Initially, the electronic medical records of 257 patients with AAV were screened. The medical records of 244 patients with AAV, in which BMI at diagnosis was clearly documented, were retrospectively reviewed. The inclusion criteria were as follows: i) patients who were diagnosed with AAV at this institute between October 2000 and January 2021; ii) patients who were reclassified as AAV based on the classification algorithm for AAV and polyarteritis nodosa proposed by the European Medicine Agency in 2007 (the 2007 algorithm) [2], and the revised nomenclature of vasculitides suggested by the Chapel Hill Conference Consensus in 2012 (the 2012 definitions) [1]; iii) patients who were followed up for more than 3 months; iv) patients whose medical records were well documented and include AAV-specific indices and clinical and laboratory data recorded at diagnosis and during follow-up; and v) patients who had BMI at diagnosis ≥18.5 kg/m². The exclusion criteria were as follows: i) patients who had serious medical conditions such as malignancy, severe infection, and autoimmune diseases other than AAV [12,13]; and ii) patients who had ever received immunosuppressive drugs for AAV treatment of prior to the diagnosis of AAV. Of 244 patients, to avoid the effect of underweight on all-cause mortality and to minimise the effects of confounding factors [9], 41 patients with BMI at diagnosis <18.5 kg/m² were excluded. Finally, 203 patients with BMI ≥18.5 kg/m² were included in this study and analysed. This study was approved by the Institutional Review Board of this institute, which waived the need for patients’ written informed consent, as this was a retrospective study (IRB protocol number 4-2020-1071).

Information regarding age, BMI and sex was collected as the demographic data. Further, information on AAV subtypes, ANCA types and positivity, and AAV-specific indices including Birmingham vasculitis activity score (BVAS, version 3), and five-factor score (FFS) was also obtained [14,15]. Patients in whom the ANCA results were negative by antigen-specific assay but were positive by immunofluorescence assay, were considered to have ANCA when AAV was strongly suspected based on the clinical and laboratory features [16]. Clinical manifestations based on BVAS, comorbidities, and acute-phase reactants were reviewed. Diabetes mellitus, hypertension, chronic kidney disease (stages 3, 4, and 5), hyperlipidaemia, and interstitial lung disease were assessed as comorbidities [17].

mBMI was calculated using the following equation: mBMI (kg · g/m² · L)=BMI (kg/m²)×serum albumin (g/L) [or 10×serum albumin (g/dL)] [11].

All-cause mortality, which was defined as death by any cause, was considered as a poor outcome in this study. The follow-up duration based on all-cause mortality was defined as the period from AAV diagnosis to death for deceased patients, and the period from AAV diagnosis to the last visit for surviving patients. Medicines that were administered during the follow-up period were recorded.

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 25. (IBM Corp., Armonk, NY, USA). Continuous variables are expressed as median with the interquartile range, whereas categorical variables are expressed as numbers (%). Significant differences between the two categorical variables were analysed using the Chi-square and Fisher’s exact tests. The correlation coefficient (r) between the two variables was obtained using either the Pearson correlation analysis or the univariable linear regression analysis. Also, multicollinearity was evaluated using the univariable linear regression analysis. The optimal cut-off was extrapolated by performing the receiver operator characteristic (ROC) curve analysis and one value having the maximised sum of sensitivity and specificity was selected. Comparison of the cumulative survivals rates between the two groups was analysed by the Kaplan–Meier survival analysis with the log-rank test. The multivariable Cox hazard model using variables with statistical significance 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 statistically significant.

The median age of the included patients was 59.0 years (35.5% were male) and the median BMI was 22.8 kg/m². Of the 203 patients with AAV, 112 had MPA, 54 had GPA and 37 had EGPA. One hundred thirty-six patients had MPO-ANCA (P-ANCA), while 41 patients did not have ANCA. The median BVAS and FFS were 12.0 and 1.0, respectively. The most commonly involved organ was the kidneys (62.6%), followed by the lungs (59.1%), and the common comorbidity was hypertension (38.4%). The median serum albumin level was 3.7 g/dL and the median mBMI was 813.2 kg · g/m² · L (Table 1).

Of the 203 patients, 25 patients (12.3%, 16 MPA, and 9 GPA patients) died during follow (median, 36.6 months). Glucocorticoids were administered to 189 patients (93.1%). In addition, 101 patients received cyclophosphamide and 113 patients received azathioprine (Table 1).

mBMI was significantly correlated with age (r=–0.248), BVAS (r=–0.333), FFS (r=–0.270), erythrocyte sedimentation rate (ESR) (r=–0.476) and C-reactive protein (CRP) (r=–0.555) (Supplementary Figure 1).

There were no significant differences between surviving and deceased patients in terms of medications administered during follow-up (Supplementary Table 1). BMI at diagnosis did not significantly differ between surviving and deceased patients. However, deceased patients exhibited significantly lower serum albumin (p<0.001) and mBMI (p=0.001) at diagnosis than surviving patients (Figure 1).

When the cut-offs of mBMI for all-cause mortality were obtained using the ROC curve (area 0.702, 95% confidence interval [CI] 0.598~0.807), mBMI of 570.1 was chosen for the optimal one due to the maximised sum of sensitivity and specificity. The sensitivity and specificity were 91.0% and 40.0% as the cut-off was set as mBMI ≤570.1 kg · g/m² · L (Figure 2).

When we divided patients into two groups based on mBMI ≤570.1 kg · g/m² · L, 26 of 203 patients were assigned to the group with mBMI ≤570.1 kg · g/m² · L.

All-cause mortality was identified more frequently in AAV patients with mBMI ≤570.1 kg · g/m² · L than those with mBMI >570.1 kg · g/m² · L (38.5% vs. 8.5%, p<0.001).

Furthermore, AAV patients with mBMI ≤570.1 kg · g/m² · L exhibited a significantly higher risk for all-cause mortality than those with mBMI >570.1 kg · g/m² · L (relative risk [RR] 6.750, 95% CI 2.608~17.468) (Figure 2).

AAV patients with mBMI ≤570.1 kg · g/m² · L showed a significantly lower cumulative patients’ survival rate than those with mBMI >570.1 kg · g/m² · L (p<0.001) (Figure 3).

In the univariable Cox hazards model analysis, age, BVAS, FFS, chronic kidney disease (stage 3, 4, and 5), interstitial lung disease, CRP, serum albumin and mBMI at diagnosis were significantly associated with all-cause mortality during follow-up. However, BMI at diagnosis was not associated with all-cause mortality during follow-up (Table 2). In the multivariable Cox hazards model analysis using variables with statistical significance in the univariable analysis, since there was multicollinearity between serum albumin and mBMI (condition number=22.929), the multivariable analysis was conducted twice with either serum albumin or mBMI. In the multivariable analysis using serum albumin, the independent predictors of all-cause mortality in AAV patients were BVAS (HR 1.086, 95% CI 1.018~1.159), FFS (HR 2.099, 95% CI 1.262~3.491), interstitial lung disease (HR 4.699, 95% CI 1.825~12.097) and serum albumin (HR 0.207, 95% CI 0.087~0.492) at diagnosis. Whereas, in the multivariable analysis using mBMI, the independent predictors of all-cause mortality were BVAS (HR 1.091, 95% CI 1.025~1.162), FFS (HR 1.966, 95% CI 1.198~3.226), interstitial lung disease (HR 4.472, 95% CI 1.721~11.619) and mBMI (HR 0.995, 95% CI 0.992~0.998) at diagnosis (Table 3).