Pathological severity determines the renal recovery for anti-myeloperoxidase antibody-associated vasculitis requiring dialysis at disease onset: a retrospective study

Between January 2005 and May 2017, 244 patients with AAV were diagnosed at the Tianjin Medical University General Hospital. All patients fulfilled the Chapel Hill Consensus Conference classification [9]. Only 7 patients were proteinase 3 (PR3)-ANCA positive, all other patients were myeloperoxidase (MPO)-ANCA positive. Patients with PR3-ANCA were excluded because of insufficient numbers. Among the 237 patients with MPO-ANCA, 217 patients had kidney injury (including haematuria, proteinuria and abnormality of kidney function), but 102 patients did not undergo kidney biopsy. Among these 102 patients, 74 patients did not agree to kidney biopsy (Sixteen patients needed dialysis at presentation and 10 patients recovered kidney function within 3 months after treatment) and 28 patients did not receive kidney biopsy because of the insufficient thickness of kidney cortex (Twelve patients needed dialysis at presentation and none recovered kidney function within 3 months following treatment). Among the 115 patients who received kidney biopsy, 46 patients required dialysis at disease onset. In the current study, dialysis dependence was defined as when the patients could not acquire dialysis independence for at least 3 months after treatment, so 6 patients were excluded because they died before getting dialysis independence within 3 months. In summary, 40 patients were included in this study (Fig. 1). The research was in compliance of the declaration of Helsinki and the protocol was approved by the institutional review board of Tianjin Medical University General Hospital. Informed consent was not required because the study was retrospective.

Clinical data included the following: gender, age (years), time from first symptom (days) and the level of Birmingham Vasculitis Activity Score (BVAS). Blood samples were taken for laboratory test. Estimated glomerular filtration rate (eGFR) was calculated with an equation described previously [10].

Renal biopsies were performed and evaluated using direct immunofluorescence and light microscopy by two pathologists. According to the pathological classification described previously, the biopsy was classified as focal, crescentic, mixed and sclerotic class [8]. The presence of fibrinoid necrosis, cellular crescents and global glomerulosclerosis of each glomerulus was scored. The presence of glomerular lesions was calculated as the percentage of the total number of glomeruli. Interstitial fibrosis (−/+/++), tubular atrophy (−/+/++) and arterial wall thickening (−/+) were scored semiquantitatively [11].

All patients received hemodialysis following admission. Indications of dialysis included serum creatinine > 500 μmol/L, severe electrolyte and acid-base abnormalities, volume overload, pericarditis and encephalopathy [12]. Dialysis was terminated when eGFR surpassed 15 ml/min/1.73m² and there was no hyperkalemia, heart failure or edema. In the induction of remission, all patients received corticosteroid treatment. Oral or intravenous corticosteroid was prescribed at an initial dosage of 1 mg/kg/day for 1 month, with reducing doses over time. Due to the severe kidney injury, it was recommended that all patients receive PE treatment, but due to restrictions in the plasma supply, only 14 patients received PE treatment (2000 ml per treatment session, 3–10 treatment sessions). The other 26 patients who could not be provided enough plasma did not receive PE treatment; they received corticosteroid pulse therapy. Intravenous methylprednisolone pulse therapy (MP) was prescribed for 3 consecutive days (200–500 mg/d) and the MP regimen was repeated after a week if necessary. The majority of patients (31 of 40) received intravenous or oral cyclophosphamide (CTX). 400 mg CTX was administered intravenously every two weeks (Pulse CTX) or by daily oral dose of 50–100 mg according to body weight (Oral CTX). Oral corticosteroid+CTX or corticosteroid+mycophenolate (0.5-2 g/d) was administered for more than 1 year for maintenance therapy. Oral prednisone was prescribed at an initial dosage of 1 mg/kg/day for 1 month, with reducing doses over time. No patient received daily oral azathioprine (AZA) treatment because AZA was not available in our hospital.

In our study, all patients were dialysis dependent on admission. After standardized treatment, 25 patients obtained dialysis independence within 3 months. Among the 25 patients who got dialysis independence, 13 patients obtained complete renal recovery (CR, eGFR> = 60 mL/min/1.73m²) and the other 12 patients acquired partial renal recovery (PR, 15 mL/min/1.73m² < eGFR< 60 mL/min/1.73m²). As shown in Table 1, there were no differences of the clinical laboratory characteristics, with the exception of the platelet counts, among the 3 groups. Patients with renal recovery had higher platelet counts (CR: 326.3 ± 160.4 10⁹/L, PR: 269.3 ± 137.1 10⁹/L) than patients who remained dialysis dependent (201.2 ± 66.4 10⁹/L, P = 0.040). There was a trend of higher hemoglobin in patients with renal recovery (CR: 8.5 ± 1.7 g/L, PR: 7.9 ± 1.0 g/L) compared to those who remained dialysis dependent (7.3 ± 1.5 g/L), but the difference was not statistically significant (P = 0.095).

All patients received kidney biopsy, and all biopsy specimens contained more than 10 glomeruli. As shown in Table 2, the percentages of normal glomeruli and cellular crescent of patients with CR or PR were significantly higher than the patients who remained dialysis dependent (P = 0.007 and P = 0.015), while the percentages of fibrous crescent and global glomerulosclerosis of patients with CR or PR were significantly lower than those of patients who remained dialysis dependent (both P < 0.001). All patients with PR or patients who remained dialysis dependent had arterial wall thickening, while 38.5% of patients with CR did not have arterial wall thickening (P = 0.003).

The distributions of 4 pathological categories were investigated. No sclerotic class was found among the 40 patients. Only 2 patients were diagnosed as focal class, and both obtained renal recovery within 3 months. Among 18 patients in the crescentic class, 15 acquired renal recovery within 3 months and the other 3 stayed dialysis dependent. Among the 20 patients in the mixed class, 8 got renal recovery and the other 12 kept dialysis dependent. The renal recovery rate of the mixed class was significantly lower than that of crescentic class (p = 0.006) (Fig. 2a).

We compared the eGFR at admission, at 3 months, and at 1 year among patients in the different pathological categories. There was no difference in baseline eGFR among patients with focal, crescentic or mixed class. After 3 months of treatment, the eGFR of all 3 groups increased, but the eGFR of patients in the mixed class was lower than that of patients in the crescentic or focal classes. After following up for 1 year, 4 out of 18 patients in the crescentic class died, while 3 out of 20 patients in the mixed class died. Among the living patients, the eGFR of mixed class patients was still lower than that of the crescentic and focal classes (Fig. 2b).

Since 12 patients who recovered renal function obtained PR but not CR, the renal survival of these patients at 1 year was analyzed. We observed that 3 patients with mixed class entered ESRD, and these 3 patients had higher a proportion of global glomerulosclerosis than patients who did not enter ESRD (Table 3).

We then compared the combined end point (Death and ESRD) among patients with different pathological categories. As shown in Fig. 2c, the order of the survival was focal, crescentic and mixed classes (p = 0.009).

ROC curves were calculated to explore whether there were pathological parameters which could predict the renal prognosis. The ROC curves are shown in Fig. 3a. The AUC of global glomerulosclerosis was the greatest (0.853) among all 4 glomerular parameters. Since fibrous crescent usually indicates a kind of irreversible lesion, we attempted to determine whether the combination of fibrous crescent and global glomerulosclerosis could increase the accuracy of the diagnosis. As shown in Fig. 3b, the AUC of fibrous crescent+global glomerulosclerosis was greater than any other parameter. When the percentage of fibrous crescent+global glomerulosclerosis was 32.6%, the Youden index was at the peak value (sensitivity was 93.3% and specificity was 88.0%). When the percentage of fibrous crescent+global glomerulosclerosis exceeded 47.9%, renal recovery was unlikely.

Increased peripheral platelet counts were associated with reversible pathological lesions and high renal recovery rate.

Since patients who obtained renal recovery had higher platelet counts than patients who were dialysis dependent, we explored whether platelet counts could be used as a substitute for pathological lesions to determine kidney outcome. Correlation analysis indicated that platelet counts negatively correlated with the percentage of fibrous crescent+global glomerulosclerosis (Fig. 4a). There were no differences in the distributions of focal, crescentic and mixed categories between patients with normal platelets and patients with increased platelets (Increased platelets meant the platelet count was larger than the upper limit of normal range 300 × 10⁹/L) (Fig. 4b, p = 0.238), but for the majority patients with increased platelets, the percentages of fibrous crescent+global glomerulosclerosis were less than 32.6% (Fig. 4c, p = 0.002, compared with patients with normal platelets). Correspondingly, most patients with increased platelets obtained renal recovery (CR + PR) within 3 months (Fig. 4d, p = 0.045, compared with patients with normal platelets). However, the AUC of platelet counts for distinguishing renal recovery from dialysis dependence was only 0.709.

Effects of different therapy regimens on renal recovery were analyzed. All 40 patients received daily oral corticosteroid as a primary treatment. Among these patients, 7 received MP + pulse CTX treatment, 19 patients received MP + oral CTX treatment, 3 patients received PE + pulse CTX treatment and 11 patients received PE + oral CTX treatment. As shown in Table 4, there were no significant differences in the baseline BVAS, eGFR, proteinuria, or incidence of gross hematuria between the MP and PE groups or between pulse CTX and oral CTX groups. There were also no significant differences in kidney histopathology parameters among groups (data not shown). Compared with MP, PE accelerated the rate of renal recovery (29.4 ± 15.6 vs. 41.4 ± 11.7 days, p = 0.039), and there was a tendency for a greater renal recovery rate in PE group (79%) compared with MP group (53%).