Thickening of Glomerular Basement Membrane Is Associated with Long-Term Prognosis in Patients with IgA Nephropathy

Visual Abstract

Abstract

Key Points

• IgA nephropathy (IgAN) patients with thickening of glomerular basement membrane had a higher risk of progressing to ESKD independent of international risk-prediction tool in IgAN.

• Thickening of glomerular basement membrane in IgAN patients was positively correlated with mesangial hyperplasia.

Background

Glomerular basement membrane (GBM) ultrastructural abnormalities are common in IgA nephropathy (IgAN); however, few studies have focused on clinical significance and prognostic value of GBM ultrastructural changes in IgAN patients.

Methods

A retrospective longitudinal cohort with 1006 biopsy-proven primary IgAN patients was collated. GBM thickness and texture of each case were assessed under transmission electron microscope. The primary end point was ESKD. Cox proportional hazards regression model was built to determine risk factors. Immunofluorescent staining was performed on patient kidney biopsy samples to validate the correlation between mesangial proliferation and abnormal GBM thickness. Twenty single nucleotide polymorphisms independently associated with IgAN in previous genome-wide association studies were genotyped in 617 patients, and whole exome sequencing was performed in 56 patients to investigate potential variants underlying GBM ultrastructural changes.

Results

Of 1006 patients, 52% were female, and the mean age was 37.3±12.3 years old. Among all patients, 80 (8%) had abnormal thickness of GBM including 29 (3%) thickening of GBM and 51 (5%) thinning of GBM. Abnormal GBM texture was found in 25 (2%) patients. During a mean follow-up time of 46.4 months, 91 (9%) patients progressed to ESKD. By Cox regression analyses, we demonstrated that thickening of GBM at biopsy increased the risk of ESKD before (hazard ratio [HR], 3.64; 95% confidence interval [CI], 1.47 to 7.55) and after adjusted by Oxford Scoring (HR, 2.92; 95% CI, 1.12 to 6.48) or international risk-prediction tool in IgAN (HR, 3.51; 95% CI, 1.41 to 7.29). Relevance analyses showed that GBM thickening was positively correlated with mesangial hyperplasia and proliferation, but not genetic variants in IgAN patients.

Conclusions

Thickening of GBM correlated with mesangial hyperplasia and proliferation was associated with ESKD in IgAN patients, demonstrating the potential of incorporating ultrastructural changes into the pathologic evaluation system of IgAN.

Introduction

IgA nephropathy (IgAN) is the most prevalent primary GN worldwide, with around 20%–40% IgAN patients progressing to ESKD. The disease is defined by the presence of IgA1-dominant or codominant glomerular deposits.¹ The formation of immune complex by galactose-deficient IgA1 and antiglycan antibodies is regarded as a key process of the immunopathogenesis of IgAN. Genetic and environment factors are related to susceptibility to IgAN and the risk of IgAN progression as well.^2–9

The clinical and pathologic manifestations of IgAN are highly diverse,^10–12 making it challenging to predict the prognosis of IgAN, especially at an early stage. Several pathologic classification systems have been established attempting to tell different IgAN subtypes apart. The most widely acknowledged pathologic classification method of IgAN is the Oxford classification, which scores IgAN histologic lesions based on evaluating the presence and severity of mesangial hypercellularity (M), endocapillary cellularity (E), segmental sclerosis (S), interstitial fibrosis/tubular atrophy (T), and crescent (C).^13,14 Among these histologic parameters, T was the most frequently validated score correlated with the prognosis, while other scores had different validation results by subsequent studies.^1,15–17 Besides pathologic changes under light microscopy, several studies found that Ig and complement deposits in the glomeruli detected by immunofluorescent staining were associated with clinical outcomes of IgAN patients.^18–21 In addition, electron microscopy (EM) assessments are used when diagnosing IgAN; however, few studies have focused on the clinical significance and prognostic value of ultrastructural changes identified by EM in IgAN.^1,14,22–25 Ultrastructural changes in IgAN include thinning or thickening of glomerular basement membrane (GBM) and texture abnormalities such as splitting or rupture of GBM.^22–25 Interestingly, an observational study found that IgAN patients with thinning of GBM had a different IgA glycosylation profile from those with normal GBM thickness.²⁵ Yuan et al. showed recently that IgAN patients with thinned GBM had a milder disease phenotype, nearly one third among which had diagnostic rare variants in COL4A3-5 genes.²⁶ This indicated ultrastructural changes of GBM may reflect different IgAN subtypes and correlate with the prognosis of patients with IgAN. Therefore, a comprehensive study of the clinical significance and mechanism of the ultrastructure changes of GBM is helpful for better understanding the development and prognosis of IgAN.

In this study, we evaluated the ultrastructural changes of GBM and investigated the prognostic value, clinical, and genetic relevance of these GBM changes based on an extended IgAN cohort with long-term follow-up. We found that IgAN patients with thickening of GBM had a higher risk of progressing to ESKD before and after adjustments of Oxford classifications or international risk-prediction tool in IgAN.²⁷ Thickening of GBM was positively correlated with mesangial hyperplasia, while genetic variants did not correlate to GBM thickening in IgAN patients. Our findings suggested that IgAN patients with thickening of GBM had a worse prognosis and proposed the importance of incorporating GBM ultrastructural changes into the pathologic evaluation system of IgAN.

Methods

Study Design

We retrospectively enrolled 1006 patients with biopsy-proven IgAN and available GBM EM assessment who underwent a renal biopsy at Shanghai Ruijin Hospital from January 2008 to September 2020. The inclusion criteria were (1) age older than 14 years, (2) biopsy proven primary IgAN, and (3) available renal biopsy EM assessment. Exclusion criteria included (1) renal biopsy specimen did not present glomeruli or presented only sclerotic or collapsed glomeruli and (2) follow-up time less than 3 months (Figure 1). The primary outcome of the study was ESKD (defined as eGFR <15 ml/min per 1.73 m², dialysis, or kidney transplantation) according to the Kidney Disease Improving Global Outcomes criteria.²⁸ The Oxford mesangial hypercellularity, segmental sclerosis, interstitial fibrosis/tubular atrophy, crescents classification was scored based on established criteria.¹⁴ Renal IgA, IgM, IgG, and C3 deposition were assessed by immunofluorescence staining on a scale of 0–3. All the pathologic features were extracted from diagnostic biopsies. This study was approved by the Ethics Committee of Shanghai Ruijin Hospital (2016 Clinical Ethics Review No. 2016-15).

Figure 1.

Flow diagram of cohort participant inclusion. GBM, glomerular basement membrane; IgAN, IgA nephropathy.

GBM Ultrastructural Analysis

GBM of diagnostic biopsies was assessed under transmission electron microscope. GBM thickness was measured by orthogonal intercept method.²⁶ In brief, the images were covered by a 2.5 cm² grid. Where the gridlines transected the endothelial surface of the GBM, the shortest distance between the endothelial cytoplasmic membrane to the outer lining of lamina rara externa underneath the cytoplasmic membrane of the epithelial foot process was measured.²⁹ Metrics in the Nephrotic Syndrome Study Network EM study were used; thickening of GBM was defined as at least 25% if the GBM appeared thicker than normal; thinning of GBM was defined as at least 25% if the GBM appeared thinner than normal.³⁰ Normal GBM thickness was described when none of the above changes occurred. For GBM texture, splitting or rupture of GBM was investigated and absence of splitting or rupture of GBM was described as normal GBM texture. Representative EM images of GBM thickness and texture abnormalities are shown in Supplemental Figure 1. GBM thickness was quantified in all patients identified as thickening of GBM and thinning of GBM and in 523 patients with normal GBM thickness (Supplemental Figure 2).

Results

Characteristics of the Study Cohort

We retrospectively enrolled 1006 biopsy-proven primary IgAN patients with a mean age of 37.3±12.3 years, 52% female and a median eGFR of 74.9 (interquartile range, 51.0) ml/min per 1.73 m². At the time of renal biopsy, the median urine protein was 1.2 (interquartile range, 1.7) grams per 24 hours, the median mean arterial pressure (MAP) was 94.0 (interquartile range, 16.0) mm Hg, and the median serum IgA level was 315.0 (interquartile range, 144.8) mg/ml. Among all patients, 80 (8%) had abnormal thickness of GBM including 29 (3%) thickening of GBM and 51 (5%) thinning of GBM. Abnormal GBM texture was found in 25 (2%) patients. Representative images of GBM abnormalities are shown in Figure 2. During a mean follow-up time of 46.4 months, 91 (9%) patients progressed to ESKD (Table 1).

Figure 2.

Kaplan-Meier curves showing the probability of ESKD-free survival of patients by GBM ultrastructural characteristic. (A) Median ESKD-free time: normal GBM thickness=160.8 months, thickening of GBM=92.7 months, thinning of GBM=127.0 months; thickening of GBM versus normal GBM thickness, log-rank test, P value = 0.014; thinning of GBM versus normal GBM thickness, log-rank test, P value = 0.14 (P values are adjusted by Bonferroni method). (B) Median ESKD-free time: normal GBM texture=160.8 months; splitting of GBM versus normal GBM texture, log-rank test, P value = 0.5.

Table 1.

Clinical characteristics of the enrolled IgA nephropathy patients

Characteristic	Total
N	1006
Age (yr)	37.3±12.3
Female, n (%)	525 (52)
DM, n (%)	13 (1)
UP (g/d)	1.2 (1.7)
eGFR (ml/min per 1.73 m2)	74.9 (51.0)
MAP (mm Hg)	94.0 (16.0)
Hb (g/L)a	128.0 (26.0)
Serum IgA (mg/ml)b	315.0 (144.8)
Serum C3 (mg/ml)c	92.0 (31.0)
Oxford scoring for IgANd, n (%)
Oxford-M1	392 (40)
Oxford-E1	340 (34)
Oxford-S1	872 (88)
Oxford-T1/T2	364 (37)
Oxford-C1/C2	477 (48)
Renal deposition (%)
IgA +/++/+++	14/56/30
C3 −/+/++/+++e	21/28/40/11
Ultrastructural GBM characteristics, (%)
Normal thickness/thickening/thinning	91/3/5
Normal texture/splitting or rupture	98/2
Average GBM thickness (nm)
Normal thickness	297.0 (59.5)
Thickening	465.0 (52.3)
Thinning	240.0 (11.8)
RAASi, n (%)	812 (80)
ESKD, n (%)	91 (9)
Follow-up time (mo)	46.4±38.4

DM, diabetes mellitus; GBM, glomerular basement membrane; Hb, hemoglobulin; IgAN, IgA nephropathy; MAP, mean arterial pressure; Oxford-C, Oxford Classification crescent; Oxford-E, Oxford Classification endocapillary cellularity; Oxford-M, Oxford Classification mesangial hypercellularity; Oxford-S, Oxford Classification segmental sclerosis; Oxford-T, Oxford Classification interstitial fibrosis/tubular atrophy; RAASi, renin-angiotensin-aldosterone system inhibitor; UP, urine protein.

^aData of seven individuals are not available.

^bData of 73 individuals are not available.

^cData of 101 individuals are not available.

^dTwenty-four individuals do not meet the Oxford scoring criteria.

^eData of 72 individuals are not available.

GBM Ultrastructural Changes Predict IgAN Prognosis

Kaplan-Meier survival analyses showed that median ESKD-free time of patients with thickening of GBM was significantly shorter than that of patients with normal GBM thickness (92.7 versus 160.8 months, P = 0.014). However, there was no statistical difference in median ESKD-free survival time between patients with thinning of GBM and patients with normal GBM thickness (127.0 versus 160.8 months, P = 0.14; Figure 2A). By Cox regression analyses, patients with thickening of GBM had an increased risk of ESKD compared with patients with normal GBM thickness before (hazard ratio [HR], 3.64; 95% confidence interval [CI], 1.47 to 7.55) and after being adjusted by Oxford Scoring (mesangial hypercellularity, segmental sclerosis, interstitial fibrosis/tubular atrophy, crescents) (HR, 2.92; 95% CI, 1.12 to 6.48) or international risk-prediction tool of IgAN²⁷ (HR, 3.51; 95% CI, 1.41 to 7.29). However, thinning of GBM or abnormal texture of GBM was not associated with the ESKD (Figure 2B and Table 2). To further investigate the effect of GBM thickening on renal outcome, we performed sensitivity analyses by using inverse probability of treatment weighting, multiply imputation approach, and by using composite renal outcome (defined as sustained ≥50% decline in eGFR, sustained ≥50% increase in serum creatinine or development of ESKD) as an end point; similar results were observed in all these analyses (Supplemental Table 1). In addition, when patients were divided into three groups according to average GBM thickness, those in the thickest GBM group had worse renal outcome compared with patients in the thinnest GBM group (Supplemental Figure 3 and Supplemental Table 2).

Table 2.

Associations between glomerular basement membrane ultrastructural changes and time to ESKD by Cox regression analysis

Effect	Model 1		Model 2		Model 3		Model 4
	HR	95% CI	HR	95% CI	HR	95% CI	HR	95% CI
Normal GBM thickness (ref)	—	—	—	—	—	—	—	—
Thickening of GBM (n=29)	3.64	1.47 to 7.55a	3.54	1.42 to 7.38a	2.92	1.12 to 6.48b	3.51	1.41 to 7.29a
Thinning of GBM (n=51)	1.21	0.40 to 2.81	1.36	0.45 to 3.20	1.45	0.45 to 3.63	1.34	0.44 to 3.09
Normal GBM texture (ref)	—	—	—	—	—	—	—	—
Splitting or rupture of GBM (n=25)	2.11	0.44 to 6.14	2.30	0.48 to 6.72	1.07	0.12 to 4.08	1.75	0.20 to 6.47

Model 1 (N=1006) is unadjusted.

Model 2 (n=1006) is adjusted by age, sex, and diabetes mellitus status.

Model 3 (n=982) is adjusted by Oxford scoring (MEST-C) for IgA nephropathy.

Model 4 (n=982) is adjusted by the full model with race of international risk-prediction tool in IgA nephropathy.²⁷ CI, confidence interval; GBM, glomerular basement membrane; HR, hazard ratio; MEST-C, mesangial hypercellularity, segmental sclerosis, interstitial fibrosis/tubular atrophy, crescents.

^aP value < 0.01.

^bP value < 0.05.

Genomic Rare Variants Analyses

To identify whether genomic rare variants played a role in GBM ultrastructural abnormalities, we performed whole exome sequencing in 56 patients, of which 32 were patients with normal GBM thickness, 17 were patients with thickening of GBM, and seven were patients with thinning of GBM. Thirteen patients had splitting or rupture of GBM. Rare variants in known kidney disease–related genes were scored according to the American College of Medical Genetics criteria.³¹ We observed trends toward detection of rare deleterious variants in patients with thinning of GBM; however, the statistical significance was limited by the small number of cases (Supplemental Table 3). The detected disease causal variant COL4A4 p.G478R was found in a 30-year-old woman with thinning of GBM and normal GBM texture. The variant reached likely pathogenic (PM1_Strong+PM2_Supporting+PP3+PP4) by American College of Medical Genetics scoring. Followed up for 21.2 months, the patients did not reach the outcome of ESKD.

Genomic Common Variants Analyses

To figure out whether genomic common variants of IgAN patients were involved in IgAN-related GBM ultrastructural changes, we included 20 susceptibility single nucleotide polymorphisms of IgAN from previous genome-wide association studies^4,32–35 and genotyped them in 617 patients. Correlations between known IgAN associated SNPs and GBM ultrastructural changes were investigated and adjusted by Benjamin-Hochberg procedure with a false discovery rate of 0.1. We did not find correlations between known IgAN associated SNPs and GBM thickness abnormalities (Supplemental Tables 4 and 5). However, the AA genotype of rs2071543, as well as TT genotype of HLA-DPB2 rs3129269, was associated with splitting or rupture of GBM in IgAN (Supplemental Table 6). Both SNPs reside within MHC locus. The SNP rs2071543 is a missense variant in PSMB8 (p.Gln49Lys), while SNP rs3129269 located in a noncoding area right next to HLA-DPB2. In addition, we calculated two published genetic risk score based on IgAN susceptibility loci. No significant relationship was seen between GBM ultrastructural features and the two genetic risk scores (Supplemental Table 7).

Association between GBM Ultrastructural Changes and Clinical and Pathologic Parameters

To investigate possible causes of GBM ultrastructural changes in IgAN patients, we analyzed the relevance between GBM features and clinical parameters as well as light microscopy and immunofluorescent pathologic parameters. Logistic regression analyses revealed that thickening of GBM was correlated with higher urine protein level and MAP at biopsy (Table 3). To further investigate whether thickening of GBM lead to a worse outcome in IgAN independent of high arterial pressure, we assessed the risk of ESKD in patients with thickened GBM and adjusted for MAP and confirmed that thickening of GBM had an increased risk of ESKD compared with patients with normal GBM thickness after being adjusted for MAP (HR, 2.89; 95% CI, 1.16 to 6.00). Thinning of GBM was associated with female sex and less renal IgA fluorescence intensity (Supplemental Table 8). Besides, splitting or rupture of GBM were inversely associated with renal C3 fluorescence intensity (Supplemental Table 9).

Table 3.

Correlation between clinical and pathologic parameters and glomerular basement membrane thickening by logistic regression analyses

Characteristic	GBM Thickness
	Normal	Thickening	OR (95% CI)a	OR (95% CI)b
Age at biopsy (yr)	37.2±12.3	39.6±14.5	1.02 (0.99 to 1.04)	—
Female, n (%)	469 (51)	13 (45)	0.78 (0.37 to 1.64)	—
DM, n (%)	12 (1)	1 (3)	2.70 (0.15 to 14.46)	—
UP (g/d)	1.2 (1.7)	2.4 (2.8)	1.18 (1.03 to 1.33)c	1.14 (0.97 to 1.30)
eGFR (ml/min per 1.73 m2)	75.0 (51.3)	56.6 (60.2)	0.99 (0.98 to 1.00)	—
MAP (mm Hg)	94.0 (16.0)	100.0 (8.0)	1.03 (1.00 to 1.06)c	1.02 (0.99 to 1.05)
Hb (g/L)d	128.0 (26.0)	129.0 (28.0)	1.00 (0.98 to 1.02)	—
Serum IgA (mg/ml)e	318.0 (145.0)	261.0 (155.5)	1.00 (0.99 to 1.00)	—
Serum C3 (mg/ml)f	93.0 (32.0)	92.0 (21.8)	1.00 (0.98 to 1.01)	—
Oxford scoring for IgANg, n (%)
Oxford-M1	354 (39)	18 (62)	2.52 (1.19 to 5.57)c	2.30 (1.08 to 5.12)c
Oxford-E1	308 (34)	8 (28)	0.73 (0.30 to 1.61)	—
Oxford-S1	793 (88)	25 (86)	0.84 (0.32 to 2.88)	—
Oxford-T1/T2	330 (37)	13 (45)	1.40 (0.65 to 2.95)	—
Oxford-C1/C2	440 (49)	8 (28)	0.40 (0.16 to 0.87)	—
Renal deposition (%)
IgA +/++/+++	13/56/31	21/45/28	0.92 (0.52 to 1.64)	—
C3 −/+/++/+++h	21/28/40/11	32/21/39/7	0.79 (0.53 to 1.18)	—
RAASi, n (%)	733 (80)	25 (86)	—	—
ESKD, n (%)	79 (9)	6 (21)	—	—
Follow-up time (mo)	46.8±38.9	39.4±31.0	—	—

CI, confidence interval; DM, diabetes mellitus; GBM, glomerular basement membrane; Hb, hemoglobulin; IgAN, IgA nephropathy; MAP, mean arterial pressure; OR, odds ratio; Oxford-C, Oxford Classification crescent; Oxford-E, Oxford Classification endocapillary cellularity; Oxford-M, Oxford Classification mesangial hypercellularity; Oxford-S, Oxford Classification segmental sclerosis; Oxford-T, Oxford Classification interstitial fibrosis/tubular atrophy; RAASi, renin-angiotensin-aldosterone system inhibitors; UP, urine protein.

^aUnivariate logistic regression, normal glomerular basement membrane thickness as reference.

^bMultivariate logistic regression, including factors of significance from univariate regression, normal glomerular basement membrane thickness as reference.

^cP value < 0.05.

^dData of seven individuals are not available.

^eData of 73 individuals are not available.

^fData of 101 individuals are not available.

^gTwenty-four individuals do not meet the Oxford scoring criteria.

^hData of 72 individuals are not available.

Notably, thickening of GBM was significantly associated with Oxford classification M1 before and after adjustment by baseline urine protein level and MAP (Table 3). Thinning of GBM showed a negative correlation with Oxford classification M1 by univariate logistic regression analysis; however, the odds ratio did not reach significance in the multivariate model, which may be due to a much stronger effect of female sex on thinning of GBM and limited number of GBM thinned patients included (Supplemental Table 8). The above results indicated that glomerular mesangial proliferation may correlate with GBM thickening in IgAN patients.

Mesangial Proliferation Increases in IgAN Patients with Thickened GBM

To test our hypothesis, we stained renal biopsy samples from four GBM thickened IgAN patients and four normal GBM thickness IgAN patients with mesangial cell marker GATA3 and cell proliferation marker Ki67.³⁶ The number of GATA3-expressing cell per total cell (4′-6-diamidino-2-phenylindole stained) per glomerular cross-section area increased in patients with thickening of GBM (Figure 3A). Moreover, assessment of proliferating mesangial cells (Ki67⁺, GATA3⁺) revealed that compared with normal GBM thickness patients, patients with thickening of GBM had an increased mesangial proliferation rate (Figure 3). These results indicated that IgAN patients with thickened GBM had more proliferating mesangial cells than those with normal GBM thickness. Mesangial cell proliferation may contribute to GBM thickening in IgAN patients; however, further studies are needed to address this question.

Figure 3.

IF staining of kidney biopsy samples revealed that patients with thickening of GBM (n=4) had an increase of mesangial cells (GATA3⁺) as well as mesangial proliferation (GATA3⁺, Ki67⁺) compared with patients with normal GBM thickness (n=4). (A) Assessment of the GATA3⁺ cells per total cell (DAPI stained) per GCS area. Each dot represents a mean at least five GCS areas per patient, and mean±SED per group is shown. (B) Assessment of the GATA3⁺ Ki67⁺ cells per total cell (DAPI stained) per GCS area. Each dot represents a mean at least five GCS areas per patient, and mean±SED per group is shown. (C) Representative sections of kidney biopsy samples from patients with IgAN stained by GATA3 and Ki67. Arrow shows GATA3 and Ki67 localization in proliferated mesangial cell. *P value < 0.05. DAPI, 4′-6-diamidino-2-phenylindole; GCS, glomerular cross-section; IF, immunofluorescent; SED, standard error of difference.

Discussion

The high heterogeneity of IgAN makes it challenging to treat patients based on different pathogenesis and different risk for adverse renal outcome, especially at an early stage. Ultrastructural alterations of GBM have been reported in IgAN patients.^22–25 Studies have shown that IgAN patients with thinned GBM had different glycosylation profile of serum IgA1 and are more likely to have rare variants in COL4A3-5 genes compared with classic IgAN,^25,26 suggesting that GBM alteration may result from unique disease pathogenesis in IgAN. However, existing studies are limited to small cohorts and focused on only thinning of GBM in IgAN. The prognostic value of GBM changes in IgAN has not been studied. In this study, based on an extended IgAN cohort, we investigated the prevalence, clinical relevance, genetic relevance, and the prognostic value of GBM changes including thickness and texture abnormalities in IgAN patients. Our findings suggested that thickening of GBM was independently associated with ESKD in IgAN patients.

In this cohort of 1006 patients older than 14 years, we found that 8% of patients had abnormal GBM thickness and 2% had abnormal GBM texture. More than 30 years ago, Yasuyuki et al. described an IgAN cohort of 73 children younger than 15 years, showing that a notable proportion (51%) of patients had GBM lesions including thickening with layering or splitting of the lamina densa and irregular attenuation or widening of the GBM.²⁴ However, the authors pointed out that the GBM lesions they observed were usually distributed only in some part of the glomerulus, which may lead to a much higher incidence of GBM abnormalities than that reported by this study. Different pathogenesis features between pediatric IgAN and adolescent and adult IgAN may also cause the difference. Regarding the prevalence of thinning of GBM in IgAN patients, Yuan et al. recently reported that 1.6% IgAN patients had thin GBM lesions, which was consistent with our finding.²⁶

Thickening of GBM was previously reported to be associated with poor renal outcome in other glomerular diseases including FSGS³⁰ and diabetic nephropathy.³⁷ In our study, we demonstrated that thickening of GBM was an independent risk factor of IgAN progression to ESKD, before and after adjusting for international risk-prediction tool in IgAN²⁷ or IgAN Oxford classification.^13,14 To the best of our knowledge, this is the first report of GBM ultrastructural changes predicting renal outcome in IgAN patients. Genomic analyses did not reveal rare or common genetic variants as a potential cause underlying GBM thickening in our cohort. However, by performing correlation analyses between GBM ultrastructural characteristics and clinical parameters, we found that the Oxford classification M1 score, which indicates mesangial hypercellularity, positively correlated with thickening of GBM. By IF staining, we found that mesangial proliferation increased in IgAN patients with thickening of GBM. Notably, Masuda et al. previously demonstrated an altered constituent of thickened GBM of IgAN, where expression of type 4 collagen α5 decreased and expression of type 4 collagen α2 increased.³⁸ Further studies are needed to investigate whether proliferated mesangial cells plays a role in GBM thickening and constituent changes.

Thinning of GBM was significantly related to female sex, which was consistent with previous reports.^25,39 There was also a trend toward less mesangial hypercellularity and less renal IgA deposition in patients with thinning of GBM, which was different from classic IgAN. Indeed, a previous observational study has shown that IgAN patients with thinning of GBM had a normal level of serum IgA1 glycosylation.²⁵ Recently, Yuan et al. revealed that over two thirds of IgA patients with thin GBM lesions had diagnostic variants of COL4A3–5 genes.²⁶ In our study, we did not find that thinning of GBM was associated with rare or common variants among IgAN-related genes, although the statistical power was limited by the rarity of variants and relatively small number of cases.

Our study has some limitations. First, it is a retrospective study from a single center. Second, although this is a large cohort, the low prevalence of GBM ultrastructural changes in IgAN patients limits the case number of IgAN with abnormal GBM. Third, owing to the limited number of patients who underwent genetic analyses, we did not have enough power to illustrate the contribution of genomic variants in GBM alterations in IgAN patients. Besides, the patients who underwent whole exome sequencing did not represent the entire cohort well due to the high cost of whole exome sequencing and limited access to patients' samples. Fourth, potential biomarkers of IgAN such as serum galactose-deficient IgA1 were not available in our study. Furthermore, we only included GBM ultrastructural changes in this study. Other ultrastructural pathologic features such as changes of podocyte that might contribute to disease progression remain to be evaluated. In addition, it is hard to completely rule out concurrent renal diseases which subsequently developed during the follow-up time due to lack of repeated biopsies. Further prospective studies with larger and diverse population will be helpful to overcome these limitations.

In conclusion, we demonstrated that thickening of GBM was an independent risk factor for ESKD in IgAN patients and observed a correlation between thickened GBM and mesangial hypercellularity. Our findings proposed the importance of incorporating GBM ultrastructural changes into the pathologic evaluation system of IgAN.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/KN9/B284.

Author Contributions

Conceptualization: Xie Jingyuan.

Data curation: Xu Jing, Yang Mingxin, Wei Wenjie, Fang Zhengying.

Formal analysis: Fang Zhengying.

Funding acquisition: Xie Jingyuan.

Investigation: Yang Mingxin, Wei Wenjie, Fang Zhengying.

Methodology: Xu Jing, Xie Jingyuan, Fang Zhengying.

Resources: Ren Hong, Ma Jun, Chen Nan, Wang Weiming, Du Wen, Pan Xiaoxia, Yan Ouyang, Chen Zijin.

Supervision: Xie Jingyuan.

Writing – original draft: Fang Zhengying.

Writing – review & editing: Xie Jingyuan, Yan Ouyang.

Funding

J. Xie: Key Technologies Research and Development Program (2024YFC2511001), Major International (Regional) Joint Research Program of National Natural Science Foundation of China (82120108007), National Natural Science Foundation of China (82370711), Shanghai Shenkang Hospital Development Center 2025 Shanghai Clinical Cohort Project (SHDC2025CCS015), National Facility for Translational Medicine (Shanghai) Open subjects (TMSK-2024-101, NRCTM[SH]-2025-10 [Ruijin Base]), Program of Shanghai Academic/Technology Research Leader (21XD1402000), Science and Technology Innovation Action Plan of Shanghai Science and Technology Commettee (22140904000), Shanghai Municipal Education Commission Gaofeng Clinical Medicine Grant (20152207), Shanghai Shenkang Hospital Development Center “Three-year Action Plan for Promoting Clinical Skills and Clinical Innovation in Municipal Hospitals” (SHDC2020CR6017), Research Foundation of Ruijin Hospital (JZ202408), and Shanghai Municipal Key Clinical Specialty (shslczdzk02502). Y. Ouyang: National Natural Science Foundation of China (No.82270739); This work has been made possible by Yan Ouyang's ISN-funded Fellowship.

Declarative Statements

This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. All patients provided written informed consent. This study includes clinical experimentation and complies with the Declaration of Helsinki.

Data Availability Statements

Partial restrictions to the data and/or materials apply. Explanation Why Data Cannot Be Shared: According to the rule of our institution, the health care data are not allowed to be uploaded elsewhere. Data can be provided on request to the principal investigators of the study.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/KN9/B285.

Supplemental Figure 1. Representative EM images of GBM thickness and texture abnormalities.

Supplemental Figure 2. Average GBM thickness in patients with thickening of GBM, normal GBM thickness, and thinning of GBM.

Supplemental Figure 3. Kaplan-Meier curves showing the probability of ESKD-free survival of patients by average GBM thickness.

Supplemental Table 1. Associations between GBM thickening, compared with normal GBM thickness, and time to ESKD by Cox regression with sensitivity analyses.

Supplemental Table 2. Associations between average GBM thickness and time to ESKD by Cox regression analysis.

Supplemental Table 3. Rare deleterious variants discovered in patients by whole exome sequencing.

Supplemental Table 4. Logistic regression analysis for thickening of GBM (versus normal GBM thickness) and previously identified IgAN-related SNPs.

Supplemental Table 5. Logistic regression analysis for thinning of GBM (versus normal GBM thickness) and previously identified IgAN-related SNPs.

Supplemental Table 6. Logistic regression analysis for splitting or rupture of GBM (versus normal GBM texture) and previously identified IgAN-related SNPs.

Supplemental Table 7. Logistic regression analysis for GBM ultrastructural abnormalities and previously reported genetic risk score for IgAN.

Supplemental Table 8. Correlation between clinical and pathologic parameters and GBM thinning by logistic regression analyses.

Supplemental Table 9. Correlation between clinical and pathologic parameters and GBM texture by univariate logistic regression analyses.

Supplemental Methods