Using MEST-C Scores and the International Study of Kidney Disease in Children Classification to Predict Outcomes of Henoch–Schönlein Purpura Nephritis in Children

Introduction: Henoch–Schönlein purpura nephritis (HSPN) and IgA nephropathy (IgAN) bear similarities in some aspects. The histological classification of HSPN was built on the International Study of Kidney Disease in Children (ISKDC) criteria, while IgAN was established on the 2016 Oxford classification (MEST-C scores). The purpose of this paper was to discuss the predictive value of the ISKDC classification and MEST-C scores in children with HSPN. Methods: We performed a retrospective study of 877 children with HSPN in a single center between 2001 and 2019. The primary outcome was defined as chronic kidney disease—estimated glomerular filtration rate (eGFR) <90 ml/min/1.73 m2. Results: During the follow-up period of 23.3 (10.9–47.9) months, 51 (5.8%) patients reached the primary outcome. As revealed in a Kaplan–Meier plot, segmental glomerulosclerosis (S) (P < 0.001) and tubular atrophy/interstitial fibrosis (T) (P < 0.001) significantly predict poor renal outcome. Other Oxford lesions and the ISKDC classification, however, did not show a significant difference in a worse outcome. In a multivariate Cox model adjusted for pathological and clinical factors, eGFR [hazard ratio (HR) = 2.831, 95% confidence interval (95% CI) = 1.359–5.896], S lesion (HR = 3.936, 95% CI = 2.078–7.457), and T lesion (HR = 4.002, 95% CI = 1.733–9.242) were independent risk factors for the renal outcome. Conclusion: This series constitutes the largest series reported so far in the literature of such patients. According to our findings, S and T of the Oxford classification, which are ignored by the ISKDC classification, could be applied to predict the renal prognosis of children with HSPN.


INTRODUCTION
Henoch-Schönlein purpura (HSP), also referred to as IgA vasculitis, is the most common systemic small vasculitis in childhood. It is estimated that 6 to 24 per 100,000 children under the age of 17 will develop HSP (1), which is affected by ethnic background (1). In Asia, in the age group of 4 and 6 years, the estimated annual incidence was the highest (70.3 per 100,000) (1). About 20-60% of children with HSP had renal manifestations, and 1% of these patients developed end-stage renal failure (ESRD) (2). At present, we use the International Study of Kidney Disease in Children (ISKDC) criteria for the histological classification of Henoch-Schönlein purpura nephritis (HSPN) in children, which is primarily based on the number of crescents (3)  original four histopathologic features related to renal outcomes to five: mesangial hypercellularity (M), endocapillary proliferation (E), segmental sclerosis/adhesion (S), tubular atrophy/interstitial fibrosis (T), and cellular or fibrocellular crescents (C) (4,5). Currently, several studies have applied the updated Oxford classification to children with HSPN for predicting renal prognosis ( Table 5) (6)(7)(8)(9)(10). In this paper, we use the Oxford classification and the ISKDC classification to predict the outcomes of 877 children with HSPN.

Patients
The retrospective survey was conducted in Jinling Hospital in the Department of Pediatric Nephrology and Rheumatology. The inclusion criteria were as follows: (1) proteinuria, hematuria, and/or renal failure associated with purpura with/without joint or abdominal pain; (2) age at renal biopsy <18 years; (3) all patients performed a renal biopsy and were histologically confirmed to have IgA deposition; and (4) at least one clinical follow-up. Exclusion criteria were as follows: (1) lack of follow-up data; (2) <10 glomeruli on biopsy; and (3) patients with hepatitis B virus infection, lupus nephritis, Alport syndrome, and other primary or secondary glomerulonephritis should be excluded. All renal biopsies were performed with the consent of the children's parents. In total, 877 children were engaged in the study.

Clinical and Biological Data
All biological and clinical data of patients were collected at biopsy. General clinical data were collected, including the age at the onset of purpura, the duration from the onset of purpura to renal involvement, the duration between the onset of renal involvement and renal biopsy, gender, and follow-up time. Extrarenal manifestations, renal involvement, the presence of hypertension, and treatment were recorded. Urine and blood samples were taken from the patients at biopsy for a routine test, including hematuria, 24-h urinary protein, serum creatinine, estimated glomerular filtration rate (eGFR), serum albumin, serum uric acid, hemoglobin, and cholesterol.

Definition
The definition of primary outcome was chronic kidney disease (CKD). We defined CKD as eGFR <90 ml/min/1.73 m 2 . We calculated renal survival time from the renal biopsy to the last follow-up. eGFR was calculated with the updated Schwartz formula (11). If patients were aged over 16 years, we used the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation to calculate eGFR (12). We defined hypertension as diastolic or systolic blood pressure ≥95 th percentile for age, sex, and height (13). Hematuria was defined as macroscopic hematuria or hematuria with ≥3 red blood cells/HPF 3 times within 1 week. The definition of proteinuria was 24 h proteinuria >0.4 g. Hypercholesterolemia and hypoproteinemia were defined as cholesterol >5.7 mmol/L and serum albumin <25 g/L, respectively. Hyperuricemia was defined as more than 360 mmol/L for girls and more than 420 mmol/L for boys (14). The definition of ESRD was eGFR <15 ml/min/1.72 m 2 or requiring maintenance of renal replacement therapy for more than 3 months.

Histological Data
Histological data (light microscopy and immunofluorescence) of the renal biopsy were recorded at the time of diagnosis. Two independent nephrologists evaluated jointly renal pathology by using the ISKDC classification (3) and the 2016 Oxford classification (MEST-C scores) (4,5). If the two nephrologists scored differently, a third nephrologist would rescore the target and the results would depend on the third nephrologist.

Statistical Analysis
As for the statistical methods, we used SPSS 26.0 for Windows to analyze the data. The data of normally distributed and non-normal distribution were described as mean ± standard deviation and median (interquartile range), respectively. Their  comparisons were based on Student's t-test and the Mann-Whitney U-test. We used the Kruskal-Wallis test for the comparison of multiple non-normal distributions. Categorical variables were described as percentages and were compared by using the Fisher's exact test or Pearson chi-square test. Patients were divided into the following groups: ISKDC (grade I + II/III + IV + V), M0/1, E0/1, S0/1, T0/1 + 2, and C0/1/2. Their comparisons were analyzed with the Kaplan-Meier method and the log-rank test. We applied the Cox regression model to multivariate analyses. Pathological and clinical factors related to primary outcome reported in previous literature were included in the model. Variables with P < 0.05 in the univariate analysis were included in the multivariate analysis. These results were described as hazard ratios (HRs) with 95% confidence intervals (95% CIs). Two-sided P values < 0.05 were regarded as statistically significant.

Clinical and Biological Data of Children With HSPN at Biopsy
The biological and clinical data of children with HSPN at biopsy are shown in Table 1 All patients were divided into outcome and nooutcome groups according to renal prognosis. Compared with the no-outcome group, the duration from the onset of purpura to renal involvement was shorter, and the duration between the onset of renal involvement and renal biopsy and follow-up time were longer in the outcome group. The number of patients with hypertension in the outcome group is bigger than that in the nooutcome group. However, there were no significant differences in other clinical characteristics between the two groups.

Renal Pathological Findings of Children With HSPN at Biopsy
The median number of glomerulus at renal biopsy was 24.0 (17.0-32.0). The renal pathological findings are presented in .1%), respectively. No significant differences in M, E, and C were observed between the outcome and the no-outcome groups, except for more S and T in the former. There were differences in the composition of the ISKDC classification between the two groups.

Comparison of Clinical Characteristics at Renal Biopsy of Two Pathological Classifications in Children With HSPN
A comparison of the clinical characteristics at the renal biopsy of two pathological classifications in children with HSPN is summarized in Table 3. Compared with ISKDC I + II, patients had more proteinuria (P < 0.001), lower eGFR level (P < 0.001), and higher serum creatinine (Scr) level (P < 0.001) in ISKDC III + IV + V. The onset age of the ISKDC III + IV + V group was higher than that in the ISKDC I + II group (P = 0.021). Based on the updated Oxford classification, there was no significant difference in onset age between the M0 and M1 groups, but patients in the M1 group had more proteinuria (P < 0.001), lower eGFR level (P < 0.001), and higher Scr level (P < 0.001). There was no significant difference in Scr level, onset age, and eGFR level between the E0 and the E1 groups, but patients in the E1 group had more proteinuria than those in the E0 group (P < 0.001). Patients in the S1 group had a lower eGFR level (P = 0.026) and a higher Scr level (P = 0.001) than those in the S0 group. There was no significant difference in proteinuria and onset age between the S0 and the S1 groups. Patients in the T1/T2 group had a lower eGFR level (P < 0.001), higher Scr level (P < 0.001), higher proteinuria (P = 0.021), and higher onset age (P = 0.003) than those in the T0 group. Proteinuria (P < 0.001), Scr level (P < 0.001), and eGFR level (P < 0.001) were different among the patients in the C0 group, C1 group, and C2 group. There was no difference in onset age among the three groups. A Kaplan-Meier plot revealed that event-free renal survival was significantly shorter in patients with S1 and T1/T2 according to MEST-C scores than in those with S0 (P < 0.001) and T0 (P < 0.001) (Figure 1). However, other Oxford lesions and the ISKDC classification did not show a significant difference in event-free renal survival (Figure 1).

Univariate and Multivariate Cox Regression Analyses of Factors Associated With Renal Outcome
In Table 4, we used the univariate Cox regression analysis and found that sex (P = 0.047), hypertension (P < 0.001), eGFR (P <

DISCUSSION
The proportion of crescents is the main basis of the ISKDC classification which has been criticized as a too rough and insensitive method. Given many similarities between HSPN and IgAN in clinical features, histology, IgA immunological abnormalities, and coagulation abnormalities (15), using the Oxford classification of IgAN to predict the long-term outcomes of adults with HSPN was firstly published in Korea in 2014 (16). Xu et al. firstly applied the Oxford classification to the analysis of children with HSPN in 2017 (8), and there have been five articles concerning the Oxford classification to predict the prognosis of children with HSPN up to date ( Table 5) (6-10). We analyzed retrospectively the data from 877 patients included in a Chinese single-center study, which constitutes the largest series of children reported so far in the literature of children with HSPN. In our study, only 51 people reached the primary outcome. It may be related to the setting of endpoint events, active treatment, or age of the included subjects. It was proven by previous studies that clinical presentation and renal outcomes in patients with HSPN may be more severe in adults than in children (17), with an estimated 25 to 30% risk of developing CKD (18). However, ∼1-7% of children with HSPN progress to renal failure (1). Substantial proteinuria is a significant factor for the progression to ESRD (19). In our study, 204 patients had proteinuria at the last follow-up, and they needed a longer follow-up to observe the progress of the disease.
We found that eGFR <90 ml/min/1.73 m 2 at renal biopsy was an independent risk factor to affect the renal outcome in our study, indicating that the basic level of clinical data of patients could predict prognosis. Previous studies have shown that HSPN patients with severe clinical presentation had a poor prognosis. Coppo et al. (20) and Pillebout et al. (21) found that in adults with HSPN at presentation, renal function impairment was a negative prognostic factor. Some studies about children with HSPN did not include the role of eGFR in the multivariate regression model (7,9), while others did not demonstrate that the decline of eGFR was a risk factor for prognosis (20).
In some researches concerning the Oxford classification to predict the prognosis of HSPN with adults (16,22), E had an unfavorable influence on renal survival. However, most of the studies on children with HSPN (including our study) found that E was not related to prognosis (6-9, 23, 24). On one hand, it may indicate the difference in pathology and prognosis between adults and children with HSPN. On the other hand, several studies of IgAN have found that E1 was a risk factor for renal prognosis in patients who did not receive immunosuppressive therapy (4,25). It is also important to note that some patients had already received immunosuppression before renal biopsy in our study, which may be related to immunosuppressionassociated bias.
We found that T accounted for a certain proportion in the Oxford classification and the Hass classification of IgAN (4,26), the ISN/RPS 2003 classification of lupus nephritis (27), and the modified semiquantitative classification (SQC) of HSPN (28), indicating the importance of T. T, which is ignored in the ISKDC classification of HSPN, may have important prognostic significance. Kim et al. (16) used the Oxford classification to predict prognosis in adults with HSPN and indicated that T had lower renal survival rates than those with T0, and T1/T2 (HR  (6), and our study concluded that T could be used to assess renal outcomes of HSPN. T1/T2 accounted for only 3.0% in our study. We hypothesize that it may be related to age. First, the onset age of T1/T2 was higher than in the T0 group in our group. In studies of adults with HSPN, the proportion of T1/T2 fluctuated between 11 and 54.1% (6,16,22), while that in children with HSPN was only 5.3-18.8% (6)(7)(8)24). Second, the Oxford classification is primarily derived from the data of adult IgAN patients, and thus, it may need to be modified when evaluating pediatric HSPN. Similar to the conclusions of  9), we used the Oxford classification to predict the prognosis of HSPN patients and found that S was a risk factor for renal outcome. S and T are chronic changes, which currently available treatments do not affect (8). Besides, these chronic changes are signs of irreversible damage and are more likely to be related to a worse prognosis (29).
The ISKDC classification is based primarily on the degree of crescent formation to predict renal prognosis. However, there is a lack of consensus regarding the value of crescents as a long-term predictor. In our study, C was not significantly related to poor renal prognosis. The point shared some similarities with those done by Kim (8). Unlike the ISKDC classification, the Oxford classification accounts not only for acute lesions like M, E, and C but also chronic changes such as S and T. Acute lesions can guide the active use of immunosuppressive agents, while chronic lesions can predict prognosis.
This series constitutes the largest series reported so far in the literature of children with HSPN. The working group does not recommend the use of the Oxford classification in HSPN since patients were not included in the validation cohort. Our study provides sufficient and valid data to demonstrate that the application of the Oxford classification is effective in the prognosis of children with HSPN, but the study has some limitations. Firstly, the study is a single-center retrospective study. The incidence of HSPN in children is more common in Asian countries than in other countries due to ethnic background (1). Therefore, prospective and multinational cohort studies are needed. Secondly, the follow-up period is relatively short, so a small number of children reach the primary outcome. 204 patients had proteinuria by the last follow-up, and they needed a longer follow-up. Thirdly, treatment has an effect on prognosis. But due to the lack of established guidelines and changes in treatment during follow-up, we did not include treatment in the multivariate regression model. Fourth, immunosuppressive therapy before renal biopsy may also be a confounding factor. These factors need to be considered in future studies concerning the Oxford classification to predict renal outcomes in patients with HSPN.

CONCLUSIONS
This series constitutes the largest series reported so far in the literature of such patients. According to our findings, S and T of the Oxford classification, which are ignored by the ISKDC classification, could be applied to predict renal prognosis of children with HSPN.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by 2019NZGKJ-168. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
MW, RW, QK, JY, XF, ZS, HW, and YP conducted data acquisition and interpretation, and participated in the drafting of the manuscript. ZX, CG, and XH participated in the interpretation of the data, the evaluation of renal pathology, and the critical revision of the manuscript of important academic content. All authors participated in data acquisition, interpretation, and statistical analysis. All authors read and approved the final manuscript.