Abstract
Objective. Pulmonary involvement is common in patients with systemic sclerosis (SSc), and this condition causes substantial morbidity and mortality. Disrupted immunity from the disease or associated medication may render such patients subject to tuberculosis (TB) infection. However, the relationship between SSc and TB has not yet been investigated.
Methods. Using the Taiwan National Health Insurance Research Database, 838 patients with SSc diagnosed in Taiwan during 2000–2006 were identified and followed for emergence of TB infection. Incidence rate ratios (IRR) of TB compared to 8380 randomly selected age-, sex-, and comorbidity-matched controls without SSc were calculated. The Cox proportional hazards model was used for multivariate adjustment to identify independent risk factors for TB infection.
Results. The risk of TB infection was higher in the SSc cohort than in controls (IRR 2.81, 95% CI 1.36–5.37; p = 0.004), particularly for pulmonary TB (IRR 2.53, 95% CI 1.08–5.30; p = 0.022). Other independent risk factors for TB infection in patients with SSc were age ≥ 60 years [hazard ratio (HR) 3.52, 95% CI 1.10–11.33; p = 0.035] and pulmonary hypertension (PH; HR 6.06, 95% CI 1.59–23.17; p = 0.008). Mortality did not differ for SSc patients with or without TB.
Conclusion. In this nationwide study, the incidence of TB infection was significantly higher among patients with SSc than in controls without SSc. Special care should be taken in managing patients with SSc who are at high risk for TB, especially those aged ≥ 60 years or who also have PH.
Systemic sclerosis (SSc) is an autoimmune disorder characterized by small vasculopathy, autoantibody production, and fibrosis of multiple organs (skin, gastrointestinal tract, lung, kidney, and heart)1. Two major pulmonary involvements associated with SSc, pulmonary hypertension (PH) and interstitial lung disease (ILD), can occur either together or independently. Each pulmonary compromise, when present, is currently the leading cause of morbidity and mortality in patients with SSc2,3. The crucial components underlying the pathogenesis of SSc include endothelial dysfunction and dysregulation of fibroblasts, which results in excessive collagen production, and profound abnormalities of the immune system4.
Tuberculosis (TB) remains a major public health problem and is responsible for an estimated 1.7 million global deaths per year5. Patients with SSc may be at risk of TB infection because of defective immunity from SSc or associated immunosuppressant therapy. TB might be acquired from an active infected person or from reactivation of a preexisting quiescent lesion. Pulmonary involvement is common in SSc and usually affects the architecture of both the lung tissue and pulmonary vessels6. Previous studies demonstrated that structurally damaging lung diseases, such as chronic ILD7, chronic obstructive pulmonary disease (COPD)8, bronchiectasis9, or pneumoconiosis10, can impair local host immunity and further increase susceptibility to TB infection. However, although a few case reports have been published11,12, no large-scale study has been conducted on the incidence, risk factors, and mortality rates among patients with SSc who develop TB, to our knowledge. Therefore, we performed a longitudinal nationwide cohort study using the Taiwan National Health Insurance Research Database (NHIRD) to investigate the relationship between SSc and incident TB infection.
MATERIALS AND METHODS
Data sources
The Taiwanese government began the National Health Insurance (NHI) program in 1995 to provide comprehensive healthcare for all citizens. Enrollment in this program is mandatory, and coverage of the population reached 98.29% in 200613. Comprehensive medical care including outpatient, inpatient, emergency, dental, traditional Chinese medicine services, and prescription drugs are provided by the NHI program. In 1999, the Bureau of National Health Insurance began releasing patient data (e.g., NHI enrollment files, claims data, and a prescription drug registry) in electronic form under the NHIRD project for research.
For catastrophic illnesses such as SSc, the government registered the confirmed subjects, after strict verification14, in the special category “Catastrophic Illness” in NHIRD. We obtained data from this dataset that included all claims information for patients with catastrophic illnesses. The application of a catastrophic illness certificate for SSc requires a comprehensive review of medical records, examination reports, and the results of imaging studies by physicians. After successful certification, patients with SSc can be exempted from related medical expenses. Our study was approved by the Institutional Review Board of Taipei Veterans General Hospital (201204022BC).
Study population
We used discharge codes (International Classification of Diseases, 9th revision, and clinical modification ICD-9-CM) in the Registry of Catastrophic Illness. We identified 937 patients who were diagnosed with SSc between January 1, 2000, and December 31, 2006. We defined newly diagnosed TB as a compatible ICD-9-CM code (010–018) plus a prescription for at least 2 anti-TB medications for more than 28 days15,16. We further validated our diagnosis and reclassified the patients as non-TB cases if their initial TB-related ICD-9 codes (010–018) in the NHI reimbursement database were changed to other ICD-9 codes during the followup period, including non-TB mycobacterial infection (ICD-9 code 031), lung cancer (ICD-9 code 162), or positive tuberculin skin test (ICD-9 code 795.5), with discontinuation of anti-TB medications17,18. To assess the reliability of our findings, we conducted a secondary analysis using a definition of TB diagnosis that required the prescription for at least 3 anti-TB medications.
Patients were excluded who were under 20 years of age, had antecedent TB, were followed for less than 30 days, and/or developed TB within 30 days of diagnosis. A total of 838 patients with SSc were included. We also collected information on comorbidities, including diabetes mellitus, hypertension, COPD, chronic kidney disease, cirrhosis, malignancies, and substance abuse. We also analyzed sociodemographic characteristics, such as age, sex, income (NT < $20,000, NT $20,000–39,999, and NT ≥ $40,000), and level of urbanization (level 1, level 2, and level 3) in our analysis19. Urbanization levels are divided according to the Taiwan National Health Research Institute publications. Level 1 designates the most urbanized areas and level 3 designates the least urbanized areas. Medications, including corticosteroids and immunosuppressants, were also retrieved and converted to defined daily dose (DDD), a unit recommended by the World Health Organization. DDD is the assumed average maintenance dose per day of a drug consumed for its main indication in adults and is commonly used for research comparing the use of immunosuppressant agents between international settings20,21. Using the concept of DDD, we could compare any immunosuppressant drugs based on the same standard: (total amount of drug)/(amount of drug in a DDD) = number of DDD. Cumulative DDD (cDDD), which indicates exposed duration, was calculated as the sum of dispensed DDD of any immunosuppressant agent. This calculation allowed us to compare use with the risk of subsequent TB. The cDDD of an immunosuppressant agent within 1 year before TB diagnosis was categorized into 2 groups (< 28 and ≥ 28 cDDD). Drugs used for < 28 cDDD were defined as non-used20.
Control cohort
Subjects without SSc were used as a matched cohort and were randomly selected from 1 million NHI beneficiaries out of a population of 21.4 million enrollees throughout Taiwan in 2000. Because SSc is a relatively rare disorder and the incidence of TB in patients with SSc is even lower, we matched each patient with SSc with 10 non-SSc subjects by age, sex, and presence of comorbidities on the same index date of diagnosis. The same exclusion criteria were applied to the control cohort as to patients with SSc. A total of 8380 people served as the matched control cohort.
Major outcome measured
The 2 cohorts, SSc cohort and matched cohort, were followed for the emergence of TB infection (ICD-9-CM code 010–018). If no TB occurred, subjects were followed to death, or the end of the study period (2006). Because TB is endemic in Taiwan, the Taiwan Centers for Disease Control (CDC) has declared TB as a mandatory notifiable disease22. By law, physicians diagnosing incident TB cases or prescribing anti-TB drugs must report cases to the Taiwan CDC within 1 week, allowing all newly-diagnosed TB cases to be identified.
Statistical analyses
Extraction and computation of data were done using the Perl programming language (version 5.12.2). Microsoft SQL Server 2005 (Microsoft Corporation) was used for data linkage, processing, and sampling. All statistical analyses were done using IBM SPSS statistical software (version 20.0, IBM-SPSS). Data are expressed as means ± SD or medians (interquartile ranges) when appropriate. A chi-square test or Fisher’s exact test was used for categorical variables, and the Mann-Whitney U test was used for parametric and nonparametric variables. Incidence rates (per 10,000 person-yrs) and incidence rate ratios (IRR) were analyzed. The Kaplan-Meier method was used for estimation of cumulative incidence. For multivariate adjustment, the Cox proportional hazards model was used to compute hazard ratios (HR) as well as 95% CI. A p < 0.05 was considered significant.
RESULTS
Patient characteristics
Figure 1 shows the study entry flow chart. A total of 838 patients were included after excluding patients who were younger than age 20 when diagnosed (n = 40), had antecedent TB (n = 44), were followed for fewer than 30 days (n = 11), or were diagnosed with TB within 30 days after SSc diagnosis (n = 4). The study population had a median age of 49 years (range 20–87 yrs), and was 76.3% women. The baseline characteristics of patients with and without SSc are in Table 1. Age, sex, and underlying comorbidities were matched between the study and control groups.
Patient selection flow chart.
Baseline characteristics of patients with systemic sclerosis (SSc) and the matched cohort.
Comparison of incidence rates of TB between SSc cohort and controls cohort
Kaplan-Meier analysis showed that the cumulative incidence of TB was significantly higher in the SSc cohort than in the control cohort (log-rank test; p = 0.001; Figure 2). As shown in Table 2, among the 9218 patients (838 SSc cases and 8380 matched controls) in our study during a 7-year period, 60 (0.7%) cases of newly diagnosed TB were identified, including 12 with SSc and 48 in the matched controls. The incidence rate of TB in the SSc cohort was higher than in the matched cohort (41.4 vs 14.7 per 10,000 person-yrs, p = 0.004). The SSc cohort had a 2.81-fold greater risk of TB infection than the matched cohort (IRR 2.81, 95% CI 1.36–5.37; p = 0.004). If TB cases were further classified as pulmonary versus extrapulmonary TB, the SSc cohort still had an increased risk of developing pulmonary TB (IRR 2.53, 95% CI 1.08–5.30; p = 0.022) compared to the matched cohort. Although not significant, the SSc cohort also had an increased risk of developing extrapulmonary TB (IRR 4.22, 95% CI 0.27–17.57; p = 0.064).
Cumulative incidence of tuberculosis (TB) in patients with systemic sclerosis and the matched cohort.
Incidence of pulmonary and extrapulmonary tuberculosis (TB) in patients with systemic sclerosis (SSc) and the matched cohort.
As shown in Table 3, we performed multivariable Cox proportional hazards model analysis and found that SSc itself was an independent risk factor for TB infection (HR 2.99, 95% CI 1.58–5.63; p = 0.001). Other independent risk factors included age ≥ 60 years (HR 2.79, 95% CI 1.66–4.71; p < 0.001) and having COPD (HR 2.19, 95% CI 1.30–3.68; p = 0.003). We further conducted a secondary analysis using a definition of TB diagnosis requiring prescriptions for at least 3 anti-TB medications, which did not significantly alter the results of the primary analysis (Appendix 1).
Cox regression of risk factors for tuberculosis (TB).
TB risk factors in patients with SSc
Of 838 patients with SSc, 521 (62.2%) received corticosteroid, 102 (12.2%) received cyclophosphamide (CYC), 10 (1.2%) received cyclosporine, 32 (3.8%) received methotrexate, and 46 (5.5%) received azathioprine. Univariate Cox regression analysis identified 4 risk factors for TB in patients with SSc: age ≥ 60, PH, use of corticosteroids, and use of CYC (Table 4). Multivariate Cox regression hazards analysis indicated that 2 variables were independent risk factors for TB in patients with SSc: age ≥ 60 (HR 3.52, 95% CI 1.10–11.33; p = 0.035) and having PH (HR 6.06, 95% CI 1.59–23.17; p = 0.008). Other factors, including use of corticosteroids or CYC, were not significant in the multivariable analysis. Of the patients with SSc during the study period, 4 of 12 patients (33.3%) with TB died, compared to 119 of 826 patients (14.4%) without TB (p = 0.066).
Cox regression of risk factors for tuberculosis (TB) in the systemic sclerosis (SSc) cohort.
DISCUSSION
This is the first study, to our knowledge, to investigate the association between SSc and TB using a nationwide population-based cohort. We compared the cumulative incidence of TB in patients with SSc to the matched cohort. This study had a long followup period and its findings provide several new insights: (1) patients with SSc had a significantly increased risk of developing TB (IRR 2.81) compared to the matched controls during the 7-year study period; (2) SSc itself might be an independent risk factor for TB development; (3) patients with SSc had 2.53 times greater risk of developing pulmonary TB than patients in the matched cohort; and (4) age ≥ 60 years and PH were risk factors for TB in the SSc cohort.
In 1987, Cowie23 reported a 40% incidence of prior TB in black gold miners with SSc. Despite noting a possible association between TB and SSc, the results of that earlier study might be confounded by silica exposure/silicosis. To date, previous studies that have explored the association between SSc and TB are a report of a Japanese-Canadian woman24, 2 Russian cases25,26, and an Indian case series of 13 patients with SSc who developed TB11. However, the findings of the Indian report were limited by a small sample size, a lack of a control group, and a high women-to-men ratio (23:1). Our study found a 2.81-fold higher incidence of TB in patients with SSc (41.4 per 10,000 person-yrs) than in matched controls (14.7 per 10,000 person-yrs). Moreover, our study demonstrated that SSc itself was an independent risk factor for TB infection after adjusting for other confounders.
SSc is an autoimmune disorder characterized by abnormalities of both the cellular and humoral immunity27. SSc patients have reduced numbers of circulating T lymphocytes, and reportedly decreased lymphocyte proliferation28,29. The subpopulation of T lymphocytes regulating cell-mediated immunity is also selectively reduced30. Impaired immunity could predispose patients with SSc to TB infection.
Our study found that compared to controls, patients with SSc were at a greater risk for developing pulmonary TB. The risk for extrapulmonary TB showed a trend toward significance (IRR = 4.22, p = 0.064). Up to 70% to 90% of patients with SSc had pulmonary involvement, which might predispose patients to pulmonary TB31,32. To our knowledge, this is the first study to demonstrate that SSc with pulmonary involvement, especially PH, was an independent risk factor for TB (HR 6.06). PH in SSc can result from fibroblast stimulation by a number of cytokines produced by activated macrophages including transforming growth factor (TGF)-β, tumor necrosis factor (TNF)-α, platelet-derived growth factor, and fibronectin. Fibroblast stimulation also leads to increased vascular endothelial growth factor and thus aberrant neoangiogenesis33. Increased levels of TNF-α and TGF-β are reported to suppress the protective immune response against TB34,35,36. However, work is still needed to elucidate the mechanism of the close relationship between SSc and TB. Further, patients with SSc and ILD had a 3.02 times greater risk for TB, although these results were not significant (p = 0.097). Previous studies found evidence of ILD in 60% to 91% of patients with SSc, as evaluated by high-resolution computed tomography. However, only 26% of the patients were evaluated by physical examination and only 22% by chest radiograph37,38. A high rate of underdiagnosis of ILD in patients with SSc could be the reason the results were not significant.
We observed a trend of higher extrapulmonary involvement in patients with SSc who developed TB than in the matched cohort (IRR 4.22). These data were not significant, possibly because of the limited number of TB subjects (p = 0.064). Our results are similar to previous findings on patients with other autoimmune disorders, systemic lupus erythematous39,40, rheumatoid arthritis treated with biological agents41,42, and inflammatory myositis, all of which can lead to TB with greater extrapulmonary spread43. This is in part due to intrinsic immune dysregulation, permitting the TB bacilli to activate in organs where it is less often successful. The application of systemic corticosteroids and immunosuppressant agents might increase the risk of TB reactivation and dissemination to extrapulmonary sites. A case series by Ahmad, et al11 found an increased incidence of TB among patients with SSc receiving dexamethasone therapy. Although we observed that use of corticosteroids was associated with an elevated HR for TB, the result was not statistically significant in the multivariable model. Perhaps use of corticosteroid is a relatively small TB risk factor in this cohort.
Our study has several limitations. First, the administrative data in the registry database did not contain information about smoking status, physical activity, body mass index, or laboratory and image data. Additionally, data regarding some clinical characteristics (such as the distinction of localized, limited, or diffuse SSc) were not recorded in the NHI database. Therefore, we could not explore the link between these clinical features and TB in patients with SSc. However, this common limitation has been noted in previous registry studies15,44. Besides, we did not use the standard procedure for TB diagnosis (acid-fast stain and culturing). Second, subjects who had antecedent TB were not included in our study. The analyses do not determine whether patients with SSc who developed TB had a recent new infection or reactivation of a remote latent infection. Third, physicians apply for a catastrophic illness certificate of SSc based on whether there is the presence of characteristic cutaneous finding of skin thickening, additional extracutaneous manifestations, capillaroscopic abnormalities, and SSc-related autoantibodies. However, the results of diagnostic tests for SSc were not available in our registry data. Finally, we were unable to evaluate the risk of TB development during childhood or adolescence in patients with SSc because we excluded patients < 20 years old. TB in childhood or adolescence is different from adult TB. Child/adolescent TB often comes from transmission from the community or school and is more likely to be complicated by extrapulmonary involvement45,46.
Patients with SSc had a higher risk for pulmonary TB. Hence, a high index of suspicion and more thorough investigation of suspected TB cases, especially in patients with SSc and PH, could minimize the delay in TB diagnosis and the potential harm to public health in a TB-endemic country.
APPENDIX 1.
Cox regression of risk factors for tuberculosis (TB).
Footnotes
-
Supported by a grant from Taipei Veterans General Hospital (V101D-001-2). The study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health and managed by National Health Research Institutes. The interpretation and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, or National Health Research Institutes.
- Accepted for publication April 7, 2014.
REFERENCES
In this issue
