Our center enrolled a total of 436 patients with AAV between 2015 and 2022, out of which 90 patients were diagnosed with MPA and underwent renal biopsy. These patients had adequate clinical and laboratory data available at the time of diagnosis and during their follow-up period. The main characteristics of the patients are presented in Table 1. The majority of patients were female (63%) and the median (interquartile range, IQR) age at diagnosis was 63 (58–68) years. The most common symptoms were fever and arthralgia. Kidneys and lungs were the organs most frequently affected. Renal involvement was usually severe in MPA, with a mean eGFR of 25 (12–50) ml/min per 1.73 m2, and renal impairment was the most common kidney syndrome, observed in 44 (49%) patients. Thirty (33%) patients had RPGN, and 12 (13%) patients required kidney replacement therapy (KRT) at onset. Eleven patients (12%) were classified as focal, 31 (35%) as mixed-class, 28 (31%) as crescentic, and 20 (22%) as sclerosing. In this cohort, 61 (68%) patients had pulmonary involvement, with usual interstitial pneumonia (UIP) or nonspecific interstitial pneumonia (NSIP) observed in 42 (47%) patients. Sixteen (18%) cases had bronchiectasis, 2 (2%) patients had alveolar hemorrhage, and 1 (1%) case had a pulmonary mass.

All 90 consecutive patients were given remission-induction therapies. The majority (88%) received intravenous cyclophosphamide (IVC) pulses and steroids, while rituximab plus steroids was given to 6% of patients, and 1% received glucocorticoids alone. The remaining patients received various combinations of glucocorticoids, mycophenolate mofetil, methotrexate, and Tripterygium wilfordii. Four patients (4.4%) with severe renal involvement or diffuse alveolar hemorrhage received additional therapy with plasma exchange. Maintenance treatment in 92% of patients included azathioprine, methotrexate, mycophenolate mofetil, or low-dose cyclophosphamide (once every 3 months).

After 6 months, 79 patients (88%) were followed-up. Of these, 33% achieved CR, 33% achieved PR, 17% had uncontrolled disease progression, and 4.4% showed no response. In the final follow-up of 78 patients, 36% achieved CR, 20% achieved PR, 29% had uncontrolled disease progression, and 2.2% showed no response. During follow-up, 19% of patients experienced at least one recurrence, with a mean time to relapse of 24 (17–54.5) months. The main cause of relapse was discontinuation or withdrawal of immunosuppressive agents after infection.

The median (IQR) time to kidney failure or last follow-up was 22.5 (3.8–54.3) months. At diagnosis, 13% of patients were on KRT, while 11% were classified as kidney failure (CKD 5) at month 6 and 22% at last follow-up. Seventeen patients (19%) were on dialysis finally. Infection was a common occurrence during treatment, with 36% of patients developing an infection. The most common type was pulmonary infection (32%, n = 29), followed by herpes zoster (5.6%, n = 5). Seven patients (7.8%) ultimately died during the follow-up period. The causes of death were severe pulmonary infection in three patients, cerebral hemorrhage in one, sudden death in one, status epilepticus in one, and lung cancer in one. The 3-year and 5-year survival rates after diagnosis were 91.7% and 88.7%, respectively (Fig. 4).

In order to identify possible subtypes of MPA, we utilized a method of unsupervised clustering of the clinical characteristics. After fitting candidate cluster models using the k-prototypes clustering method, we concluded that the four-subgroup model had the most favorable fit statistics (Supplementary Fig. 1). These four subgroups were found to be clinically significant (Table 1, Figs. 1 and 2). Specifically, we defined the four groups as follows: renal impairment type (cluster 1), pure type (cluster 2), systemic inflammation type (cluster 3), and rapid progress type (cluster 4). Furthermore, we discovered that these subgroups were associated with different prognoses. We conducted an analysis of renal outcomes and survival among the four subgroups using Kaplan–Meier survival curves (Figs. 3 and 4). The distribution of the different CKD stages at baseline and last follow-up between the four subgroups were shown in Supplementary Fig. 2

The correlations between clinical features and clusters. The correlation matrix indicating the percentage of a given feature (row) in a cluster (column). The size of each blue dot on the visualization indicating the proportion of patients in that cluster. ENT, ear, nose, and throat, UIP, usual interstitial pneumonia, NSIP, nonspecific interstitial pneumonia

The identification of clinical subtypes of MPA patients. A A UMAP visualization of both clinical and laboratory data from 90 patients at baseline was performed, which revealed distinct grouping of patients into 4 clusters. B Laboratory results projected onto UMAP plots, including Cr (creatinine, a), eGFR (estimated glomerular filtration rate, b), CRP (C-reactive protein CRP, c), BVAS (Birmingham Vasculitis Activity Score, d), and total UPro24h (urine protein of 24 h, e). The scale showed the original data without transformation

Kaplan–Meier estimates and forest plots indicating the risks of kidney survival, according to clusters

Kaplan–Meier estimates and forest plots indicating the risks of survival, according to clusters

As shown in Table 1 and Fig. 1, 35 patients (39%) were classified as having the renal impairment type (cluster 1). Among these patients, 71.4% were female, and 80% exhibited renal impairment and had a higher average concentration of serum creatinine (243 (184–308) μmol/L) at onset. About half of these patients had the crescentic types, and 69% experienced respiratory system involvement. The response to treatment was poor; only 26% achieved complete remission (CR), 43% of patients suffered from infection, 11% dying, and 17% developed renal failure at the final follow-up (Figs. 3 and 4, Supplementary Fig. 2).

In contrast, 20 patients (22%) were classified as cluster 2, almost all of whom were female (90%). Most of these patients had kidney involvement exclusively, and the histological types were mixed (65%), with crescentic (15%) and focal (15%) types. These patients exhibited the lowest levels of BVAS (12 (10–13)) and serum creatinine concentration (103 (82–163) μmol/L). The response to treatment was favorable; 55% received CR, and none of them died or progressed to renal failure. The total infection rate was only 10% during follow-up.

Patients in cluster 3 (26%, n = 23) were predominantly male (70%) and presented with fever and high C-reactive protein (CRP) levels as the main characteristics. Both the kidney and respiratory systems were all involved (100%), as well as ear, nose, and throat (ENT) (13%) and nervous system (13%). Nearly half of the histological types were mixed (48%). Despite with the highest BVAS levels (18 (16–19)), the creatinine level (127 (109–277) μmol/L) was relatively lower. The patients responded well to treatment, with 53% achieving CR, 9% dying, and 4% developing renal failure (Figs. 3 and 4, Supplementary Fig. 2).

During follow-up, we found that patients in this group had the highest infection rate (52%, n = 12).

Finally, cluster 4 (13%, n = 12) had the lowest proportion of patients and was characterized by rapidly progressive glomerulonephritis, with most patients requiring kidney replacement therapy (KRT) at the time of diagnosis. Some of these patients also exhibited diffuse alveolar hemorrhage (17%). Histological types showed two types: crescentic (58.3%) and sclerotic (41.7%). Despite aggressive treatment, none of the patients achieved CR, 8% dying, and almost all of them (75%) developed kidney failure and required dialysis (Figs. 3 and 4, Supplementary Fig. 2).