Association of kidney biopsy findings with short- and medium-term outcomes in children with moderate-to-severe IgA vasculitis nephritis

Assessing the initial severity of immunoglobulin A vasculitis nephritis (IgAV-N) is important due to its determining effect on kidney management and outcomes. This paper describes a multicentre paediatric cohort of IgAV-N patients and discusses relationships among clinical presentation, histological features, and kidney outcome. We retrospectively studied a cohort of 170 children with biopsy-proven IgAV-N, diagnosed between 2007 and 2017. One-quarter of the cohort (27%) presented with initial nephrotic syndrome (NS). Kidney biopsy revealed International Study of Kidney Disease (ISKDC) grade II or grade III in 83% of cases. Endocapillary proliferation was observed in 73% of patients, and chronic lesions were observed in 25%. Data analysis showed a significant association between NS at onset and endocapillary proliferation and cellular crescents. After a median follow-up of 21 months (IQR 12–39), 30% of patients had persistent proteinuria or decreased eGFR. At the end of follow-up, kidney impairment was more often observed in patients with NS at onset and those with cellular crescents and chronic lesions on initial kidney biopsy. Conclusion: This study highlights the relationship between the clinical and histological presentation of IgAV-N and the factors that affect kidney outcome. The ISKDC classification may be improved by including lesions that are more discriminating for disease severity and prognosis. What is Known: • Nephrotic syndrome (NS) or kidney failure at diagnosis and cellular crescents in more than 50% of the glomeruli are recognized as risk factors for poor kidney outcome in immunoglobulin A vasculitis nephritis (IgAV-N). • The reference histological classification of the International Study of Kidney Disease in Children (ISKDC) is primarily based on the presence and number of affected glomeruli (mesangial proliferation, cellular crescents). The updated Oxford classification, which emphasizes tubular atrophy and interstitial fibrosis, is also used to group pathological features of IgAV-N. Both classifications have limitations. What is New: • Medical treatment should not be postponed in patients with IgAV-N and NS until after biopsy, as NS at diagnosis is associated with initial histological severity and poorer kidney outcome. This proposal needs to be verified in further studies. • Endocapillary proliferation is associated with the initial severity of IgAV-N at diagnosis, while chronic glomerular changes and interstitial fibrosis are associated with poorer short- and medium-term kidney outcomes. What is Known: • Nephrotic syndrome (NS) or kidney failure at diagnosis and cellular crescents in more than 50% of the glomeruli are recognized as risk factors for poor kidney outcome in immunoglobulin A vasculitis nephritis (IgAV-N). • The reference histological classification of the International Study of Kidney Disease in Children (ISKDC) is primarily based on the presence and number of affected glomeruli (mesangial proliferation, cellular crescents). The updated Oxford classification, which emphasizes tubular atrophy and interstitial fibrosis, is also used to group pathological features of IgAV-N. Both classifications have limitations. What is New: • Medical treatment should not be postponed in patients with IgAV-N and NS until after biopsy, as NS at diagnosis is associated with initial histological severity and poorer kidney outcome. This proposal needs to be verified in further studies. • Endocapillary proliferation is associated with the initial severity of IgAV-N at diagnosis, while chronic glomerular changes and interstitial fibrosis are associated with poorer short- and medium-term kidney outcomes.

Introduction but kidney impairment is the main cause of morbidity and mortality and the main determinant of long-term outcome [2]. IgAV nephritis (IgAV-N) occurs in approximately onethird of cases. In the literature, about 20% of IgAV-N patients develop nephrotic syndrome (NS) at diagnosis, occasionally associated with acute kidney injury [3][4][5]. The risk of progression to chronic kidney disease (CKD) has been reported to range from 1 to 36% [6][7][8][9][10]. This wide variability can be explained by selection bias (such as the inclusion of patients with more-or-less severe IgAV-N depending on the type of treatment centre, secondary paediatric care, or tertiary referral centre), differences in the definition of CKD, or the follow-up duration.
The clinical severity (NS, acute kidney injury) or histological features (cellular crescents in more than 50% of the glomeruli) of IgAV-N at disease onset have been considered to be predictive of poor kidney outcome in children [7,8,11,12]. However, in many other studies, these histological features are not associated with IgAV-N prognosis [8,10,11,13,14]. To date, the literature findings on the association between a specific histological lesion and the IgAV-N prognosis are conflicting.
Kidney biopsy remains the gold standard in assessing the severity of IgAV-N with the reference histological classification of IgAV-N by the International Study of Kidney Disease in Children (ISKDC), dating back to 1977 [15]. The ISKDC classification is based mainly on the presence and number of affected glomeruli (mesangial proliferation, cellular crescents). In addition, the Oxford classification of IgA nephropathy was developed as a classification of biopsy findings in adults with IgA-N. This is a more extensive classification than the ISKDC classification and incorporates not only mesangial hypercellularity but also endocapillary proliferation, glomerulosclerosis, and interstitial fibrosis/tubular atrophy [16]. IgAV-N and IgA nephropathy have previously been considered the two different manifestations of a single disease, with many similarities. The Oxford classification has therefore been applied for reviewing biopsies of adults and children with IgAV-N [17].
In 6 paediatric nephrology units in Southern France, we have shared the same diagnostic algorithm and treatment protocol for IgAV-N for the past 10 years. In evaluating our kidney biopsy practices, this retrospective analysis aims to describe the presentation of biopsy-proven IgAV-N in a paediatric cohort and the relationship between clinical and histological features at diagnosis and during short-and mediumterm follow-ups of kidney disease.

Study design
We performed a retrospective multicentre analysis of a historic cohort of children with IgAV-N in paediatric nephrology units of Bordeaux, Limoges, Marseille, Montpellier, Nice, and Toulouse from 2007 to 2017. All patients who were 1 to 18 years of age at the time of IgAV onset with a clinical diagnosis of IgAV according to the European League Against Rheumatism (EULAR) classification [18] and biopsy-proven IgAV-N were considered for inclusion. The patients were identified through pathology databases of the participating centres. The management of patients was based on a common protocol implemented in 2007 (Supplementary data S1). The study was approved by the ethics committee of the French Society of Paediatrics (CER_SFP_2018_079-3).

Data collection
All children were examined by general practitioners or referred to local hospitals at the onset of purpura and extrarenal symptoms. The children were then systematically referred to paediatric nephrology units a few days after onset of IgAV-N. All kidney biopsies were performed in trained paediatric nephrology centres and were performed according to the same pre-established protocol (see Supplementary data S1). The indications for the first kidney biopsies included uncertainty about the aetiology of the nephritis, moderate IgAV-N of undetermined onset, severe IgAV-N, rapidly progressive glomerulonephritis, and/or inefficacy of a well-conducted treatment after 1 month.
The following clinical and biochemical data, obtained at time of kidney biopsy (or in the preceding week), were retrospectively collected from hospitalization files: extrarenal symptoms, blood pressure, kidney function, serum albumin, and proteinuria. Hypertension was defined as a systolic or diastolic blood pressure ≥ the 95th percentile for age, sex, and height [19]. Gastrointestinal involvement was defined as severe in cases of intussusception, gastrointestinal bleeding, or the need for enteral and/or parenteral nutrition support. The glomerular filtration rate (eGFR) was estimated using the bedside Schwartz formula [20]. Decreased kidney function was defined as an estimated GFR below 90 mL/min per 1.73 m 2 [21]. Proteinuria was measured in spot urine samples prior to kidney biopsy. Highest protein content measurements (expressed as g/L) were recorded. Protein-to-creatinine ratio was expressed as mg/mmol. NS was defined in accordance with France's national health agency (HAS) [22] as serum albumin < 30 g/L and nephrotic-range proteinuria (proteinto-creatinine ratio > 200 mg/mmol or 24-h urinary protein > 50 mg/kg).
All kidney biopsy reports were retrospectively reviewed by two experienced kidney pathologists and graded according to the ISKDC (Supplementary data S2) [15]. Acute lesions (endocapillary proliferation and cellular crescents) were graded as < 50% or > 50% of the glomeruli. Endocapillary proliferation was defined by the presence of hypercellularity internal to the capillary wall with endothelial cell swelling, subendothelial expansion, and inflammatory cells, clearly leading to luminal narrowing. Interstitial fibrosis, glomerular sclerosis, and fibrous crescents were defined as chronic lesions. A second kidney biopsy was performed in cases of insufficient treatment efficacy at 1 month (less than a 50% reduction in the protein-to-creatinine ratio) or persistence of a protein-to-creatinine ratio > 100 mg/mmol at 2 months after the onset of treatment.
The treatments used were oral corticosteroids, methylprednisolone pulse therapy followed by oral corticosteroids, or other immunosuppressive therapy, and angiotensin-converting enzyme inhibitors (ACEis) or angiotensin receptor blockers (ARBs). Patients with ISKDC grade ≥ III were usually treated with methylprednisolone pulses followed by oral corticosteroid or other immunosuppressive therapy. Patients with ISKDC grade I were treated with only ACEis or ARBs, and ISKDC grade II was treated with oral corticosteroids most often (see Supplementary data S1).
We collected follow-up data until the last documented outpatient visit. IgAV-N remission was defined as negative proteinuria (protein-to-creatinine ratio < 20 mg/mmol) and normal kidney function (eGFR ≥ 90 mL/min per 1.73 m 2 ). Kidney outcomes at the last follow-up were classified as follows: (A) normal (no hypertension, normal proteinuria, and kidney function), (B) mild-to-moderate urinary abnormalities (protein-to-creatinine ratio between 20 and 200 mg/mmol), and (C) more severe kidney impairment (protein-to-creatinine ratio > 200 mg/mmol and/or reduced eGFR < 90 mL/min per 1.73 m 2 ).

Statistical analyses
Quantitative variables are expressed as medians and interquartile ranges (IQRs), and qualitative variables are expressed as counts and percentages. Cumulative incidence was calculated using the Kaplan-Meier method. Qualitative variables were compared using chi-square or Fisher tests, and quantitative variables were compared using Student's or Mann-Whitney tests.
The relationship between the protein-to-creatinine ratio during follow-up and several baseline factors was evaluated by linear regression analysis. Based on prior knowledge and the literature, sex, age, gastrointestinal involvement, NS status, histological severity (ISKDC grade, endocapillary proliferation, cellular crescents, and interstitial fibrosis), time from IgAV-N diagnosis to corticosteroid therapy initiation, and treatments received were included in the model. Statistical significance was defined as p < 0.05. The statistical analyses were performed using SPSS version 20.0 (SPSS Inc., Chicago, IL, USA).

Patient characteristics
Patient characteristics at baseline are reported in Table 1. A total of 170 patients (101 males) with IgAV-N were included. The median age at diagnosis was 7.8 years (IQR 5.7-10.9). The median time from IgAV onset to IgAV-N diagnosis was 17 days (IQR 10-34).
The median time from IgAV-N diagnosis to kidney biopsy was 17 days (IQR 10-39). The kidney biopsy reports were reviewed by two pathologists (no full review of biopsies) and were classified as ISKDC grade I in 23 patients (14%), grade II in 73 patients (43%), grade III in 68 patients (40%) including 44 patients with grade IIIA and 24 with grade IIIB, grade IV in two patients (1.2%), and grade VI in four patients (2.4%). Endocapillary proliferation was described in 125 patients, 87 patients with endocapillary proliferation observed in < 50% of the glomeruli and 38 patients with endocapillary proliferation in > 50% of the glomeruli. Chronic lesions were present in 42 patients (25%). Specifically, interstitial fibrosis was present in 22 biopsies (13%), glomerular sclerosis in 15 biopsies (8.8%) with a median number of 5.5% (IQR 3.7-9.7) per biopsy, and fibrous crescents in 15 biopsies (8.8%) with a median number of 6.3% (IQR 4.8-5.7) per biopsy. In patients with chronic lesions, the median time from IgAV-N diagnosis to kidney biopsy was 30 days (IQR 12-59) vs 17 days (IQR 9-32) in patients without chronic lesions (p < 0.01). The median follow-up duration was 21 months (IQR 12-39), which was similar for patients with and without chronic lesions.
One patient with a severe clinical and histological presentation was treated by plasma exchange (10 sessions). At 1 month from IgAV-N diagnosis, 5 patients received two methylprednisolone pulses. Three of them had already received methylprednisolone pulses at diagnosis of IgAV-N.
In the multivariate analysis, interstitial fibrosis on initial kidney biopsy was associated with the protein-to-creatinine ratio at 3 months (β = +178, p = 0.008). In addition, the time between IgAV-N diagnosis and the introduction of corticosteroids was significantly associated with the protein-tocreatinine ratio at the 6-month follow-up (β = +0.1, p < 0.0001); the shorter the time between IgAV-N diagnosis and the introduction of corticosteroids, the lower was the proteinto-creatinine ratio at 6 months (Supplementary data S3).

Outcome at last follow-up
The median follow-up was 21 months, with 30% of patients being followed for more than 3 years. At the last follow-up, two-thirds of patients (115/170, 68%) had no kidney sequelae (no residual proteinuria or decreased kidney function, i.e. group A). Of these 115 patients, 105 had been treated with oral corticosteroids, of whom 69 (60%) also received methylprednisolone pulses, and 10 (9%) received additional immunosuppressive therapy. Thirty percent of patients (51/170) had moderate urinary abnormalities (group B). Of these 51 patients, only five received ACEis, and 13 patients received several treatments (ACEis, corticosteroids, and immunosuppressants). Only 2 patients had more severe kidney impairment (group C). Both had severe IgAV-N with NS and an ISKDC grade IIIB kidney biopsy at IgAV-N onset. Two patients were lost to follow-up immediately after the diagnosis of IgAV-N.
At the last follow-up, the kidney outcome was significantly associated with confirmed NS at IgAV-N diagnosis (p = 0.045) and the presence of cellular crescents (p = 0.027) and chronic lesions such as glomerular sclerosis (p = 0.046) and fibrous crescents (p = 0.046) in the kidney biopsy (Table 4).

Discussion
This study retrospectively examined a large paediatric cohort of patients with biopsy-proven IgAV-N and identified associations between clinical and histological presentations and kidney prognosis. The main findings are that endocapillary proliferation on initial kidney biopsy was associated with the severity of IgAV-N at diagnosis and that chronic glomerular changes and interstitial fibrosis on initial kidney biopsy were associated with a poorer medium-term kidney outcome. Approximately one-third of patients had persistent kidney disease (proteinuria in 30% of patients or decreased eGFR in 1% of patients) after a median follow-up of 21 months. The demographic characteristics of our population are similar to those in other studies, with a predominance of male patients and a median onset age of 8 years [11,13,23,24]. In accordance with previous studies, kidney involvement occurred within 6 months following IgAV diagnosis in almost all our patients (96%) [4,23,25]. In our study, kidney biopsy was usually performed quickly after the diagnosis, with a median interval of 17 days. Acute histological lesions (mostly endocapillary proliferation and, less frequently, cellular crescents) were described in most of the biopsies. In contrast, chronic lesions (glomerular and tubular) were less common. In the literature, endocapillary lesions are rarely documented in IgAV-N and may be reversible by corticosteroids and immunosuppressive therapy in IgA vasculitis [26]. In our study, endocapillary proliferation was the only acute lesion associated with the level of proteinuria at diagnosis in the multivariate analysis and therefore reflected the initial clinical severity. This association is not always mentioned in published studies. Recently, a French study by Delbet et al. did not find any association between endocapillary proliferation and kidney prognosis [27]. Conversely, cellular crescents have been identified as poor prognostic factors in some studies [7,12,[28][29][30], especially if crescents are present in more than 50% of glomeruli [31]. In our study, crescents in the initial biopsy were more predictive of the longer-term kidney outcome than endocapillary proliferation was. In addition to acute lesions, some 25% of biopsies showed chronic histological lesions (glomerular changes and interstitial fibrosis). In patients with chronic lesions, the median time from IgAV-N diagnosis to kidney biopsy was twice as long as that in patients without chronic lesions (30 vs 17 days). These results were similar to those in the studies by Hennies et al. [23] and Delbet et al. [27], which reported a longer delay between nephritis and kidney biopsy and a higher proportion of chronic lesions among children with worse outcomes. We also found that the presence of chronic lesions was significantly associated with a poorer kidney outcome, as other studies did [32]. Therefore, the identification of chronic lesions on initial biopsies may reflect the duration of IgAV-N, regardless of clinical manifestations, suggesting a late diagnosis or indicating a silent period with kidney damage in IgAV-N before clinical symptoms. Additionally, some patients may develop chronic lesions more rapidly than others due to their genetic background favouring fibrotic healing from an inflammatory state. According to these results, we suggest that endocapillary proliferation and chronic lesions be analysed in IgAV-N. In our cohort, 25% of patients with grade II ISKDC had persistent kidney disease at the last follow-up, despite receiving corticosteroid therapy. Coppo et al. found CKD in 23% of patients (adults and children) with grade II ISKDC after a follow-up of 4.8 years and stated that IgAV-N should not be considered a more benign condition in children than in adults [14]. Additionally, Delbet et al. found persistent proteinuria in 25% of these patients during the 1.5-year follow-up [25]. The prognosis of patients with mild histological lesions (no crescents but mesangial and/or endocapillary proliferation) may often also be unfavourable in the long term and should not be underestimated. Therefore, these histological lesions should be taken into account in a new classification. The ISKDC classification, the reference currently for IgAV-N, is of limited prognostic value. Other histological classifications or biomarkers need further evaluations to better define the histological severity and improve the risk stratification in IgAV-N. In adult nephrology, the classification of Pillebout et al. is used to identify focal and diffuse endocapillary proliferation as the most frequent lesions in IgAV-N [33]. Furthermore, given the pathophysiological and histological similarities between IgAV-N and IgA nephropathy and the limitations of the ISKDC classification, several authors have proposed using the Oxford classification in adults and children with IgAV-N [17,32,34]. The Oxford classification assesses kidney pathology in terms of mesangial cellularity, endocapillary hypercellularity, glomerular sclerosis, and interstitial fibrosis [16]. A significant association between the presence of endocapillary proliferation or interstitial fibrosis and longterm kidney prognosis has been reported using the Oxford classification [32]. However, extracapillary proliferation, considered a marker of severity in IgAV-N, is not considered in this classification.
On the other hand, clinical features as risk factors for poor kidney outcome remain controversial. Several studies have reported that the clinical severity at diagnosis of IgAV-N is predictive of kidney outcome [4,7,8,11,12,35,36]. Narchi et al. reported CKD in 20% of children with NS and only in 1.6% of children with isolated urinary abnormalities at  [4]. In contrast, Tudorache et al. found no association between initial clinical severity and long-term kidney prognosis [10]. In our cohort, the presence of NS at diagnosis was significantly associated with persistent kidney disease after 21 months. Furthermore, the presence of NS was also significantly associated with initial histological severity, with increased endocapillary proliferation and cellular crescents, associated with a poorer kidney outcome, as described above [7,12,28,29,31,37]. These data suggest that corticosteroid treatment in IgAV-N patients with NS should not be delayed and that the decision for treatment should rely more on clinical presentation than on histological findings. Moreover, these findings question the usefulness of a kidney biopsy in cases of severe clinical features of IgAV-N. Kidney biopsy may be performed as a second step to identify lesions associated with poor prognosis and to adapt treatment. In accordance with the literature [28,29,38,39] and because the 6-month protein-tocreatinine ratio was significantly associated with the delay of IgAV-N corticosteroid treatment, our analysis indirectly suggests that kidney outcomes may be improved by early treatment.
Our study has several limitations. These include the retrospective, multicentric, and uncontrolled design of the study and the selection bias of only including patients with biopsyproven IgAV-N. Patients with benign or moderate kidney impairment that improved spontaneously or with ACEi therapy were not included. There was likely an overrepresentation of severe cases of IgAV-N. All kidney biopsy reports were retrospectively reviewed by two experienced kidney pathologists, but there was no full review of slide biopsies. Moreover, the relatively short follow-up period of the study precludes an adequate assessment of long-term kidney outcomes.
In conclusion, in this retrospective multicentre cohort analysis, endocapillary proliferation on initial kidney biopsy was associated with the clinical severity of IgAV-N at diagnosis, and chronic lesions were associated with poorer short-and medium-term kidney outcomes. These lesions are not included in the ISKDC reference classification and should be considered to better delineate the disease severity in paediatric IgAV-N with a new histological classification. We suggest that corticosteroid treatment in IgAV-N patients with NS should not be delayed by kidney biopsy.