Mepolizumab initiation by day 30 and glucocorticoid exposure in newly diagnosed EGPA: a single-center 30-day landmark cohort study Authors: Shota Kaburaki, Toru Tanaka, Koichiro Kamio, Yosuke Tanaka, Namiko Taniuchi, Kazuo Kasahara, Masahiro Seike Published: 2026 ABSTRACT AbstractBackgroundEvidence on mepolizumab soon after glucocorticoid initiation in newly diagnosed eosinophilic granulomatosis with polyangiitis (EGPA) is limited. We compared nonrandomized initial treatment strategies defined by mepolizumab initiation within 30 days after glucocorticoid initiation with respect to post-landmark oral glucocorticoid exposure.MethodsWe performed a single-center pilot retrospective cohort study of 30 patients with newly diagnosed EGPA who started systemic glucocorticoids. Primary analyses used a 30-day landmark cohort of 27 patients who remained under observation through day 30. Patients were classified into a mepolizumab-by-day-30 group and a no-mepolizumab-by-day-30 group; the latter comprised patients without initiation by day 30, regardless of later initiation. The primary outcome was cumulative oral prednisolone exposure from day 30 to 12 months.ResultsThe landmark cohort included 6 patients in the mepolizumab-by-day-30 group and 21 in the no-mepolizumab-by-day-30 group. Cumulative oral prednisolone exposure from day 30 to 12 months was lower in the mepolizumab-by-day-30 group than in the no-mepolizumab-by-day-30 group (24.7 vs. 68.6 mg/kg). Similar differences were seen from day 30 to 6 months (17.6 vs. 43.6 mg/kg) and in oral prednisolone dose at 12 months (0.03 vs. 0.12 mg/kg/day). When intravenous methylprednisolone pulse therapy was added using observed dose and duration as prednisolone-equivalent systemic glucocorticoid exposure, the between-group difference was attenuated, especially over 0–6 months. By 3 months, the Birmingham Vasculitis Activity Score was 0 in all evaluable patients in both groups, and remission differences mainly reflected attainment of prednisolone ≤ 7.5 mg/day.ConclusionsIn this single-center pilot cohort comparing nonrandomized initial treatment strategies, mepolizumab initiation within 30 days after glucocorticoid initiation was associated with lower post-landmark oral glucocorticoid exposure. These findings are hypothesis-generating and do not isolate the causal effect of mepolizumab from calendar-time, treatment-indication, or concomitant-treatment differences.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12890-026-04427-4.Keywords: Asthma, Eosinophilic Granulomatosis with Polyangiitis, Glucocorticoids, Mepolizumab, Retrospective Studies Background Eosinophilic granulomatosis with polyangiitis (EGPA) is a rare systemic vasculitis characterized by asthma, eosinophilia, and granulomatous and/or vasculitic involvement of multiple organs. Induction therapy for active EGPA relies on systemic glucocorticoids, with additional immunosuppressants considered for patients with poor prognostic factors or organ- or life-threatening manifestations [1–4]. Relapse during glucocorticoid tapering is common, and cumulative glucocorticoid exposure contributes substantially to treatment-related morbidity. Cytotoxic agents such as cyclophosphamide can also cause clinically important adverse effects, including infection [2–6]. Mepolizumab, a monoclonal antibody targeting interleukin-5 (IL-5), has established efficacy in relapsing or refractory EGPA and can reduce oral glucocorticoid use [7–10]. Guideline documents allow consideration of mepolizumab in selected patients with active non-severe disease, but most direct evidence comes from relapsing or refractory cohorts [3, 4, 7–11]. Evidence regarding mepolizumab soon after glucocorticoid initiation in newly diagnosed EGPA remains limited to case reports, small series, and a few retrospective studies [12–17]. We therefore compared nonrandomized initial treatment strategies defined by mepolizumab initiation within 30 days after glucocorticoid initiation with respect to post-landmark oral glucocorticoid exposure in newly diagnosed EGPA. Rather than assess routine biologic use at diagnosis, we sought to describe outcomes under different physician-selected initial treatment strategies in routine care. Because exposure was defined during the first 30 days, we used a 30-day landmark framework and analyzed Birmingham Vasculitis Activity Score (BVAS) = 0 and prednisolone ≤ 7.5 mg/day separately to distinguish disease control from tapering policy. Methods Study design and patient selection This single-center retrospective cohort study was conducted at Nippon Medical School Hospital. We reviewed Diagnosis Procedure Combination (DPC)-coded electronic medical records and International Classification of Diseases, Tenth Revision (ICD-10) code M30.1 entries to identify patients with a confirmed or suspected diagnosis of EGPA who had at least one visit to the Department of Pulmonary Medicine and Oncology between April 2010 and August 2025. Patients were included if they met the diagnostic criteria for EGPA proposed by the Japanese Ministry of Health, Labour and Welfare and fulfilled the 2022 American College of Rheumatology/European Alliance of Associations for Rheumatology classification criteria for EGPA [18, 19]. Treatment initiation was defined as the chart-recorded start date of systemic glucocorticoids for EGPA. After chart review, patients who did not have EGPA and those whose EGPA treatment had already been started at another institution were excluded. Patients with follow-up shorter than 30 days were included in the screening flow but excluded from the 30-day landmark cohort used for comparative tables, longitudinal figures, and post-landmark analyses. This study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Nippon Medical School Hospital (approval number B-2024-1026). The Ethics Committee waived the requirement for written informed consent to participate because this retrospective chart review used existing clinical data, imposed no additional intervention, and analyzed de-identified data. Exposure definition and 30-day landmark design Patients were classified according to mepolizumab initiation during the first 30 days after starting systemic glucocorticoids. The mepolizumab-by-day-30 group included patients who initiated mepolizumab at the approved EGPA dose in Japan (300 mg subcutaneously every 4 weeks) within 30 days after glucocorticoid start. All mepolizumab-treated patients received it in routine care as physician-selected add-on therapy after systemic glucocorticoid initiation, for patients judged to have insufficient control with existing glucocorticoid-based treatment. The no-mepolizumab-by-day-30 group comprised patients who had not initiated mepolizumab by day 30 and who were otherwise managed with glucocorticoids with or without other immunosuppressive agents. Treatment decisions, including use of mepolizumab, intravenous methylprednisolone pulse therapy, cyclophosphamide, and azathioprine, were not protocolized and were selected by treating physicians according to organ involvement, disease activity, eosinophilic or airway-predominant manifestations, comorbidities, perceived glucocorticoid-related risk, and drug availability or reimbursement. Because treatment indications were not recorded in a standardized form and the cohort was small, confounding by indication could not be reliably adjusted in this pilot cohort. Some patients in the no-mepolizumab-by-day-30 group later initiated mepolizumab as rescue therapy at the discretion of the treating physicians. Accordingly, the primary estimand was a comparison of nonrandomized initial treatment strategies by mepolizumab timing, rather than ever use versus never use of mepolizumab, with post-landmark outcomes. Because exposure status depended on treatment decisions made during the first 30 days after glucocorticoid initiation, the primary inferential analyses used a 30-day landmark design to align exposure classification with the start of post-classification follow-up and thereby reduce bias arising from time-dependent exposure assignment [20, 21]. The 30-day window was selected because it corresponds to the initial induction-treatment period after systemic glucocorticoid initiation, during which early add-on treatment decisions, inpatient-to-outpatient transition, and the first reassessment of glucocorticoid response commonly occur in routine practice. Data collection and outcomes Clinical data were extracted from the medical records, including demographics, disease characteristics, laboratory findings, pulmonary function data when available, and treatment details. Baseline variables were taken from the diagnostic evaluation performed at or near treatment initiation. Peripheral eosinophil counts were collected at diagnosis and at follow-up time points. Baseline myeloperoxidase antineutrophil cytoplasmic antibody (MPO-ANCA) titers and total immunoglobulin E levels were extracted at diagnosis. Baseline Birmingham Vasculitis Activity Score (BVAS) and Five-Factor Score (FFS) were recorded as standard disease-activity and prognostic measures, respectively [22, 23]. For the descriptive summary of time from glucocorticoid start to the first mepolizumab dose, patients in the mepolizumab-by-day-30 group who received a 3-day intravenous methylprednisolone pulse at initiation were counted from the first day of the pulse course. FFS component items were reported as prognostic-score variables rather than comprehensive organ-involvement variables. The FFS severe gastrointestinal item was defined as severe gastrointestinal involvement used for Five-Factor Score calculation, such as gastrointestinal bleeding, ischemia or infarction, perforation, or pancreatitis, and should not be interpreted as all EGPA-related gastrointestinal involvement [23]. Cumulative oral prednisolone exposure was calculated as the area under the curve using the trapezoidal rule based on body-weight-normalized oral prednisolone doses at day 30 and at 3, 6, and 12 months. Baseline body weight was used for normalization. Dose changes between scheduled assessment time points were not reconstructed from daily prescription records; instead, exposure between adjacent time points was approximated by linear interpolation. The day-30 prednisolone dose was used as the starting value for the post-landmark area-under-the-curve calculation and was not interpreted as an independent post-landmark outcome. Secondary outcomes were cumulative oral prednisolone exposure from day 30 to 6 months, oral prednisolone dose at 12 months, longitudinal changes in BVAS and eosinophil counts, relapse within 12 months, and infection events within 12 months. No institutional protocol mandated a fixed glucocorticoid tapering schedule during the study period. Prednisolone tapering was determined by treating physicians according to disease activity, organ involvement, eosinophil counts, adverse-event risk, and clinical response. Complete remission was defined as BVAS = 0 with prednisolone ≤ 7.5 mg/day. To separate disease activity control from glucocorticoid tapering policy, BVAS = 0 and prednisolone ≤ 7.5 mg/day were also analyzed individually. Patients were considered evaluable at each scheduled outcome time point when the corresponding routine-care assessment was available in the medical record. Date-derived follow-up duration was used to summarize observation time, whereas chart-confirmed 12-month outcome assessments were used to determine evaluability for 12-month outcomes. Relapse was defined as new or recurrent EGPA-related clinical manifestations after initial improvement that led the treating physician to increase systemic glucocorticoids, initiate or intensify immunosuppressive therapy, or add biologic therapy. Isolated laboratory eosinophilia without treatment escalation was not counted as relapse. Infection events were defined as clinically diagnosed infections requiring antimicrobial treatment, hospitalization, or interruption of immunosuppressive or biologic therapy. Severity was classified descriptively as outpatient-treated, hospitalization-requiring, severe/life-threatening, or unknown when severity information was not consistently recorded. Statistical analysis No formal sample size calculation was performed because of the exploratory design and the rarity of EGPA. Continuous variables are reported as medians with interquartile ranges (IQRs), and categorical variables as counts and percentages. Between-group comparisons used the Mann–Whitney U test for continuous variables and Fisher’s exact test for categorical variables. Missing data were handled by available-case analysis for each variable, and no imputation was performed. Given the exploratory design and the small exposed sample, analyses were unadjusted and intended to estimate associations descriptively rather than confirm causality. Because treatment indication was incompletely captured and the cohort was small, confounding by indication could not be reliably adjusted in this pilot cohort. Additional sensitivity analyses were restriction to patients treated from 2022 onward to reduce calendar-time imbalance, exclusion of patients in the no-mepolizumab-by-day-30 group who later initiated mepolizumab within 1 year, and inclusion of intravenous methylprednisolone pulse therapy as prednisolone-equivalent exposure. For the pulse-inclusive sensitivity analysis, intravenous methylprednisolone pulse therapy was incorporated using the actual administered daily dose and duration recorded in the medical chart. The total methylprednisolone dose was converted to prednisolone-equivalent dose using a conversion factor of 1.25 and normalized by baseline body weight. This pulse exposure was added to oral prednisolone area-under-the-curve exposure for the corresponding analysis. All analyses were performed using Python 3.11. Because of the exploratory design and small sample size, p-values are reported descriptively, and no adjustment for multiple comparisons was performed. Standardized mean differences were calculated to describe covariate imbalance. For continuous variables, standardized mean differences were calculated using pooled standard deviations; for binary variables, they were calculated using the difference in proportions divided by the square root of the average binomial variance. Bootstrap 95% confidence intervals for selected effect sizes were calculated using 10,000 resamples with a fixed random seed and interpreted descriptively. Because follow-up assessments were performed in routine care rather than at protocol-mandated event-assessment visits, attainment of BVAS = 0 and prednisolone ≤ 7.5 mg/day was summarized at day 30 and at 3, 6, and 12 months with numbers at risk and censoring status, rather than analyzed as time-to-event outcomes. After completion of the analyses and interpretation by the authors, large language models (ChatGPT, OpenAI, and Gemini, Google) were used to assist with drafting and language refinement of the manuscript. All generated text was reviewed and edited by the authors, who take full responsibility for the final content. Results Patient characteristics Thirty-four patients were identified through the DPC/ICD-10 code M30.1 search. Four were excluded after chart review (3 did not meet EGPA diagnostic criteria and 1 had already started treatment at a prior hospital). Among the 30 included patients, 6 initiated mepolizumab within 30 days after glucocorticoid initiation and 24 did not. For the 30-day landmark analyses, 3 patients in the no-mepolizumab-by-day-30 group who had less than 30 days of follow-up were excluded, leaving a landmark cohort of 27 patients: 6 with mepolizumab initiation within 30 days and 21 with no mepolizumab initiation by day 30. Baseline and treatment characteristics of the landmark cohort Baseline characteristics of the 30-day landmark cohort are summarized in Table 1, concomitant treatment during the first year is summarized in Table 2, and phenotype-related variables and FFS prognostic components with standardized mean differences are shown in Table S1. The landmark cohort comprised 27 patients: 6 in the mepolizumab-by-day-30 group and 21 in the no-mepolizumab-by-day-30 group. At treatment initiation, the mepolizumab-by-day-30 group started with a lower oral prednisolone dose (0.41 vs. 0.84 mg/kg/day; p = 0.012), and intravenous methylprednisolone pulse therapy was numerically more frequent (83.3% vs. 42.9%; p = 0.165). Among the 21 patients in the no-mepolizumab-by-day-30 group, 7 eventually initiated mepolizumab during total follow-up. Of these, 2 initiated mepolizumab within the prespecified 12-month outcome window at a median of 185 days (IQR 151–219), whereas 5 initiated mepolizumab after 12 months. No patients received biologics other than mepolizumab during the 12-month outcome window. Table 1Baseline characteristics of the 30-day landmark cohortVariableNo mepolizumab by day 30 (n = 21)Mepolizumab by day 30 (n = 6)p valueA. DemographicsAge, years57.00 (51.00–66.00)63.50 (51.25–68.25)0.661Female sex, n (%)12/21 (57.1)4/6 (66.7)1.000BMI, kg/m²22.05 (20.39–26.02)22.13 (19.98–24.53)0.887Never smoker, n (%)17/21 (81.0)6/6 (100.0)0.545Calendar year of treatment start2018.0 (2015.0–2022.0)2024.0 (2024.0–2024.8)0.001Treatment start in 2022 or later, n (%)6/21 (28.6)6/6 (100.0)0.003Diagnosis-to-treatment interval, days0.0 (0.0–2.0)0.5 (0.0–1.8)0.753B. Disease activity and severityBVAS at diagnosis19.0 (13.0–22.0)14.0 (12.0–16.0)0.084Five-Factor Score1.00 (0.00–1.00)0.50 (0.00–1.75)0.899FFS cardiac item, n (%)3/21 (14.3)0/6 (0.0)1.000FFS severe gastrointestinal item, n (%)0/21 (0.0)0/6 (0.0)1.000FFS renal item (serum creatinine > 1.69 mg/dL), n (%)0/21 (0.0)0/6 (0.0)1.000Age > 65 years, n (%)6/21 (28.6)3/6 (50.0)0.367No upper airway lesions, n (%)5/21 (23.8)2/6 (33.3)0.633C. Laboratory dataPeripheral eosinophils, /µL11,077 (5838–18318)6923 (3325–11022)0.239Total IgE, IU/mL1288 (923–1567)421 (103–865)0.036MPO-ANCA, U/mL0.0 (0.0–180.8)3.9 (0.0–22.4)0.873MPO-ANCA positivity, n/N (%)9/21 (42.9)3/6 (50.0)1.000D. Pulmonary function testsFVC, L3.17 (2.62–3.44)2.44 (2.33–2.49)0.065FVC, % predicted108.9 (99.8–127.1)100.0 (80.6–101.1)0.048FEV1, L2.28 (1.98–2.59)1.67 (1.66–1.94)0.126FEV1, % predicted91.4 (84.6–105.6)83.8 (60.5–95.2)0.328FEV1/FVC ratio, %74.4 (64.8–79.1)71.7 (68.2–77.9)0.879DLCO, mL/min/mmHg16.20 (13.38–18.12)18.01 (15.93–19.28)0.604DLCO, % predicted98.0 (94.2–123.8)92.0 (84.6–99.0)0.604Note: Data are presented as median (interquartile range) or n/N (%). Pulmonary function variables were analyzed among patients with available baseline measurements. MPO-ANCA positivity was defined as a value of ≥ 3.5 U/mL. p values are descriptive and were not adjusted for multiple comparisonsFFS cardiac, renal, and severe gastrointestinal items are Five-Factor Score prognostic components and do not represent comprehensive organ-involvement variables. The FFS severe gastrointestinal item was defined as severe gastrointestinal involvement used for Five-Factor Score calculation, such as gastrointestinal bleeding, ischemia or infarction, perforation, or pancreatitisAbbreviations: BMI Body mass index, BVAS Birmingham Vasculitis Activity Score, DLCO Diffusing capacity of the lung for carbon monoxide, FEV1 Forced expiratory volume in 1 s, FFS Five-Factor Score, FVC Forced vital capacity, IgE Immunoglobulin E, MPO-ANCA Myeloperoxidase-antineutrophil cytoplasmic antibody Baseline characteristics of the 30-day landmark cohort Note: Data are presented as median (interquartile range) or n/N (%). Pulmonary function variables were analyzed among patients with available baseline measurements. MPO-ANCA positivity was defined as a value of ≥ 3.5 U/mL. p values are descriptive and were not adjusted for multiple comparisons FFS cardiac, renal, and severe gastrointestinal items are Five-Factor Score prognostic components and do not represent comprehensive organ-involvement variables. The FFS severe gastrointestinal item was defined as severe gastrointestinal involvement used for Five-Factor Score calculation, such as gastrointestinal bleeding, ischemia or infarction, perforation, or pancreatitis Abbreviations: BMI Body mass index, BVAS Birmingham Vasculitis Activity Score, DLCO Diffusing capacity of the lung for carbon monoxide, FEV1 Forced expiratory volume in 1 s, FFS Five-Factor Score, FVC Forced vital capacity, IgE Immunoglobulin E, MPO-ANCA Myeloperoxidase-antineutrophil cytoplasmic antibody Table 2Concomitant treatment and mepolizumab timing during the first year in the 30-day landmark cohortParameterNo mepolizumab by day 30 (n = 21)Mepolizumab by day 30 (n = 6)p valueA. Concomitant treatment within 12 months, n (%)IV methylprednisolone pulse9/21 (42.9)5/6 (83.3)0.165Mepolizumab within 12 months2/21 (9.5)6/6 (100.0)Not testedIV cyclophosphamide9/21 (42.9)0/6 (0.0)0.071Azathioprine5/21 (23.8)0/6 (0.0)0.555IV immunoglobulin5/21 (23.8)1/6 (16.7)1.000B. Timing of first mepolizumab among recipientsDays from glucocorticoid start to first mepolizumab185.0 (151.0–219.0)(among n = 2)7.0 (6.2–7.8)—Note: Data are presented as n/N (%) or median (interquartile range). Values are shown for the 30-day landmark cohort. The interval to first mepolizumab is reported among recipients within the first year. In the no-mepolizumab-by-day-30 group, 7 patients eventually received mepolizumab during total follow-up, but only 2 initiated it within 12 months. The p value for mepolizumab use within 12 months was not tested because this variable is directly related to the day-30 exposure definition. p values are descriptive and were not adjusted for multiple comparisonsAbbreviation: IV Intravenous Concomitant treatment and mepolizumab timing during the first year in the 30-day landmark cohort 185.0 (151.0–219.0) (among n = 2) Note: Data are presented as n/N (%) or median (interquartile range). Values are shown for the 30-day landmark cohort. The interval to first mepolizumab is reported among recipients within the first year. In the no-mepolizumab-by-day-30 group, 7 patients eventually received mepolizumab during total follow-up, but only 2 initiated it within 12 months. The p value for mepolizumab use within 12 months was not tested because this variable is directly related to the day-30 exposure definition. p values are descriptive and were not adjusted for multiple comparisons Abbreviation: IV Intravenous At the 30-day landmark, oral prednisolone dose remained lower in the mepolizumab-by-day-30 group (median 0.23 vs. 0.51 mg/kg/day; p < 0.001). BVAS at 1 month was also lower (median 0 vs. 2; p = 0.032), and BVAS = 0 was observed in all six patients in the mepolizumab-by-day-30 group versus 10 of 21 patients (47.6%) in the no-mepolizumab-by-day-30 group. Because these 1-month measures coincide with the exposure-classification window, they are interpreted descriptively rather than as post-landmark comparative endpoints. Follow-up duration was defined as the interval from systemic glucocorticoid initiation to the last evaluable clinical encounter. In the 30-day landmark cohort, median follow-up was 929 days (IQR 458–2801.5; 124.0 person-years); group-specific follow-up was 2109 days (IQR 624–3272; 116.0 person-years) in the no-mepolizumab-by-day-30 group and 433.5 days (IQR 374.8–541.0; 8.0 person-years) in the mepolizumab-by-day-30 group. Post-landmark glucocorticoid exposure and disease control In the landmark cohort, cumulative oral prednisolone exposure from day 30 to 12 months was lower in the mepolizumab-by-day-30 group than in the no-mepolizumab-by-day-30 group (median 24.7 [IQR 22.7–30.2] vs. 68.6 [56.2–79.2] mg/kg; p < 0.001). The same pattern was observed from day 30 to 6 months, and oral prednisolone dose at 12 months also remained lower in the mepolizumab-by-day-30 group (Table 3; Fig. 2A and B). Table 3Post-landmark outcomes in the 30-day landmark cohortOutcomeNo mepolizumab by day 30Mepolizumab by day 30p valueEffect size (bootstrap 95% CI)Cumulative oral PSL exposure, day 30 to 12 months (mg/kg)68.6 (56.2–79.2)24.7 (22.7–30.2)< 0.001-43.9 (-55.7 to -27.4)Cumulative oral PSL exposure, day 30 to 6 months (mg/kg)43.6 (36.0–49.0)17.6 (14.7–19.9)< 0.001-26.0 (-32.5 to -17.8)Oral PSL dose at 12 months (mg/kg/day)0.12 (0.06–0.13)0.03 (0.01–0.04)0.001-0.08 (-0.13 to -0.04)BVAS at 3 months0.0 (0.0–0.0)0.0 (0.0–0.0)—Not estimableBVAS at 6 months0.0 (0.0–0.0)0.0 (0.0–0.0)—Not estimableBVAS at 12 months0.0 (0.0–0.0)0.0 (0.0–0.0)—Not estimableBVAS = 0 at 3 months21/21 (100.0)6/6 (100.0)1.000Not estimableBVAS = 0 at 6 months21/21 (100.0)6/6 (100.0)1.000Not estimableBVAS = 0 at 12 months21/21 (100.0)6/6 (100.0)1.000Not estimablePSL ≤ 7.5 mg/day at 3 months0/21 (0.0)5/6 (83.3)< 0.0010.833 (0.500 to 1.000)PSL ≤ 7.5 mg/day at 6 months7/21 (33.3)6/6 (100.0)0.0060.667 (0.476 to 0.857)PSL ≤ 7.5 mg/day at 12 months15/21 (71.4)6/6 (100.0)0.2840.286 (0.095 to 0.476)Complete remission at 3 months0/21 (0.0)5/6 (83.3)< 0.0010.833 (0.500 to 1.000)Complete remission at 6 months7/21 (33.3)6/6 (100.0)0.0060.667 (0.476 to 0.857)Complete remission at 12 months15/21 (71.4)6/6 (100.0)0.2840.286 (0.095 to 0.476)Relapse within 12 months2/21 (9.5)0/6 (0.0)1.000-0.095 (-0.238 to 0.000)Infection within 12 months3/21 (14.3)0/6 (0.0)1.000-0.143 (-0.286 to 0.000)Note: Data are presented as median (interquartile range) or n/N (%). Effect sizes are reported as the mepolizumab-by-day-30 group minus the no-mepolizumab-by-day-30 group. Bootstrap confidence intervals are descriptive. A dash indicates that a p value was not calculated because all observed values were identical. Not estimable indicates that the effect size could not be calculated because of no between-patient variability or no events in both groups. p values are descriptive and were not adjusted for multiple comparisonsAbbreviations: BVAS Birmingham Vasculitis Activity Score, CI Confidence interval, PSL Prednisolone Post-landmark outcomes in the 30-day landmark cohort Note: Data are presented as median (interquartile range) or n/N (%). Effect sizes are reported as the mepolizumab-by-day-30 group minus the no-mepolizumab-by-day-30 group. Bootstrap confidence intervals are descriptive. A dash indicates that a p value was not calculated because all observed values were identical. Not estimable indicates that the effect size could not be calculated because of no between-patient variability or no events in both groups. p values are descriptive and were not adjusted for multiple comparisons Abbreviations: BVAS Birmingham Vasculitis Activity Score, CI Confidence interval, PSL Prednisolone Longitudinal trajectories in the 30-day landmark cohort. Left-column panels show the no-mepolizumab-by-day-30 group (n = 21), and right-column panels show the mepolizumab-by-day-30 group (n = 6). Panels A and B depict oral prednisolone dose normalized by body weight (mg/kg/day), panels C and D depict BVAS, and panels E and F depict peripheral eosinophil counts on a logarithmic scale. Time points are baseline and 1, 3, 6, and 12 months after glucocorticoid initiation. Thin lines represent individual patients, and thick lines represent group medians. The vertical dashed line marks the 30-day landmark, separating the exposure-classification window from post-landmark follow-up. For logarithmic visualization, observed values of 0/µL were plotted as 1/µL; no actual measurements of 1/µL were recorded. Abbreviations: BL, baseline; BVAS, Birmingham Vasculitis Activity Score Screening flow of the DPC/ICD-10-identified cohort. Thirty-four patients were identified through the DPC/ICD-10 code M30.1 search. Four were excluded after chart review (3 did not meet EGPA diagnostic criteria and 1 had already started treatment at a prior hospital). Among the 30 included patients, 6 initiated mepolizumab within 30 days after glucocorticoid initiation and 24 did not. For the 30-day landmark analyses, 3 patients in the no-mepolizumab-by-day-30 group with < 30 days of follow-up were excluded, leaving 21 patients in the no-mepolizumab-by-day-30 group and 6 in the mepolizumab-by-day-30 group. Among the 21 patients in the no-mepolizumab-by-day-30 group, 7 later initiated mepolizumab during total follow-up, including 2 within 12 months and 5 after 12 months. Abbreviations: DPC, Diagnosis Procedure Combination; EGPA, eosinophilic granulomatosis with polyangiitis; ICD-10, International Classification of Diseases, Tenth Revision By 3 months, BVAS = 0 was achieved in all evaluable patients in both groups. In contrast, prednisolone ≤ 7.5 mg/day was more frequent in the mepolizumab-by-day-30 group at 3 months (83.3% vs. 0.0%; p < 0.001) and 6 months (100.0% vs. 33.3%; p = 0.006), whereas the difference was less marked by 12 months (100.0% vs. 71.4%; p = 0.284). Numbers at risk, attainment of BVAS = 0 and prednisolone ≤ 7.5 mg/day, and censoring status at day 30 and at 3, 6, and 12 months are shown in Table S2. Descriptive effect sizes with bootstrap 95% confidence intervals for primary and secondary outcomes are shown in Table 3. Peripheral eosinophil counts were lower in the mepolizumab-by-day-30 group at 6 months (median 8.2 vs. 132.8 /µL; p = 0.006) and 12 months (29.9 vs. 262.3 /µL; p = 0.002). Relapse within 12 months occurred in 0 of 6 patients in the mepolizumab-by-day-30 group and 2 of 21 patients in the no-mepolizumab-by-day-30 group. Infection events within 12 months occurred in 0 of 6 versus 3 of 21 patients, respectively, although the numbers were too small for stable inference. The three infection events in the no-mepolizumab-by-day-30 group comprised two outpatient-treated herpes zoster events and one hospitalization-requiring bacterial pneumonia. Group-specific individual trajectories and group medians for oral prednisolone, BVAS, and peripheral eosinophils are shown in Fig. 2. Sensitivity analyses The association between the mepolizumab-by-day-30 strategy and lower post-landmark oral prednisolone exposure remained directionally similar in all sensitivity analyses (Table S3). In the restriction to patients treated from 2022 onward (n = 12; 6 mepolizumab-by-day-30 and 6 no-mepolizumab-by-day-30), cumulative oral prednisolone exposure from day 30 to 12 months remained lower in the mepolizumab-by-day-30 group (24.7 vs. 54.0 mg/kg; p = 0.005). Similar differences were seen from day 30 to 6 months (17.6 vs. 35.9 mg/kg; p = 0.005) and in oral prednisolone dose at 12 months (0.03 vs. 0.08 mg/kg/day; p = 0.013). The result was also similar after excluding patients in the no-mepolizumab-by-day-30 group who later initiated mepolizumab within 1 year (24.7 vs. 68.6 mg/kg; p = 0.001). When intravenous methylprednisolone pulse therapy was added using observed dose and duration as prednisolone-equivalent systemic glucocorticoid exposure, the between-group difference was attenuated, especially over 0–6 months (50.1 vs. 61.6 mg/kg; p = 0.431), whereas a smaller difference remained over 0–12 months (60.1 vs. 80.8 mg/kg; p = 0.044). Discussion In this single-center pilot retrospective cohort of newly diagnosed EGPA, the nonrandomized mepolizumab-by-day-30 strategy was associated with lower oral glucocorticoid exposure after the 30-day landmark, whereas post-landmark differences in BVAS were modest. This pattern should be interpreted as a comparison of combined initial treatment strategies rather than an isolated effect of mepolizumab. The observed difference may reflect greater early use of methylprednisolone pulse therapy, physician-selected tapering, treatment era, and phenotype differences, as well as mepolizumab timing. This interpretation fits the established anti–IL-5 trial literature, but only in part. In the pivotal phase 3 placebo-controlled trial, mepolizumab increased accrued remission and reduced glucocorticoid use in relapsing or refractory EGPA, and the later head-to-head benralizumab-versus-mepolizumab trial showed noninferior remission induction in the same relapsing or refractory setting [7–10, 24]. However, those trials were not designed to test biologic use during the first month after glucocorticoid initiation for newly diagnosed EGPA. Thus, the present study addresses early add-on mepolizumab after glucocorticoid initiation rather than biologic-first treatment at diagnosis. The pattern observed here is biologically plausible given the dual pathobiology of EGPA. EGPA spans eosinophilic tissue inflammation and airway disease on the one hand, and antineutrophil cytoplasmic antibody-associated or vasculitic organ injury on the other.25–27 Anti–IL-5 therapy is most directly expected to suppress eosinophilic inflammation, asthma, and sinonasal activity, whereas conventional induction treatment remains central when clinicians are concerned about organ-threatening vasculitic manifestations. Consistent with that framework, the clearest between-group difference in our cohort involved ongoing oral prednisolone exposure, whereas differences in total systemic glucocorticoid exposure were smaller because pulse methylprednisolone was used more often in the mepolizumab-by-day-30 group, and BVAS = 0 rates were already similar by 3 months. Thus, once initial systemic inflammation had been brought under control with glucocorticoids, mepolizumab may have facilitated safer tapering of maintenance oral glucocorticoids rather than materially changing early vasculitis control. These findings align with the two Japanese studies that most directly examined early add-on mepolizumab. In Ueno et al., patients with severe EGPA treated with high-dose glucocorticoids plus mepolizumab or intravenous cyclophosphamide showed similar early reductions in BVAS and eosinophils, whereas the main between-group difference emerged later in glucocorticoid tapering.12 In Suzuki et al., 10 patients started mepolizumab a mean of 14 days after glucocorticoid initiation; all reached remission by week 8, and by week 104 half had discontinued glucocorticoids and 70% had reduced prednisolone to ≤ 4 mg/day, although some also received cyclophosphamide or intravenous immunoglobulin and there was no concurrent control group.17 Taken together, these studies suggest that, once glucocorticoids have started, the most consistent signal of early add-on mepolizumab is subsequent glucocorticoid sparing rather than large early separation in disease-activity scores. Our cohort adds a complementary perspective by providing a pragmatic comparison of post-classification outcomes under different initial treatment strategies, although the small sample and nonrandom treatment allocation preclude causal attribution to mepolizumab alone [28–31]. These data should therefore be interpreted as a comparison of nonrandomized initial treatment strategies rather than the isolated causal effect of mepolizumab. Treatment allocation was nonrandom and physician selected, the mepolizumab-by-day-30 group was very small, and strong calendar-time, phenotype, and concomitant-treatment imbalances were present. Table S1 summarizes these imbalances using phenotype-related variables and FFS prognostic components with standardized mean differences. Table S1 includes both phenotype-related variables available from chart review and FFS prognostic components; therefore, FFS cardiac, renal, and severe gastrointestinal items were not interpreted as comprehensive organ-involvement variables. The pulse-inclusive sensitivity analysis using observed methylprednisolone doses attenuated the systemic glucocorticoid difference, especially over 0–6 months, supporting the interpretation that early pulse use and tapering practice contributed to the observed exposure pattern. In addition, 7 of 21 patients in the no-mepolizumab-by-day-30 group later initiated mepolizumab during total follow-up, including 2 within 12 months, so the analysis addresses timing of initiation strategy rather than early versus never use. Given the small sample, p-values and between-group differences should be interpreted descriptively. The landmark cohort was too small for reliable multivariable adjustment, and confounding by indication could not be reliably adjusted. Although landmark analysis is appropriate when exposure is defined during an initial assessment window, it excludes patients not observed to the landmark; our estimates therefore apply to patients observed through day 30. This single-center pulmonary cohort may overrepresent airway-predominant EGPA, limiting generalizability. Tapering schedules were not protocolized, and cumulative glucocorticoid exposure was estimated from scheduled dose time points rather than complete prescription histories. Conclusions In newly diagnosed EGPA managed in routine care, a nonrandomized initial strategy defined by mepolizumab initiation within 30 days after glucocorticoid initiation was associated with lower post-landmark oral glucocorticoid exposure, whereas post-landmark disease-activity control was similar after 3 months. Because this study compared physician-selected strategies across markedly different calendar eras, phenotype distributions, and concomitant-treatment patterns, the findings are hypothesis-generating and should not be interpreted as evidence of the isolated causal effect of mepolizumab or as support for routine universal biologic use at diagnosis. Prospective multicenter studies with explicit capture of treatment indication, organ phenotype, tapering decisions, and time-varying exposure are needed. Supplementary Information Supplementary Material 1. Acknowledgements The authors thank the clinical staff involved in the care of patients with EGPA at Nippon Medical School Hospital. Abbreviations Antineutrophil cytoplasmic antibody Birmingham Vasculitis Activity Score Diagnosis Procedure Combination Eosinophilic granulomatosis with polyangiitis Forced expiratory volume in 1 s Five-Factor Score Forced vital capacity International Classification of Diseases, Tenth Revision Interleukin-5 Interquartile range Myeloperoxidase antineutrophil cytoplasmic antibody Prednisolone Standardized mean difference Authors’ contributions S.K. contributed to conceptualization, data curation, formal analysis, investigation, methodology, visualization, and writing—original draft. T.T. contributed to data curation, investigation, supervision, and writing—review and editing. K.Ka. contributed to project administration, supervision, and writing—review and editing. Y.T. contributed to writing—review and editing. N.T. contributed to writing—review and editing. K.Kas. contributed to writing—review and editing. M.S. contributed to conceptualization, methodology, project administration, resources, supervision, and writing—review and editing. All authors read and approved the final manuscript. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Data availability The individual-level clinical dataset is not publicly available because it contains potentially re-identifiable information from a small rare-disease cohort. De-identified data sufficient to reproduce the main analyses are available from the corresponding author on reasonable request, subject to institutional and ethical approval where applicable. Declarations Ethics approval and consent to participate This study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Nippon Medical School Hospital (approval number B-2024-1026). The Ethics Committee waived the requirement for written informed consent to participate because this retrospective chart review used existing clinical data, imposed no additional intervention, and analyzed de-identified data. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The individual-level clinical dataset is not publicly available because it contains potentially re-identifiable information from a small rare-disease cohort. De-identified data sufficient to reproduce the main analyses are available from the corresponding author on reasonable request, subject to institutional and ethical approval where applicable.