Association of Cigarette Smoking With Antineutrophil Cytoplasmic Antibody–Associated Vasculitis

Key Points

Question  Is cigarette smoking associated with an increased odds of having antineutrophil cytoplasmic antibody–associated vasculitis (AAV)?

Findings  In this case-control study of 473 patients with AAV compared with 1419 matched controls, cigarette smoking was associated with an increased odds of having AAV. This association was especially strong in patients with antimyeloperoxidase antibodies, who are increasingly recognized to have differences in genetic risk, pathogenesis, and response to treatment compared with those with proteinase 3 antibodies.

Meaning  The association between AAV and cigarette smoking identifies a modifiable risk factor for AAV and may suggest a possible pathogenic mechanism between respiratory exposures and development of AAV.

Importance  Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is a systemic small vessel vasculitis characterized by circulating ANCAs targeting proteinase 3 (PR3) or myeloperoxidase (MPO) and associated with excess morbidity and mortality. Myeloperoxidase-ANCA–positive AAV and PR3-ANCA–positive AAV are increasingly recognized to have differences in genetic risk, pathogenesis, and response to treatment. Risk factors for AAV, including cigarette smoking, are poorly understood.

Objective  To examine the association between cigarette smoking and AAV.

Design, Setting, and Participants  This case-control study included a consecutive inception cohort of 484 patients with AAV diagnosed from 2002 to 2017 compared with a cohort of sex-, race-, and age-matched controls. Eleven cases were excluded owing to discordant smoking information in the electronic health record. Controls were randomly selected from participants recruited to the Partners HealthCare Biobank between its inception in 2010 and 2018 and who completed a smoking questionnaire and were not diagnosed with AAV (n = 30 536).

Exposures  Smoking status (current, former, never) and pack-years of cigarette smoking were determined from review of the electronic medical record and smoking questionnaires.

Main Outcomes and Measures  Patients with AAV were individually matched with 3 randomly-selected controls based on sex, race, and age (within 2 years difference). Conditional logistic regression was performed to examine the association between cigarette smoking and AAV using odds ratios (OR) and 95% confidence intervals (CIs).

Results  Overall, 473 cases were matched with 1419 controls (mean [SD] age, 59 [16] years; 281 women [59%], 396 white [84%]). Patients with AAV were more likely to be former (OR, 1.6; 95% CI, 1.3-2.0) or current smokers (OR, 2.7; 95% CI, 1.8-4.1); there was a dose-response relationship according to pack-years of exposure (P < .001). These associations were especially strong among participants with MPO-ANCA–positive disease (former smokers: OR, 1.7; 95% CI, 1.3-2.3; current smokers: OR, 3.5; 95% CI, 2.1-6.1) but not in participants with PR3-ANCA–positive AAV (former smokers: OR, 1.3; 95% CI, 0.9-2.0; current smokers: OR, 1.7; 95% CI, 0.8-3.5). After stratifying by selected demographics and disease manifestations, these associations remained strong.

Conclusions and Relevance  Cigarette smoking was associated with AAV, especially MPO-ANCA–positive AAV. Further studies are needed to investigate a potential pathogenic mechanism.

Antineutrophil cytoplasmic antibody–associated vasculitis (AAV) is a systemic small vessel vasculitis of unclear etiology associated with excess morbidity and mortality compared with the general population.¹ In most cases, AAV is characterized by the presence of circulating antineutrophil cytoplasmic antibodies (ANCA) that target proteinase 3 (PR3) or myeloperoxidase (MPO) and are considered pathogenic.^2,3 Proteinase 3-ANCA– and MPO-ANCA–positive AAV are increasingly recognized as distinct conditions characterized by differences in genetic risk, pathogenesis, disease manifestations, and response to treatment.^4,5

The contribution of both genetic factors and inhaled environmental exposures, including silica and asbestos, to the development of AAV has been previously reported.^6-11 Although little is known about the association of cigarette smoking with the risk of AAV, several observational studies have suggested that smoking may influence AAV presentation.^12,13 Smokers have been reported to have increased risk of AAV disease relapse¹⁴ and a lower prevalence of ear, nose, and throat damage¹⁵ compared with nonsmokers.

Studies of the association between cigarette smoking and the risk of AAV, however, have generated conflicting results.^7,9,16-19 Some found no association,^9,16 1 documented a nonstatistically significant trend suggesting an association,¹⁹ and others have reported a potentially protective effect of smoking.^17,18 However, these studies have been limited by small sample sizes, inconsistent use of reference groups, and a focus on patients with granulomatosis with polyangiitis (GPA) who were typically PR3-ANCA–positive. To date, there has been no large case-control study investigating the association between cigarette smoking and AAV or the differences in these associations according to ANCA type, to our knowledge.

We performed a large case-control study to examine the association of cigarette smoking, a potentially modifiable factor, with the odds of having AAV.

Cases were identified from the Partners inception AAV cohort, which included participants with AAV who were evaluated and treated at Partners HealthCare, a large health care system including tertiary care and community hospitals, as well as primary care and specialty outpatient clinics in the greater Boston area.²⁰ Consecutive cases diagnosed between January 1, 2002, and December 31, 2017, were identified using a validated approach.²⁰ The diagnosis of AAV was systematically confirmed by manual medical chart review using a previously described consensus algorithm.²¹ All patients were PR3-ANCA or MPO-ANCA positive. Eosinophilic granulomatosis with polyangiitis (EGPA) cases were excluded. The study was approved by the Partners HealthCare institutional review board. Informed consent was waived because of the retrospective nature of the study.

Controls were obtained from the Partners HealthCare Biobank, a large sample and clinical data collection program that recruits patients from inpatient and outpatient sites throughout the Partners HealthCare system.²² The Biobank specifically recruits patients from a diversity of settings to achieve a representative sample of patients who receive their care at Partners HealthCare system. Patients who had completed a smoking questionnaire and entered the Partners Biobank between January 1, 2010, and May 4, 2018, were considered eligible (N = 30 536). Each case was matched to 3 randomly selected controls based on year of birth (within 2 years apart), sex, and race.

For participants with AAV, smoking status was determined based on a review of the electronic medical record. We defined never smoking as less than 100 cigarettes smoked in a patient’s lifetime. Former or current smoking status was determined by smoking status at the time of treatment initiation (index date). For those with current or past cigarette smoking, a cumulative smoking exposure in pack-years was determined by calculating the product of years smoked and average packs of cigarettes smoked per day. For controls, smoking status and pack-years of exposure at the same age as the matched case was determined using responses to a smoking questionnaire administered on enrollment in the Partners Biobank.

For both cases and controls, the date of the first encounter in the Partners HealthCare system was determined. Self-reported race and education were extracted from the electronic medical record. For cases, PR3- and MPO-ANCA status was extracted from the electronic medical record and baseline clinical manifestations, including head and neck, renal, and pulmonary disease, and the Birmingham Vasculitis Activity Score/Wegener Granulomatosis (BVAS/WG) score were determined via manual medical chart review.²³ The date of treatment initiation for AAV was determined by medical chart review.²⁰

Continuous variables are reported as mean (SD). Categorical variables are reported as number (%). The association between cigarette smoking and AAV was examined using conditional logistic regression. Odds ratios (ORs) and 95% confidence intervals (CIs) for current and former smokers were calculated using never smokers as the reference group.

Cases and controls were assigned to mutually exclusive groups based on cumulative lifetime smoking exposure: never smokers, 0 to 20 pack years, 21 to 40 pack years, 41 to 60 pack years, and more than 60 pack years of smoking. The association between cigarette exposure and AAV was examined for each pack-year group using conditional logistic regression and treating the pack-year group as an ordinal variable where the never smoker group was the reference group.

We performed several sensitivity analyses. To assess for confounding by education level, cases were matched to 3 randomly selected controls matched for year of birth (within 2 years apart), sex, and education level; in this sensitivity analysis, 1 case was matched more liberally owing to young age at time of AAV diagnosis. To control for potential changing smoking trends over time, we stratified our primary analysis according to date of treatment initiation (before or after 2010) and also repeated the pack-year analysis described herein with the same date stratification. In addition, we performed an analysis restricted to cases and their respective controls that had established care (>5 outpatient visits) in the Partners HealthCare System at least 1 year prior to their AAV diagnosis.

A 2-tailed P value of <.05 was used as the threshold for statistical significance and SAS statistical software (version 9.4; SAS Institute, Inc) was used for all statistical analyses.

Baseline characteristics of participants with AAV and controls are included in Table 1. Of 484 participants with AAV, 1 did not have smoking status documented and 10 participants with AAV had discordant smoking statuses recorded in several notes and/or structured data fields and were excluded from the study. A total of 473 AAV cases were matched to 1419 controls from the Partners Biobank.

The mean (SD) age for cases and controls was 59 (16) years; 281 cases (59%) and 843 controls (59%) were women and 84% were white (281/743 cases and 1188/1419 controls). Most participants with AAV were MPO-ANCA positive (309 [65%]). Renal involvement was present in 304 patients (64%) at baseline, whereas 194 cases (41%) had pulmonary and 216 (46%) had head and neck involvement at baseline. The mean (SD) baseline BVAS/WG score was 4.7 (2.1).

There was a greater proportion of ever smokers (current or former) among participants with AAV (25/475 [54%]) than controls (596/1419 [42%]) and smoking was associated with an increased odds of having AAV (OR, 1.7; 95% CI, 1.4-2.2). The average pack-years of cigarette exposure was 13.3 (21.1) for AAV cases and 3.5 (8.8) for controls (P < .001). When stratified by smoking status, the odds of AAV were higher for both former smokers (OR, 1.6; 95% CI, 1.3-2.0) and current smokers (OR, 2.7; 95% CI, 1.8-4.1) compared with never smokers (Table 2). We observed a strong dose-response relationship such that the odds of having AAV rose with increasing pack-years of exposure (P for trend <.001) (Figure). Patients with the greatest cumulative pack-year exposure (≥60 pack-years) to cigarette smoking had the greatest odds of having AAV (OR, 30.3; 95% CI, 8.7-105.6) compared with never smokers.

These associations remained strong among male patients (former smokers: OR, 2.1; 95% CI, 1.5-3.1; current smokers: OR, 3.6; 95% CI, 1.7-7.6) after stratification by sex. Among female patients, current smoking was associated with AAV (OR, 2.3; 95% CI, 1.4-3.9) and a similar trend was observed for female former smokers (OR, 1.3; 95% CI, 0.95-1.8).

We performed several analyses after stratifying the AAV cases according to disease-specific features. When stratified by ANCA type, smoking was strongly associated with AAV among patients who were MPO-ANCA positive (former smokers: OR, 1.7; 95% CI, 1.3-2.3; current smokers: OR, 3.5; 95% CI, 2.1-6.1). A similar trend was observed among PR3-ANCA–positive patients but did not reach significance (former smokers: OR, 1.3; 95% CI, 0.9-2.0; current smokers: OR, 1.7; 95% CI, 0.8-3.5). Associations between being a former or current smoker and the odds of having AAV were observed across subgroups of patients with selected AAV manifestations. Among the cases with baseline renal involvement, former and current smokers had an increased odds of having AAV (OR, 1.8; 95% CI,1.3-2.4 and OR, 2.3; 95% CI, 1.3-4.1, respectively). Similarly, among those with baseline head and neck disease, former and current smokers had an increased odds of having AAV (OR, 1.6; 95% CI, 1.1-2.3 and OR, 2.0; 95% CI, 1.01-3.7, respectively). Among those with baseline pulmonary manifestations, former and current smokers had an increased odds of having AAV (OR, 2.2; 95% CI, 1.5-3.2 and OR, 2.1; 95% CI, 1.1-4.2, respectively).

In sensitivity analyses (Table 3), the observed associations between cigarette smoking and the odds of having AAV persisted after accounting for education status, whether or not a patient had established outpatient care in the Partners HealthCare System at least 1 year prior to the diagnosis of AAV, and when limiting our analysis to cases diagnosed after January 1, 2010, the date that the Biobank began enrolling participants (eTable 1 and eFigures 1A and 1B in the Supplement).

In this large case-control study of patients with AAV and matched controls identified from the same health care system, a history of cigarette smoking was strongly associated with an increased odds of having AAV. Moreover, we observed a dose-response association such that greater cumulative smoking exposure was associated with a greater odds of having AAV. This association was especially strong among MPO-ANCA–positive patients and persisted across subgroups of patients with certain AAV manifestations, including renal, head/neck, and pulmonary disease. Our findings remained robust in multiple sensitivity analyses.

To our knowledge, this is the first large US-based case-control study evaluating the association between cigarette smoking and AAV among patients with diverse AAV manifestations and in a cohort enriched for MPO-ANCA–positive disease. It is also the first study to stratify cases by cumulative smoking exposure (eg, pack-years) and demonstrate a dose-response relationship between smoking and the odds of having AAV. Our findings differ from those of some prior studies. In their case-control study of patients with biopsy-proven ANCA-associated glomerulonephritis, Hogan and colleagues¹⁹ found an increased ever-smoking rate (62%) in their 129 cases compared with matched controls (51%). However, this difference did not meet the threshold for statistical significance, likely because of the sample size. Notably, the difference in rates of ever-smoking between cases and controls in this study was similar to the difference observed by Hogan, et al¹⁹ (12% vs 11%).

Lane and colleagues⁹ found no significant difference in smoking rates between the vasculitis and nonvasculitis control groups in their study of 75 patients with small vessel vasculitis. In addition to the larger AAV sample size in this study, differences in the proportion of MPO-ANCA–positive patients between the study by Lane et al⁹ and this study may account for some of the observed differences. Whereas a minority of patients in the studies by Hogan et al¹⁹ and Lane et al⁹ had MPO-ANCA–positive disease (47% and 25% respectively), MPO-ANCA–positive patients accounted for 65% of the patients in this study (309/473) and the association between smoking and AAV was especially strong in this subgroup. Two additional European studies^17,18 noted lower smoking rates among patients with AAV compared with overall rates in the general population, but it is difficult to interpret these observations given the lack of matched control groups. Benarous et al¹⁸ reported that only 16% of their cohort were ever smokers. This smoking rate was substantially lower than reported in other cohorts from the US (71%),⁹ Germany (41%),¹⁷ and New Zealand (57%),²⁴ as well as the 54% ever smoking rate observed in this study. It remains unclear why the cohort in the study by Benarous et al¹⁸ differed from those reported elsewhere and how their observed rates would compare with those observed in a matched control group. This study corroborates observations by Yamaguchi et al,¹⁴ who noted an increased relapse rate among smokers in their AAV cohort, which was mostly (97%) MPO-ANCA–positive patients.

An association between cigarette smoking and other autoimmune diseases, including rheumatoid arthritis, systemic lupus erythematosus, and giant cell arteritis has also been observed.^25-29 The mechanism by which cigarette smoking may contribute to AAV pathogenesis is unclear, but the observed association between smoking and the odds of having AAV in the context of other known environmental risk factors like asbestos and silica suggests that, in a subset of patients, ANCA autoantibody formation may be influenced by respiratory exposures.⁹ A similar phenomenon has been well demonstrated in anticitrullinated peptide antibody–positive patients with rheumatoid arthritis.³⁰ The possibility of a pathogenic link between AAV and respiratory exposures such as cigarette smoking may also contribute to differences in disease presentation between MPO-ANCA– and PR3-ANCA–positive patients. Future studies will be needed to understand whether smoking is associated with certain disease manifestations in AAV, such as fibrotic lung disease previously associated with MPO-ANCA–positive AAV.^31,32 Alternatively, cigarette smoking may increase the odds of having AAV through its effects on oxidative stress, depletion of nitric oxide, and endothelial damage.^33,34 Future studies are needed to examine the association between smoking status and AAV disease progression.

In addition to cigarette smoking, other respiratory exposures have been identified as potential risk factors for autoimmune conditions, including AAV. In patients with AAV, the most well-described of these is silica exposure.⁹ We did not have information regarding occupational history in cases and controls to evaluate the potential association between other respiratory exposures and the odds of having AAV in this study. However, after matching according to education status, which may be a surrogate for occupation, our observations remained robust. Additional studies are necessary to clarify whether the association between cigarette smoking and the odds of having AAV is modified by exposure to silica.³⁵

Strengths of our study include the large sample size and the high proportion of patients who were MPO-ANCA positive. Furthermore, we included patients with AAV with diverse organ manifestations of disease, including patients with renal and nonrenal manifestations. Most prior studies of environmental risk factors for AAV have included a predominance of PR3-ANCA–positive and/or GPA patients who had renal disease.^9,11,12,19 Moreover, we identified cases and controls from within the same health care system; it is important to note that patients were enrolled into the Partners Biobank from across the spectrum of care, including outpatient primary care and specialty clinics as well as the inpatient setting. Finally, in addition to detecting associations between being a former or current smoker and the odds of having AAV, we also found a dose-response relationship such that a greater cumulative exposure to smoking was associated with a higher odds of having AAV.

This study also has certain limitations. This is an observational study that has found associations rather than establish causal effects. Nevertheless, our observed associations between cigarette smoking and the odds of having AAV were robust and similar to other smoking-associated rheumatic conditions.^25-29 The ascertainment of cases and controls from a single health care system and the fact that most participants in the study were white may limit the generalizability of these findings. However, Partners HealthCare System includes several large, tertiary care hospitals, as well as community hospitals, primary care clinics, and specialty clinics around New England. Our approach to assessing smoking status in the case and control groups may also be a limitation because we used smoking status in the medical record for patients with AAV and self-reported survey for controls. Nevertheless, smoking status in both groups was ascertained in the setting of receiving care at Partners HealthCare. A cohort study design would allow for prospective uniform assessment of smoking status but AAV is too rare for such a study design.

We observed a strong association between current and former cigarette smoking and the odds of having AAV. These findings expand the list of potential risk factors for AAV, including genetics and silica exposure.^6,7 This association was especially strong for patients with MPO-ANCA. Further studies to confirm these results and investigate a potential pathogenic mechanism are needed.

Corresponding Author: Zachary S. Wallace, MD, MSc, Clinical Epidemiology Program, Division of Rheumatology, Allergy, and Immunology, Massachusetts General Hospital, 100 Cambridge St, 16th floor, Boston, MA 02114 (zswallace@mgh.harvard.edu).

Accepted for Publication: February 14, 2020.

Published Online: April 13, 2020. doi:10.1001/jamainternmed.2020.0675

Author Contributions: Dr Wallace had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: McDermott, Zhang, Choi, Wallace.

Acquisition, analysis, or interpretation of data: McDermott, Fu, Stone, Wallwork, Zhang, Choi, Wallace.

Drafting of the manuscript: McDermott, Choi, Wallace.

Critical revision of the manuscript for important intellectual content: McDermott, Fu, Stone, Wallwork, Zhang, Choi, Wallace.

Statistical analysis: Fu, Zhang, Choi, Wallace.

Obtained funding: Choi, Wallace.

Administrative, technical, or material support: Zhang, Choi.

Study supervision: Stone, Choi.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was funded by the National Institutes of Health/ National Institute of Arthritis and Musculoskeletal and Skin Diseases [K23AR073334 and L30 AR070520 to ZW].

Role of the Funder/Sponsor: The National Institutes of Health/ National Institute of Arthritis and Musculoskeletal and Skin Diseases had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.