) (18) rates in 2016. Our higher than average population of drug-resistant cases presented an opportunity to study patient characteristics in detail; however, the number of drug resistant cases in our study was still relatively low. Furthermore, we may not have had the power to detect a significant association between some true risk factors and drug resistant TB. Additionally, there could be selection bias in our study population, since DS-TB cases with less severe TB disease or with fewer comorbidities might have been less likely to be referred to our specialized center. Another potential limitation to our study is that it is representative of patients in the Toronto region (which sees the majority of TB in the province; 76% in 2016) (19), and may not represent the characteristics of patients from other Ontario cities, who may be less likely to be referred to our institution. Additionally, we did not have detailed information regarding all countries where an individual resided in before coming to Canada. Finally, we tested many patient characteristics for their association with drug resistant TB, and we may have found associations that were spurious due to multiple testing.

Conclusion
We summarize risk factors for INH-R and MDR-TB among patients seen at our institution in Toronto, Ontario. These findings may be of use to TB clinicians throughout the province by informing the initial empiric antibiotic regimen they prescribe while awaiting phenotypic DST, and by assisting them in their decision regarding whether to request rapid molecular DST. These findings may also guide policy makers and laboratory personnel regarding targeted application of molecular DST in the province.