Patients with ANCA vasculitis were enrolled at UNC Health clinics and followed in the Glomerular Disease Collaborative Network (GDCN) longitudinal patient registry (16, 17). Patients were diagnosed according to Chapel Hill Consensus Conference criteria (18, 19). Patients were initially recruited from the GDCN registry and provided informed, written consent to participate in sub-study in accordance with the UNCCH Office of Human Research Ethics Institutional Review Board guidelines (IRB study 97–0523). A total of 162 eligible participants with an existing clinical diagnosis of ANCA vasculitis were enrolled. Secondary data analysis was performed under an ancillary study IRB (22–0598), eventually including 112 participants with clinically defined disease status and longitudinal survey follow-up. The 50 participants excluded did not have clinician confirmed relapse or remission dates and/or were lost to follow-up with no available surveys.

Disease activity was defined by the 2003 Birmingham Vasculitis Activity Score (BVAS) and confirmed through clinician chart review (20). Active disease (onset or relapse) was defined as BVAS >0 with clinical and/or laboratory evidence of disease and/or new or escalation of immunosuppressive therapy. Patients with a BVAS of 0 and no clinical or laboratory evidence of active disease within 3 months of the clinical follow-up date were considered in remission. Complete remission after initial diagnosis was required prior to relapse. ANCA serotypes were assessed by indirect immunofluorescence with positive cytoplasmic (C-ANCA) or perinuclear (P-ANCA) staining and antigen-specific PR3 and MPO enzyme-linked immunosorbent assays (ELISA) (20). MPO- and/or P-ANCA were classified together and PR3- and/or C-ANCA were classified together. Patients with only P-ANCA positivity prior to MPO availability were required to have a negative Antinuclear Antibody (21).

Detailed treatment regimens were not analyzed with patient surveys for this study but all patients received comparable treatment as the cohort is out of a single center. At our center, patients who have severe organ-threatening disease are treated with induction therapy, which consists of corticosteroids, cyclophosphamide (unless contraindicated or not tolerated), and/or rituximab, and 15 to 20% also receive plasmapheresis (22). This regimen is followed by maintenance dosing of rituximab for an additional 12–18 months, which is tailored to each patient in the context of other clinical factors (such as infections). Oral maintenance agents, including azathioprine and mycophenolate mofetil, are used in cases where rituximab may not be appropriate. Patients can come off all immunosuppressive therapy for periods of time during their disease course. End of immunosuppression is defined as: the date corticosteroids, azathioprine, mycophenolate mofetil, or oral cyclophosphamide are stopped, 30 days post-treatment of intravenous cyclophosphamide, or the date CD19+ B cells are detected in circulation or 1 year after the last dose if B-cell measures were not available for rituximab (22).

Surveys were collected at three-to-six-month intervals and administered in-person during clinic visits from July 2010 to May 2022. The survey questionnaire was composed by the study team for research done under the Program Project, ANCA Glomerulonephritis: From Molecules to Man (National Institutes of Health, P01DK05833, PI: RJF). The survey questions pertained to a broad range of patient-reported events and exposures in the last three months (see Supplementary Material ). Patient-reported responses were recorded and retrospectively reviewed for this case-control study, comparing participants who relapsed to those who remained in remission following survey enrollment.

During survey collection, participants provided yes or no responses pertaining to infections and insect/tick bites with an extended request if answered yes for further details on the date and type of event. Insect bites were reported based on origin from tick, mosquito, unknown, or other. There were several questions on the survey requesting information on recent hospitalizations, new diagnoses, difficulties from disease burden, and notable life events that collected information on stressors within free text fields. Data from free text fields were systematically reviewed by one individual (MMC) and stress events were categorized as biological, environmental, cognitive, personal behavior, life situation, and other. Infections were categorized as upper respiratory infection, lower respiratory infection, urinary tract infection, or other infection. If a specific categorization for any response was not clear, the survey was reviewed and responses were categorized by committee (MMC, DC, KJ, SH) and a consensus was reached.

We reviewed the entire study cohort with survey responses and correlated clinical disease activity across survey duration to categorize participants into relapse and remission cohorts. Clinical disease activity was determined separately in clinic or in systematic review of medical records by a clinician (DPC, EYW, VKD and/or RJF) and a study team member (LNB, CEJ). A standardized 15-month timeline for all participants provided uniform cohort timeframes for comparison ( Supplementary Figure 3-1 ).

Descriptive statistics include numbers with percentage or median with interquartile range. Demographic characteristics and exposures of interest (stressors, insect bite, and infection) were compared between relapse and remission cohorts using Fisher’s exact test for categorical variables and Wilcoxon Two-Sample Test for continuous variables.

We collected records of stress, insect bites, and infection events from survey measurements. However, some survey measures were incomplete, resulting in missing event records. To address this, exclusion and inclusion criteria were used. Consequently, instead of counting the frequency of events, we calculated an average event occurrence in three months at –15, -12,…, -3 and 0 months to mitigate bias.

The average event rate was defined using the average infection event divided by total observation time for an infection type in three months. These rates over time from –15 to 0 months were compared between relapse and remission groups using Exact Cochran-Armitage Test. P-values less than 0.05 were considered statistically significant. All analyses were conducted with SAS (v.9.4, Cary, N.C.).

Survey data was analyzed from two groups as described in the Methods: those who experienced relapse (n=65 participants) and those in remission (n=47 participants). The relapse and remission cohorts were 55% and 47% female participants, respectively (p=0.44). The median age at the onset of disease was 52 years for the relapse group and 49 years for the remission group (p=0.56). The time from disease onset to enrollment in the survey was a median of 3 years for the relapse cohort and 1 year for the remission cohort. The cohort characteristics are shown in Table 1 . The breakdown of serotypes within each cohort was as follows: 58% PR3-ANCA and 42% MPO-ANCA in the relapse group, and 44% PR3-ANCA and 56% MPO-ANCA in the remission group. Additionally, 1% of the relapse group and 9% of the remission group were dual positive for MPO and PR3-ANCA, or seronegative (p=0.17). Table 2 outlines the available longitudinal survey data for each cohort. The relapse cohort (n=65) had a median follow-up of 8 years with an average of 19 surveys per participant. In comparison, the remission cohort (n=47) had a median follow-up of 3 years with an average of 7 surveys per participant.

All study participants received maintenance therapy during their clinical care during and/or outside of the observed survey period, with 88% receiving rituximab, 54% azathioprine, and 49% mycophenolate mofetil. Moreover, 58% of total participants stopped all immunosuppressive therapy at least once in their disease course.

In the relapse cohort, organ involvement at relapse was documented as 48% kidney, 42% upper respiratory, 25% joints, 12% lungs, 12% musculoskeletal, 9% neurological, and/or 8% dermatological. In the remission cohort, 70% of participants were in complete remission off therapy during the reviewed survey period, while 30% were in remission on therapy.

The frequency of external exposures reported at any time during the 15 months prior to relapse or the remission index date is shown in Table 2 , along with the patient-reported incident frequency of each category of exposure (e.g., stress, insect bites, and infections). Overall, the frequency of any external exposure was not statistically significant between the relapse and remission cohorts (97% versus 87%, respectively, p=0.065). Stress events were also not significantly different between the relapse and remission cohorts (98% versus 77%, p=0.12), and the number of any stressors reported was similar, both with a median of 2 (p=0.41). Reported insect bites were also similar between the two groups (32% in each, p=1.00). Although the reported incidence of infections was not statistically significant between the relapse and remission cohorts (71% versus 55%, p=0.11), the median number of infections reported was statistically higher among the relapse cohort compared to the remission cohort (p=0.03). This finding led us to further explore specific infection types and rates of occurrence.

Of note, subgroup analyses limited to PR3 ANCA serotype and then to MPO ANCA serotype, did not reveal differences in frequency or counts of the exposures between the relapse and remission cohorts ( Supplementary Tables 2-1 , 2-2 ).

The average infection rate of upper respiratory, pulmonary, urinary, and other infections for relapse and relapse cohorts in 3-month intervals over 15 months were compared ( Table 3 ). The relapse cohort consistently had higher rates of upper respiratory infections more than 6 months prior to the relapse as compared to this same time period prior to the remission index date (Exact Cochran-Armitage Test, P <0.0001, Supplementary Figure 3-2A ). Reported pulmonary and urinary infections were lower compared to the rate of reported upper respiratory infections but were generally similar over time between the relapse and remission cohorts (Exact Cochran-Armitage Test, P-value= 0.08 and 0.10, Supplementary Figure 3-2B, C ). “Other reported infections”, which was inclusive of those affecting the skin, eyes, oral cavity, gastrointestinal, or nervous system, were also significantly different between the relapse and remission cohorts (Exact Cochran-Armitage Test, P < 0.0001). The rate of other infections peaked at 15 and 6 months prior to the index date in the relapse cohort, in contrast to an increase in average cases from 15 to 3 months prior to remission index date ( Supplementary Figure 3-2D ). A subgroup analysis did not show serotype specific differences in the exposures ( Supplementary Tables 2-1 , 2-2 ).