Clinical implication of metabolic syndrome in nonobese patients with antineutrophil cytoplasmic antibody-associated vasculitis

Objective: We investigated the prevalence of metabolic syndrome (MetS) in all or nonobese patients with antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) and compared it with age- and gender-matched controls. Also, we assessed the effect of variables at diagnosis on the risk of cardiovascular disease (CVD) in all or nonobese AAV patients. Methods: In this study, 173 AAV patients and 344 controls were included and MetS was dened by the National Cholesterol Education Program Adults Treatment Panel III criteria. The obesity based on body mass index (BMI) was dened as BMI ≥ 25 kg/m 2 . The follow-up duration was dened as the period from diagnosis to the last visit or to each poor outcome occurrence. Results: The median age of AAV patients was 58.7 years and 57 patients were men. The prevalence of MetS was 50.9% in all AAV patients and 46.5% in nonobese AAV patients, which were signicantly higher than 37.8% in all controls and 28.2% in nonobese controls. In the Kaplan-Meier survival analysis, Mets at diagnosis signicantly reduced the cumulative CVD-free survival rate in both all and nonobese AAV patients. In the multivariable Cox hazards model analysis, CVD during follow-up was signicantly associated with both BVAS (HR 1.159) and MetS at diagnosis (HR 9.036) in nonobese AAV patients. Conclusions: The prevalence of MetS at diagnosis in all or nonobese AAV patients was signicantly higher than those in all or nonobese controls. Furthermore, both BVAS and MetS at diagnosis increased the risk of CVD in nonobese AAV patients.

through many studies and is now acknowledged that the disturbance in the immune-metabolic interaction may result in abnormal metabolic states, culminating in metabolic diseases such as MetS [8].
So far, there have been several studies investigating the association of MetS with autoimmune rheumatic diseases and the increased prevalence of MetS in patients with autoimmune rheumatic diseases such as systemic lupus erythematosus, rheumatoid arthritis and vasculitis have been reported. Especially in patients with systemic vasculitis, the prevalence of MetS was high, and consequentially to CVD as well [9].
Anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is a group of systemic necrotizing vasculitis which primarily affects the small-sized vessels and occasionally the medium-sized ones and consists of three subtypes such as microscopic polyangiitis (MPA), granulomatosis with polyangiitis (GPA) and eosinophilic granulomatosis with polyangiitis (EGPA) [10,11]. Since chronic lowgrade in ammation of AAV may provoke insulin resistance and disturb metabolic homeostasis, leading to MetS, it can be theoretically assumed that the prevalence of Mets can be gradually increased in AAV patients. Based on this assumption, a previous study conducted in the United Kingdom (UK), which compared the prevalence of Mets between AAV patients and controls, reported a higher proportion of MetS in AAV patients compared to controls [12]. MetS may have an in uence on the poor outcomes of AAV: the high risk of CVD has been reported in AAV, which, in turn, has led to the revision of the European League Against Rheumatism (EULAR) guideline for periodic CVD risk evaluation in AAV patients [13][14][15].
In addition, the effect of MetS on the risk of CVD might be clearer in nonobese people [16]. However, to our knowledge, there was no study su ciently investigating whether Mets at diagnosis could increase the risk of CVD during follow-up in AAV patients. Hence, in this study, we investigated and compared the prevalence of Mets at diagnosis between AAV patients and age-and gender-matched controls, and furthermore between nonobese AAV patients and controls who had body mass index (BMI) < 25 kg/m 2 [17]. Also, we investigated whether MetS at diagnosis could increase the risk of CVD and other poor outcomes of AAV during follow-up in both all AAV patients and nonobese AAV patients.

Study subjects
We included 173 patients with AAV, who were reclassi ed as AAV based on the 2007 European Medicines Agency algorithm for AAV and polyarteritis nodosa and the 2012 revised International Chapel Hill Consensus Conference Nomenclature of Vasculitides and we reviewed their medical records. All patients were initially diagnosed as AAV at the Division of Rheumatology, the Department of Internal Medicine, Yonsei University College of Medicine, Severance Hospital, from October 2000 to March 2019. They all had well-documented medical records with which clinical and laboratory data that reviewed ANCA positivity and both Birmingham vasculitis activity score (BVAS) version 3 and ve-factor score (FFS) were calculated at diagnosis [18,19]. AAV patients, who had serious medical conditions mimicking AAV or enabling ANCA false-positivity, such as chronic liver diseases, coexisting malignancies, serious infections, and drugs at the time of diagnosis, were excluded from this study. For controls, the medical records of general people, who had consecutively visited Severance Executive Healthcare Clinic in Severance Hospital, a university-a liated tertiary care hospital, for a comprehensive medical health check-up, were also reviewed [20]. And, 344 age-and gender-matched people without any serious medical condition were included in this study as controls. The cross-sectional diseases at diagnosis or the poor outcomes of AAV during follow-up were identi ed by the 10th revised International Classi cation Diseases-10 and drugs being administered were con rmed using the Korean Drug Utilization Review system. This study was approved by the Institutional Review Board (IRB) of Severance Hospital (4-2017-0673 for AAV patients and 4-2018-0856 for controls), who waived the need for the written informed consent, as this was a retrospective study.

Variables at diagnosis and during follow-up
In terms of the variables at diagnosis, age, male gender and BMI were obtained. In the general Korean population, obesity based on BMI was de ned as BMI ≥ 25 kg/m 2 [17], and thus, in this study, nonobese patients and controls were de ned as those having BMI < 25 kg/m 2 . AAV subtypes, ANCA positivity and both BVAS and FFS were reviewed. We reviewed the results for ANCA by both an indirect immuno uorescence assay (perinuclear (P)-ANCA and cytoplasmic (C)-ANCA) and antigen-speci c assays for ANCA (myeloperoxidase (MPO)-ANCA and proteinase 3 (PR3)-ANCA). In patients who tested positive in the indirect uorescence assay, but negative in antigen-speci c assays, P-ANCA positivity was considered as MPO-ANCA positivity and C-ANCA positivity as PR3-ANCA positivity [22,23]. Comorbidities at diagnosis, such as chronic kidney disease, diabetes mellitus, hypertension, dyslipidaemia and interstitial lung disease, were collected. The results of routine laboratory tests at diagnosis were also evaluated. In terms of the variables during follow-up, the poor outcomes of AAV were de ned as all-cause mortality, relapse, end-stage renal disease (ESRD), cerebrovascular accident (CVA) and CVD. The followup duration was de ned as the period between the date of the diagnosis of AAV and the date of the last visit for survived patients. For deceased patients, the follow-up duration based on all-cause mortality was de ned as the period between the initial diagnosis of AAV and the time of death. For patients who had any poor outcomes, the follow-up duration based on each poor outcome was de ned as the period starting from the diagnosis of AAV until each poor outcome appeared. Also, administered medications were assessed.

Statistical analyses
All statistical analyses were conducted using the SPSS software (version 23 for Windows; IBM Corp., Armonk, NY, USA). Continuous variables were expressed as a median (interquartile range), and categorical variables were expressed as number and the percentage. Signi cant differences in categorical variables between the two groups were analysed using the Chi-square and Fisher's exact tests. Signi cant differences in continuous variables between the two groups were compared using the Mann-Whitney test.
The odds ratio (OR) was assessed using the multivariable logistic regression analysis of variables with pvalues less than 0.05 in the comparative analysis. The optimal cut-off for BMI in predicting MetS at diagnosis was extrapolated by calculating the receiver operator characteristic (ROC) curve and selecting the maximised sum of the sensitivity and speci city. The relative risk (RR) was analysed using contingency tables and the chi-square test. Each cumulative poor outcome-free survival rate was analysed using the Kaplan-Meier survival analysis. The multivariable Cox hazard model using variables with p-values less than 0.05 in the univariable Cox hazard model was conducted to appropriately obtain the hazard ratios (HRs) during the considerable follow-up duration. P-values less than 0.05 were considered as statistically signi cant.

Characteristics of AAV patients
In regard to variable at diagnosis, the median age of AAV patients was 58.7 years and 57 patients were men. Ninety-seven patients were classi ed as MPA, 42 patients were classi ed as GPA and 34 patients were classi ed as EGPA. MPO-ANCA (or P-ANCA) was detected in 115 patients and ANCA was negative in 35 patients. The most common comorbidity was hypertension (46.2%), followed by chronic kidney disease (stage 3-5) (29.5%) and dyslipidaemia (28.3%). In regard to variables during the follow-up period, 14 patients died of any cause. The most frequently occurred poor outcome of AAV was relapse (32.4%), followed by ESRD (19.1%). Twelve of 173 patients (6.9%) had experienced CVD after the diagnosis of AAV. One hundred sixty-two patients had received glucocorticoid (93.6%) during follow-up. The most common immunosuppressive drug administered was cyclophosphamide (49.7%), followed by azathioprine (48.0%) ( Table 1).

Comparison of MetS-related variables between 173 AAV patients and 344 controls
There were no signi cant differences in the demographic data, in particular, the number of ex-smokers between AAV patients and controls; however, AAV patients exhibited the lower median BMI than controls (22.2 vs. 23.4 kg/m 2 ). Inversely, the number of AAV patients, who ful lled the 2005 NCEP-ATP-III criteria, was signi cantly higher than that of controls who satis ed the same criteria (50.9% vs. 37.8%). Among the ve components of the 2005 NCEP-ATP-III criteria, AAV patients exhibited central obesity (53.2% vs. 71.2%) less frequently than controls. Meanwhile, they showed hypertriglyceridemia (49.7% vs. 23.5%) and impaired glucose tolerance (60.1% vs. 24.4%) more often than controls. In regard to MetS-related laboratory results, AAV patients exhibited the higher median harmful cholesterols than controls ( Table 2).

Comparison of MetS-related variables between 144 nonobese AAV patients and 255 nonobese controls
To minimise the effect of obesity based on BMI on the prevalence of MetS, we compared the prevalence of MetS in nonobese AAV patients and controls who had BMI < 25 kg/m 2 . Sixty-seven of 144 AAV patients (46.5%) exhibited the cross-sectional MetS, whereas only 72 of 255 controls (28.2%) had it (RR 2.212, P < 0.001). Whereas, among obese AAV patients and controls who had BMI ≥ 25 kg/m 2 , the two groups showed the similar prevalence of MetS (Fig. 1).
Comparison of variables at diagnosis between 88 AAV patients with Mets and 85 AAV patients without Mets AAV patients were divided into the two groups based on MetS at diagnosis: 88 patients were assigned to the groups of AAV patients with MetS and they showed higher frequency in all ve components of the 2005 NCEP-ATP-III criteria than those without MetS. At the time of diagnosis, AAV patients with MetS were older and more obese than those without MetS: however, no difference in gender and ex-smoker between the two groups was found. In regard to ANCA positivity and AAV-speci c in ammatory indices, AAV patients with Mets showed a signi cantly higher frequency of MPO-ANCA (or P-ANCA) positivity and the higher cross-sectional BVAS and FFS than those without MetS. Among comorbidities, the proportions of chronic kidney disease, diabetes mellitus, hypertension and dyslipidaemia were signi cantly increased in AAV patients with MetS compared to those without MetS. Among laboratory results, AAV patients with MetS exhibited higher levels in all the variables than those without MetS except for haemoglobin and serum albumin which showed an opposite tendency (Table 3).

Multivariable logistic regression analysis of variables at diagnosis for the cross-sectional MetS in AAV patients
We categorised variables at diagnosis with statistical signi cance in the comparison analysis into the two groups: the conventional risk factors for MetS and AAV-speci c in ammatory indices as shown in Additional le 1: Table S1. To determine the independent predictor of the cross-sectional MetS at diagnosis, we conducted the multivariable logistic regression analysis and found that BMI (OR 1.481), diabetes mellitus (OR 7.629), hypertension (OR 16.054) and dyslipidaemia (OR 8.800) were signi cantly associated with the cross-sectional MetS. Whereas, none of the AAV-speci c in ammatory indices was associated with the cross-sectional MetS.
Comparison of the effect of MetS at diagnosis on the risk of the outcomes of AAV during follow-up We simply compared the frequencies of the poor outcomes of AAV between the two groups and found that ESRD and CVD occurred in AAV patients with MetS more frequently than those without MetS during follow-up (Table 3). We also compared the cumulative risk of each poor-outcome during the follow-up period based on each poor outcome occurrence between the two groups using the Kaplan-Meier survival analysis. AAV patients with MetS exhibited the lower cumulative CVD-free survival rate than those without MetS during the follow-up period based on CVD occurrence (Fig. 2). Furthermore, we assessed the effect of MetS at diagnosis on the risk of CVD by dividing AAV patients into two categories based on BMI of 25 kg/m 2 . Among nonobese AAV patients, MetS at diagnosis signi cantly reduced the cumulative CVD-free survival rate. However, among obese AAV patients, there was no signi cant difference in the cumulative CVD-free survival rate between AAV patients with and without MetS (Fig. 2).
Hazard ratio of variables at diagnosis for the risk of CVD during follow-up in 173 AAV patients Firstly, in cases of all AAV patients, in the univariable Cox hazards model analysis, male gender, BVAS, FFS, diabetes mellitus, dyslipidaemia, fasting glucose level and MetS at diagnosis were signi cantly associated with CVD occurrence during follow-up. In the multivariable Cox hazards model analysis of variables with signi cance in the univariable analysis, only male gender (HR 6.006, 95% con dence interval (CI) 1.486, 24.283) was signi cantly associated with CVD occurrence during follow-up. With the same assumption above, among male gender, BVAS, FFS and MetS at diagnosis, both male gender (HR 4.625, 95% CI 1.258, 16.996) and MetS at diagnosis (HR 9.864, 95% CI 1.136, 85.679) were signi cantly associated with CVD during follow-up (Additional le 2: Table S2).
Hazard ratio of variables at diagnosis for the risk of CVD during follow-up in 144 nonobese AAV patients In cases of nonobese AAV patients, in the univariable Cox hazards model analysis, BVAS, dyslipidaemia, haemoglobin, fasting glucose and MetS at diagnosis were signi cantly associated with CVD during follow-up. In the multivariable analysis, only BVAS at diagnosis (HR 1.157, 95% CI 1.038, 1.289) was signi cantly associated with CVD during follow-up. Given that diabetes, dyslipidaemia and fasting glucose are closely related to components of the 2005 NCEP-ATP-III criteria, they could be deleted in the multivariable analysis in order to clarify the effect of variables at diagnosis on the risk of CVD. Thus, only BVAS, haemoglobin, and MetS at diagnosis were included in the multivariable analysis, in which, both BVAS (HR 1.159, 95% CI 1.039, 1.293) and MetS at diagnosis (HR 9.036, 95% CI 1.011, 80.786) were signi cantly associated with CVD during follow-up (Table 4).

Discussion
In this study, we assessed the effect of variables at diagnosis on the risk of CVD during follow-up in AAV patients and found several interesting ndings. Firstly, the prevalence of MetS based on the 2005 NCEP-ATP-III criteria was 50.9% in all AAV patients, which was signi cantly higher than 37.8% in age-and gender-matched controls. The 2013 annual report regarding the prevalence of MetS in approximately 10 million Korean individuals with an average age of 50.8 years and BMI of 23.9 kg/m 2 analysed the overall prevalence of MetS as 30.5% [24]. The next version of the report, Metabolic Syndrome Fact Sheet in Korea 2018, reported the increased prevalence of Mets of Korean people of an average age of 50s up to 37.9% [25], which supports that controls in this study were representative of the general Korean population of an average age of 50s. Moreover, the prevalence of MetS in AAV patients in Korea was slightly higher than that in the UK [12]. Despite insu cient studies investigating the prevalence of MetS in AAV patients, this discordance might be considered to appear due to the different ethnic or geographical backgrounds [26].
Secondly, the prevalence of MetS was signi cantly higher in nonobese AAV patients than that in nonobese controls (46.5% vs. 28.2%). This result may suggest the contribution of the in ammatory burden of AAV to the presence of MetS in AAV patients beyond obesity and its related complications. Interestingly, in the UK study, no difference in BMI between AAV patients and controls was observed [12]. Whereas, in our study, BMI of AAV patients was signi cantly lower than that of controls, which exhibited an opposite tendency of the prevalence of MetS. Although BMI is not one of the components of the 2005 NCEP-ATP-III criteria, BMI is another independent index for determining obesity and considered one of the risks for MetS [6,27]. This inverse tendency suggests that another unique risk factor exists in AAV patients other than the conventional risk factors for MetS in normal people and it was assumed as the in ammatory burden of AAV. To prove this assumption, we tried to compare the cross-sectional BVAS or FFS between the two studies but unfortunately, we could not due to no information on BVAS in the UK study.
Thirdly, unlike the comparison analysis between AAV patients and controls, BMI was strongly associated with the cross-sectional MetS as shown in the comparison analysis between AAV patients with MetS and those without MetS. Based on this result, it might be assumed that the general association between obesity and MetS became apparent when compared only in AAV patients, resulting from minimizing the in uence of the in ammatory burden of AAV. However, the burden of in ammation was not thoroughly removed, because BVAS was assessed signi cantly elevated in AAV patients with MetS, compared to those without MetS. Thus, this result may suggest the cooperative contribution of AAV activity to the presence of MetS in AAV patients along with obesity and its related complications.
Supposed that variables directly related to the 2005 NCEP-ATP-III criteria were excluded, two categories of the risk factors at diagnosis for the cross-section MetS could be organized: one is the conventional risk factors such as age, BMI, diabetes mellitus, hypertension and dyslipidaemia; and the other is the AAVspeci c in ammatory variables such as BVAS, FFS, haemoglobin, platelet count, creatinine, serum albumin, ESR and CRP at diagnosis. Using these variables, we conducted the multivariable logistic regression analysis and found that BMI, diabetes mellitus, hypertension and dyslipidaemia were independently and signi cantly associated with the cross-sectional MetS at diagnosis. By contrast, none of the AAV-speci c in ammatory variables were signi cantly associated with the cross-sectional MetS. This analysis gave two conclusions: one is that BVAS itself might not be independently associated with the cross-sectional MetS in AAV patients, and the other is that BMI might independently contribute to the cross-sectional MetS in AAV patients and AAV activity might consolidate the association between BMI and MetS at diagnosis.
Since BMI was an independent variable that could predict the cross-sectional MetS in AAV patients, we calculated the optimal cut-off of BMI at diagnosis for the cross-sectional Mets using the ROC curve analysis. We determined the BMI of 22.9 kg/m 2 as the cut-off for a strong predictor of the cross-sectional MetS (area 0.686, 95% CI 0.606, 0.766, P < 0.001, sensitivity 62.5%, speci city 75.3%) (Additional le 3: Fig. S1A). When we classi ed AAV patients into the two groups based on the cut-off of BMI and assessed its relative risk for the occurrence of the cross-sectional MetS using the chi-square test, 76 AAV patients were partitioned into the group of BMI ≥ 22.9 kg/m 2 . The cross-sectional MetS was identi ed more frequently in AAV patients with BMI ≥ 22.9 kg/m 2 than those without (72.4% vs. 34.0%, P < 0.001).
Prior to the investigation, it should be noted that except for relapse of AAV, Mets could increase the risks of all-cause mortality [28], chronic kidney disease or ESRD [29], CVA [30] and CVD [28, 31, 32] in both AAV patients and the general population with MetS. In addition, AAV itself without MetS also could increase the risk for CVD compared to healthy people [14,33]. Therefore, it should not be ignored that both AAV entity and the cross-sectional MetS at diagnosis may be simultaneously engaged in CVD occurrence in AAV patients: MetS might signi cantly initiate CVD occurrence and AAV might accelerate it. On the other hand, unlike, the UK study [12], we could not nd any link between MetS at diagnosis and relapse during follow-up in this study.
Fourthly, Mets at diagnosis signi cantly reduced the cumulative CVD-free survival rate and both BVAS and Mets at diagnosis signi cantly associated with CVD in nonobese AAV patients. In the survival analysis, MetS at diagnosis signi cantly reduced the cumulative CVD-free survival rate only in nonobese AAV patients. Whereas, obese AAV patients showed no association between MetS at diagnosis and CVD occurrence during follow-up. This result may suggest that the independent contribution of MetS to the development of CVD would have been offset because MetS is closely related to obesity itself and obesityrelated complications in obese AAV patients. For this reason, the effect of MetS on the risk of CVD might be clearer in nonobese AAV patients. In addition, to discover the independent predictors of and contributors to CVD occurrence during follow-up in nonobese AAV patients, we conducted the multivariable Cox hazards model analysis using variables with P value less than 0.05 in the univariable analysis. In the multivariable analysis excluding variables related to the 2005 NCEP-ATP-III criteria, BVAS and MetS at diagnosis had in uence on CVD in nonobese AAV patients. This result might support our assumption that both AAV entity and metabolic abnormalities could enhance the risk of CVD in nonobese AAV patients.
Our study has several limitations. Controls, who visited Severance Executive Healthcare Clinic in Severance Hospital, and the two-thirds of AAV patients, who belong to the prospective cohort of AAV in our hospital, had information on smoking history, alcohol consumption, and family history of MetS and CVD. However, we could not gather them from all AAV patients and controls due to the nature of a retrospective study. In addition, the number of AAV patients of this study, particularly patients with CVD occurrence, was not large enough to represent all Korean patients with AAV, due to a limitation of a monocentric study. Nevertheless, we believe that this study has signi cant clinical implications as a pilot study in that we clari ed the effect of BVAS and MetS at diagnosis on the risk of CVD in all or nonobese AAV patients, for the rst time. In the near future, a prospective and multicentre study with a larger number of AAV patients will compensate for the limitations of this study.

Conclusions
The prevalence of MetS at diagnosis was 50.9% in all AAV patients and 46.5% in nonobese AAV patients, both of which were signi cantly higher than those in all controls and nonobese controls. Furthermore, both BVAS and MetS at diagnosis increased the risk of CVD in nonobese AAV patients.

Declarations
This study was conformed to the provision of the Declaration of Helsinki and approved by the Institutional Review Board of Severance Hospital (4-2020-1071), which waived the requirement for patient written informed consent due to the retrospective study design.