Multi-Autoantibody Signature and Clinical Outcome in Membranous Nephropathy

Visual Abstract

Abstract

Background and objectives

Patients with membranous nephropathy can have circulating autoantibodies against membrane-bound (phospholipase A2 receptor 1 [PLA2R1] and thrombospondin type-1 domain containing 7A [THSD7A]) and intracellular (aldose reductase, SOD2, and α-enolase) podocyte autoantigens. We studied their combined association with clinical outcomes.

Design, setting, participants, & measurements

Serum levels of anti-PLA2R1, anti-THSD7A, anti-aldose reductase, anti-SOD2, and anti−α-enolase autoantibodies were determined in 285 patients at diagnosis and during follow-up using standardized and homemade assays. An eGFR>60 ml/min per 1.73 m² and remission of proteinuria (<0.3/<3.5 g per d) after 12 months were the outcomes of interest.

Results

At diagnosis, 182 (64%), eight (3%), and 95 (33%) patients were anti-PLA2R1⁺, anti-THSD7A⁺, and double negative, respectively. The prevalence of a detectable antibody to at least one intracellular antigen was similarly distributed in patients who were anti-PLA2R1⁺ (n=118, 65%) and double negative (n=64, 67%). Positivity for anti-PLA2R1, anti-SOD2, and anti–α-enolase antibodies and higher titers at diagnosis were associated with poor clinical outcome independently to each other. Combined positivity for anti-PLA2R1, anti-SOD2, and anti−α-enolase was associated with highest risk of poor outcome (odds ratio, 5.5; 95% confidence interval, 1.2 to 24; P=0.01). In Kaplan–Meier analysis, patients who were anti-PLA2R1⁺/anti-SOD2⁺ or anti-PLA2R1⁺/anti−α-enolase⁺ had lower eGFR at 12 months compared with patients who were anti-PLA2R1⁺/anti-SOD2⁻ or anti−α-enolase⁻. Predictive tests (net reclassification index and area under the curve–receiver-operating characteristic analysis) showed that combined assessment of antibodies improved classification of outcome in 22%–34% of cases for partial remission of proteinuria and maintenance of normal eGFR. For patients with nephrotic syndrome at diagnosis, anti-SOD2 positivity and high anti-PLA2R1 titer were associated with a lack of complete remission. Patients who were anti-PLA2R1⁻/anti-intracellular antigens⁻ had the lowest proteinuria and the highest eGFR at diagnosis and the lowest risk of lower eGFR at 12 months. Epitope spreading was present in 81% of patients who were anti-PLA2R1⁺ and was associated with increased positivity for intracellular antigens and poor eGFR at diagnosis and 12 months.

Conclusions

Combined serological analysis of autoantibodies targeting membrane-bound and intracellular autoantigens identifies patients with poor clinical outcomes.

Introduction

Membranous nephropathy is a common cause of nephrotic syndrome in adults (1–3). It is characterized by thickening of the glomerular basement membrane with a well-defined pattern of subepithelial immune complex deposits, consisting in most cases of IgG4 autoantibodies targeting various autoantigens. The clinical outcome of membranous nephropathy is variable (4). About one third of patients do not require immunosuppressive therapy and reach spontaneous remission, whereas two thirds require immunosuppressive treatment (such as cyclosporine, cytotoxic agents, or anti-CD20 antibodies) and respond, or not, to treatment, with possible progression to ESRD over 5–10 years (5–10).

Antibodies against various podocyte autoantigens have been shown to play a role in the pathogenesis of various forms of membranous nephropathy. In rare familial cases of alloimmune membranous nephropathy, autoantibodies were identified against neutral endopeptidase (11). These autoantibodies developed in mothers deficient for neutral endopeptidase and crossed the placenta to induce membranous nephropathy in the fetus.

This elegant demonstration of a first causative autoantibody of membranous nephropathy was followed by the identification of phospholipase A2 receptor 1 (PLA2R1) as the major autoantigen in adult membranous nephropathy, with circulating autoantibodies present in about 70% of patients (12). Genetic studies further supported a causal role of PLA2R1 based on association of the disease with 6p21 HLA-DQA1 and 2q24 PLA2R1 loci in cohorts of White (13) and Asian (14) patients. Multiple epitopes have been identified in the CysR, CTLD1, CTLD7, and CTLD8 domains of PLA2R1 (15–20), which may be reminiscent of a mechanism of epitope spreading reflecting the maturation of the immune response from CysR to other PLA2R1 domains (21,22). Epitope spreading has also been documented in the Heymann rat nephritis model where megalin is the major autoantigen (23).

Thrombospondin type-1 domain containing 7A (THSD7A) was reported as a second autoantigen in adult membranous nephropathy, with circulating anti-THSD7A autoantibodies present in about 3% of a different group of patients (24).

Autoantibodies targeting different intracellular podocyte autoantigens including aldose reductase (AR) (25), superoxide dismutase 2 (SOD2) (26), and α-enolase (αENO) (27) have been identified in a significant number of patients with membranous nephropathy (28).

Finally, three new antigenic markers of autoimmunity (exostosin1/exostosin2 and NELL-1) have been very recently identified in immune deposits, with exostosins more present in secondary cases of membranous nephropathy (i.e., class V lupus nephritis) (29,30). The exact prevalence and significance of autoantibodies against these new autoantigens in the pathogenesis of membranous nephropathy remain to be determined, with circulating autoantibodies currently found only for NELL-1, and with a different prevalence between American and European cohorts of patients (29,30).

In this study, we aimed to: (1) detect autoantibodies directed against the well-characterized membrane-bound autoantigens PLA2R1 and THSD7A and the intracellular autoantigens SOD2, AR, and αENO in a large retrospective cohort of 285 patients with membranous nephropathy at diagnosis and after 6 and 12 months of follow-up; and (2) evaluate the clinical outcome of patients stratified into different groups based on their combined positivity or not for multiple autoantibodies directed against the different autoantigens.

Materials and Methods

Study Design and Patient Population

This is a cross-sectional study in patients with membranous nephropathy assessing the relationships between the detection of different autoantibodies at baseline and eGFR and proteinuria at 12 months. We included 285 patients at the time of a histopathological finding of idiopathic or primary membranous nephropathy (Supplemental Table 1) and followed them with clinical visits for at least 12 months. They were enrolled in the frame of two studies: 244 patients were enrolled in Italy (Italian Study Group on Membranous Nephropathy-EudraCT 2011–003942–41) and 41 in France (Cohort SOURIS NCT02199145). Criteria for enrollment were: (1) biopsy-proven diagnosis of membranous nephropathy associated with proteinuria >0.3 g/d; (2) normal complement profile; (3) negative tests for anti-nuclear antibodies, anti-double strand DNA, ANCA, cryoglobulins, and the absence of viral markers (hepatitis B surface antigen and HIV) (4); absence of clinical and biochemical signs of cancer; and (5) serum samples available at diagnosis and during follow-up for determining antibody levels. Clinical studies were approved by the relevant institutional review boards in the different hospitals and were conducted according to the principles of the Declaration of Helsinki, with written informed consent obtained from all patients. The study was presented at the Ethical Committee of Scientific Institute for Research, Hospitalization and Healthcare Giannina Gaslini. Serum was also obtained from 50 healthy donors recruited at the same institutions that participated to the sample recruitment. Healthy donors had at least one normal urine analysis and normal clinical tests in the prior 6 months (Supplemental Table 1).

Detection of Circulating Autoantibodies

Anti-PLA2R1 autoantibodies (detection of total IgG) against the extracellular domain of PLA2R1 were measured with the standardized ELISA from Euroimmun (Medizinische Labordiagnostika AG, Lübeck, Germany) (31). Anti-PLA2R1 autoantibodies (detection of IgG4 subclass) against the specific epitope-containing domains CysR, CTLD1, and CTLD7 were determined as described (19) using recombinant soluble forms of each domain harboring a hemagglutinin tag and produced in HEK293 cells. Anti-THSD7A autoantibodies (detection of IgG4 subclass) against the extracellular domain of THSD7A were determined by ELISA as described (32). Positivity was confirmed by western blot (24) and indirect immunofluorescence (33) assays. Autoantibodies against the intracellular antigens AR, SOD2, and αENO (detection of IgG4 subclass) were determined by dot blot using recombinant proteins as immobilized antigens as described (26,27). Further details on the detection methods and source of antigens are given in the Supplemental Material. For anti-PLA2R1 total IgG, we used the normal limit of positivity of 20 RU/ml given by the manufacturer (31). For anti-THSD7A and intracellular autoantibodies, normal limits were calculated from receiver-operating characteristic (ROC) curves as described (32,34). High and low levels of autoantibodies were defined as the titers above median and between median and limit of normality, respectively (Supplemental Table 2).

Outcome

Clinical outcome was evaluated on the basis of proteinuria levels at diagnosis versus after 12 months of follow-up, and was defined as complete remission (proteinuria <0.3 g/d) or partial remission (proteinuria <3.5 g/d) according to Thompson et al. (35). Failure to achieve clinical remission was defined by persistence of proteinuria >3.5 g/d. Kidney function (eGFR) was evaluated based on the Chronic Kidney Disease Epidemiology Collaboration creatinine equation (36).

Statistical Analyses

Autoantibody levels were expressed as median and interquartile range. The differences in serum concentration of all autoantibodies within patient groups were analyzed using nonparametric Mann–Whitney U test for unpaired samples; Spearman’s correlation was used to evaluate correlations among circulating levels of the various types of autoantibodies. The ROC curve analysis was used to discriminate between healthy people and membranous nephropathy patients based on levels of each autoantibody at diagnosis. The best cut-offs were selected as the values that minimized the geometric distance from 100% sensitivity and 100% specificity on the ROC curves. In these analyses, discrimination is assessed as the ability of the test (performance) to distinguish the presence of the outcome beyond chance, defined when the area under the curves (AUC) are significantly >0.5. Odds ratios and 95% confidence intervals (95% CIs) were tested to determine the associations between autoantibody positivity or high titer and proteinuria and eGFR using 2×2 contingency tables. Fisher’s exact test was used to determine the statistical significance. The P values were corrected for multiple comparisons with alpha error values of 5%. We used the Kaplan-Meier method to construct survival curves for time to events according to the level of proteinuria. Differences in outcomes were estimated using the log-rank (Mantel-Cox) test and their hazard ratio; P values ≤0.05 were considered as significant. The performance of single antibody and their synergism for predicting clinical outcome were assessed by two-category net reclassification index (NRI) and AUC-ROC analysis (34). We used R for all statistical analyses (http://www.R-project.org/).

Results

Patients and Predictors of Clinical Outcome

We included 285 patients with biopsy-proven membranous nephropathy and whose clinical characteristics are shown in Table 1 and Supplemental Table 1, according to PLA2R1 and intracellular antibody positivity. Most patients were men (n=194, 68%). At diagnosis, 67% of patients had nephrotic syndrome. In total, 79% of patients were treated with immunosuppressive treatments: steroids (33%), cytotoxic drugs (32%), rituximab (12%), and cyclosporine (25%). Antiproteinuric treatment with angiotensin converting enzyme inhibitors was added to the above therapy. Among all patients, 21% received only angiotensin converting enzyme inhibitors. Patients with nephrotic syndrome (n=177, 62%) had persistent high proteinuria after antiproteinuric treatment, with evidence for deterioration of kidney function (eGFR). When the cohort was analyzed as a whole, gender and nephrotic proteinuria at diagnosis were the only two clinical factors associated with the risk of failure to achieve complete remission of proteinuria after 12 months of follow-up (Supplemental Figure 1), in agreement with previous findings (35).

Table 1.

Clinical characteristics of patients with membranous nephropathy according to antibody profile

Characteristics	Phospholipase A2 Receptor 1+/Intracellular+ (n=118)	Phospholipase A2 Receptor 1+/Intracellular- (n=64)	Phospholipase A2 Receptor 1−/Intracellular+ (n=64)	Phospholipase A2 Receptor 1−/Intracellular− (n=31)
Sex (M/F)	84/34	45/19	47/17	21/10
Age (yr)	61 (50–72)	63 (46–70)	57 (42–67)	55 (43–71)
Membranous nephropathy histologic stage (%)
I	19	24	26	45
II	44	49	50	33
III	29	23	15	11
IV	8	4	9	11
Treatment (%)
Steroids/ACEia	80	78	85	78
Cyclosporine A	18	30	14	16
Cytotoxic	43	44	33	32
Rituximab	7	22	16	19
Clinical characteristics
Proteinuria (g/d)
Baseline	5.6 (3.3–9.7)	6.0 (3.3–9.8)	3.4 (1.8–6.4)*	4.4 (2.6–6.4)*
12 mo	1.7 (0.4–4.4)	1.8 (0.6–4.3)	1.3 (0.3–3.5)	0.5 (0.1–2.7)**
CKD-EPI (ml/min per 1.73 m2)
Baseline	69 (46–99)	66 (47–100)	72 (43–99)	89 (63–103)**
12 mo	74 (48–99)	78 (59–99)	67 (44–101)	96 (72–103)**

Patients both PLA2R1 and intracellular antibody positive; patients PLA2R1 positive and intracellular antibody negative; patients PLA2R1 negative and intracellular antibody positive; and patients both PLA2R1 negative and intracellular antibody negative. Data are reported as number and percentage or median with interquartile range. Mann–Whitney U or Kruskal–Wallis tests were used for continuous variables, respectively, for two or more of two unpaired samples, and chi-squared or Fisher’s exact tests were used for categorical variables. Two-tailed P values ≤0.05 were considered as significant. *P=0.001 versus A and B; **P=0.001 versus A, B, and C. PLA2R1, phospholipase A2 receptor 1; M, male; F, female; ACEi: angiotensin converting enzyme inhibitors; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration.

^aSteroids and ACEi were given alone or in association with other drugs.

Individual Autoantibodies: Distribution and Levels

All patients were tested at diagnosis for the presence of circulating autoantibodies recognizing PLA2R1 and THSD7A by specific ELISAs (31,32). The presence of circulating autoantibodies specifically recognizing the CysR, CTLD1, and CTLD7 domains of PLA2R1 were also analyzed, defining nonspreaders (patients with detectable reactivity limited to CysR) and spreaders (patients with detectable reactivity to CysR, CTLD1, and/or CTLD7 domains) (19,37). Patients were also tested for the presence of autoantibodies recognizing the intracellular autoantigens SOD2, AR, and αENO (26,27). Patients with single or multiple positivities to any of these intracellular antigens were referred to as intracellular⁺, whereas patients with no detectable reactivity were referred to as intracellular⁻. The same analyses were repeated after 6 and 12 months (Supplemental Figures 2, A–E).

Among all patients, 182 (64%) were positive for anti-PLA2R1 (PLA2R1⁺ patients, Figure 1A, Supplemental Figure 2A). The median level of anti-PLA2R1 autoantibodies was 122 RU/ml (interquartile range, 60–267) at baseline and decreased after 6 and 12 months of follow-up (Supplemental Figure 2A). At baseline, 118 (65%) PLA2R1⁺ patients were positive for one or more intracellular antigens, defining PLA2R1⁺/intracellular⁺ patients, whereas 64 (35%) were negative, defining PLA2R1⁺/intracellular⁻ patients (Figure 1A, Supplemental Figure 3A, Table 1). Interestingly, the percentage of intracellular⁺ patients was higher in those with high titer of anti-PLA2R1 autoantibodies (Figure 1B). Furthermore, PLA2R1⁺ patients defined as spreaders had higher anti-PLA2R1 titers than nonspreaders (Supplemental Figure 3B) and were also more positive for intracellular autoantigens (Supplemental Figure 3C, Supplemental Table 3), in particular for αENO (Supplemental Figure 3D).

Figure 1.

Detection of circulating autoantibodies against the membrane-bound autoantigens (phospholipase A2 receptor 1 [PLA2R1] and thrombospondin type-1 domain containing 7A [THSD7A]) and against the intracellular autoantigens (aldose reductase [AR], superoxide dismutase 2 [SOD2], and α-enolase [αENO]) allows the stratification of patients in different groups.(A) Patients were first divided according to the presence of circulating autoantibodies targeting the membrane-bound autoantigens PLA2R1 or THSD7A. The three groups of patients (PLA2R1⁺, THSD7A⁺, and PLA2R1⁻/THSD7A⁻) were further divided according to positivity for autoantibodies targeting any of the three intracellular podocyte autoantigens (AR, SOD2, αENO). (B) Stratification of PLA2R1⁺ patients according to anti-PLA2R1 titers below and above the median. *The percentage of patients positive for intracellular autoantibodies was higher in those with high anti-PLA2R1 titers (P=0.02, Fisher’s exact test). MN, membranous nephropathy.

Among patients, only eight (3%) were positive for anti-THSD7A (THSD7A⁺, Figure 1A, Supplemental Figure 2B). One was double positive for PLA2R1 and THSD7A (Supplemental Figure 2B). Seven of these patients had additional autoantibodies against at least one intracellular autoantigen (Figure 1A, Supplemental Figure 3E). Because of their low prevalence, anti-THSD7A⁺ patients were not considered for further analysis.

Thus, among all patients, 95 patients (33%) were identified as double negative for PLA2R1 and THSD7A. Among these latter patients, 64 (67%) were also positive for one or more intracellular antigens, defining patients who were double negative/intracellular⁺, whereas 31 (33%) were negative, defining patients who were double negative/intracellular⁻ (Figure 1A, Supplemental Figure 3F, Table 1).

Overall, autoantibodies against AR, SOD2, and αENO were detected in 85, 85, and 117 patients, respectively (Supplemental Figure 2, C–E). The median levels of each intracellular autoantibody decreased at 6 and 12 months of follow-up (Supplemental Figure 2, C–E). Positivity and titer levels for anti-PLA2R1, anti-SOD2, anti-AR, and anti-αENO autoantibodies at baseline did not correlate to each other (Supplemental Figure 3G).

Association of the Different Antibodies with Clinical Outcome

Considering all patients, positivity for anti-PLA2R1, anti-SOD2, and anti-αENO autoantibodies at diagnosis was variably and independently associated with complete or partial remission of proteinuria and eGFR after 12 months (Table 2). Concomitant positivity of the three antibodies had the highest odds ratio for complete remission of proteinuria after 12 months (odds ratio, 5.5; 95% CI, 1.2 to 24; P=0.01). High titers of anti-PLA2R1 and anti-SOD2 above the median (Supplemental Table 2, n=91 and n=43, respectively) were similarly associated with lack of partial or complete remission of proteinuria after 12 months compared with those with low titers (Table 2). Kaplan–Meier analysis showed that combined positivity for anti-PLA2R1 and anti-SOD2 or anti-αENO was associated with a higher risk of low eGFR (<60 ml/min per 1.73 m²) after 12 months compared with positivity for only anti-PLA2R1 (Figure 2). Additionally, high titers of anti-PLA2R1 and anti-αENO were associated with reduced eGFR after 12 months, as compared with low titers (Table 2). Considering only patients with nephrotic syndrome at diagnosis, the associations were different (Table 3). Only positivity for anti-SOD2 antibodies and concomitant positivity for anti-PLA2R1 and anti-SOD2 were associated with persistence of proteinuria after 12 months. High titers of anti-PLA2R1 were also associated with persistence of proteinuria. Considering treatment, positivity and high levels of anti-PLA2R1 and anti-SOD2 were the main features associated with lack of partial and complete remission after 12 months in patients who had been treated with cytotoxic drugs, cyclosporine A, or rituximab (Supplemental Table 4).

Table 2.

Two-way contingency table showing the association of antibody levels with indexes of kidney outcome and the interaction between autoantibodies against phospholipase A2 receptor 1 and intracellular antigens for the whole cohort

Antibody	Proteinuria >0.3 g/d at 12 mo	Proteinuria ≤0.3 g/d at 12 mo	Odds Ratio (95% CI)	Proteinuria >3.5 g/d at 12 mo	Proteinuria ≤3.5 g/d at 12 mo	Odds Ratio (95% CI)	eGFR<60 ml/min per 1.73 m2 at 12 mo	eGFR>60 ml/min per 1.73 m2 at 12 mo	Odds Ratio (95% CI)
Anti-PLA2R1 (P)
Positive Negative	149 (54%) 60 (22%)	33 (12%) 35 (13%)	2.6 (1.5 to 4.6)	55 (20%) 19 (7%)	127 (46%) 76 (27%)	1.7 (0.9 to 3.1)	67 (24%) 31 (11%)	115 (42%) 64 (23%)	1.2 (0.7 to 2.1)
Anti-SOD2 (S)
Positive Negative	74 (27%) 135 (49%)	11 (4%) 57 (21%)	2.8 (1.4 to 5.7)	31 (11%) 43 (16%)	54 (19%) 149 (54%)	2 (1.1 to 3.5)	34 (12%) 64 (23%)	51 (18%) 128 (46%)	1.3 (0.8 to 2.3)
Anti-αENO (E)
Positive Negative	95 (34%) 114 (41%)	22 (8%) 46 (17%)	1.7 (0.98 to 3.1)	40 (14%) 34 (12%)	77 (28%) 126 (45%)	1.9 (1.1 to 3.3)	51 (18%) 47 (17%)	66 (24%) 113 (41%)	1.8 (1.1 to 3.1)
P+/S+a
Positive Negative	55 (30%) 94 (52%)	6 (3%) 27 (15%)	2.6 (1.02 to 6.8)	22 (12%) 33 (18%)	39 (21%) 88 (48%)	1.5 (0.8 to 2.9)	24 (13%) 43 (24%)	37 (20%) 78 (43%)	1.2 (0.6 to 2.2)
P+/S+/E+a
Positive Negative	39 (21%) 110 (60%)	2 (1%) 31 (17%)	5.5 (1.2 to 24)	16 (9%) 39 (21%)	25 (14%) 102 (56%)	1.7 (0.8 to 3.5)	16 (9%) 51 (28%)	25 (14%) 90 (49%)	1.1 (0.5 to 2.3)
Anti-PLA2R1 (P)
High titer Low titer	82 (45%) 67 (37%)	9 (5%) 24 (13%)	3.3 (1.4 to 7.5)	36 (20%) 19 (10%)	55 (30%) 72 (40%)	2.5 (1.3 to 4.8)	44 (24%) 23 (13%)	47 (26%) 68 (37%)	2.8 (1.5 to 5.2)
Anti-SOD2 (S)
High titer Low titer	40 (47%) 27 (32%)	3 (4%) 15 (18%)	4.6 (1.1 to 19)	22 (26%) 9 (11%)	21 (25%) 33 (39%)	3.8 (1.5 to 9.9)	19 (22%) 15 (18%)	24 (28%) 27 (32%)	1.4 (0.6 to 3.4)
Anti-αENO (E)
High titer Low titer	49 (42%) 46 (39%)	10 (9%) 12 (10%)	1.3 (0.5 to 3.2)	26 (22%) 14 (12%)	32 (27%) 45 (38%)	2.6 (1.2 to 5.8)	32 (27%) 19 (16%)	27 (23%) 39 (33%)	2.4 (1.1 to 5.1)
High P+/S+b
High titer P/S+ High titer P/S−	21 (34%) 34 (56%)	3 (5%) 3 (5%)	0.6 (0.1 to 3.3)	8 (13%) 14 (23%)	16 (26%) 23 (38%)	0.8 (0.3 to 2.4)	10 (16%) 14 (23%)	14 (23%) 23 (38%)	1.2 (0.4 to 3.3)

The 2×2 contingency table reports the association with OR and 95% CI between antibody titer and clinical outcome. Data are calculated from all patients excluding the eight anti-THSD7A positive patients (n=277). The upper section shows the association between positivity versus negativity of each autoantibody at diagnosis with clinical outcome of proteinuria (complete (≤0.3 g/d) or partial (<3.5 g/d) remission) and eGFR after 12 mo. The lower section shows the association of high versus low titers of each antibody with the same parameters of proteinuria and eGFR. The additive effect of positivity for more than one antibody is indicated as P⁺/S⁺ and P⁺/S⁺, /E⁺ (anti-PLA2R1⁺/anti-SOD2⁺/anti-αENO⁺). PLA2R1, phospholipase A2 receptor 1; αENO, α-enolase. S, SOD; OR, odds ratio; 95% CI, 95% confidence interval; curve analysis; THSD7A, thrombospondin type-1 domain containing 7A.

^aIn these cases, P⁺/S⁺ were compared with P⁺/S⁻, and P⁺/S⁺/E⁺ were compared with P⁺/S⁻/E⁻.

^bIn this case, high P/S (high P titer and S positivity) were compared with high P titer and S negativity.

Figure 2.

Positivity for anti-SOD2 and anti-αENO has an additive effect with positivity for anti-PLA2R1 for decreased eGFR at 12 months. Kaplan–Meier curve showing the proportion of patients in the different groups who reached the endpoint of reduction of kidney function (eGFR <60 ml/min per 1.73 m²) after 12 months of follow-up based on the concomitant positivity or not for anti-PLA2R1 and anti-SOD2 (A) or anti-αENO (B). The Kaplan–Meier curve of the difference was calculated with the log-rank (Mantel-Cox) test and the hazard ratios with confidence intervals for the different groups.

Table 3.

Two-way contingency table showing the association of antibody levels with indexes of kidney outcome and the interaction between autoantibodies against phospholipase A2 receptor 1 and intracellular antigens for nephrotic patients at diagnosis

Antibody	Proteinuria >0.3 g/d at 12 mo	Proteinuria ≤0.3 g/d at 12 mo	Odds Ratio (95% CI)	Proteinuria >3.5 g/d at 12 mo	Proteinuria ≤3.5 g/d at 12 mo	Odds Ratio (95% CI)	eGFR<60 ml/min per 1.73 m2 at 12 mo	eGFR>60 ml/min per 1.73 m2 at 12 mo	Odds Ratio (95% CI)
Anti-PLA2R1 (P)
Positive Negative	109 (62%) 35 (20%)	22 (12%) 11 (6%)	1.5 (0.7 to 3.5)	44 (25%) 11 (6%)	87 (49%) 35 (20%)	1.6 (0.7 to 3.5)	53 (30%) 16 (9%)	78 (44%) 30 (17%)	1.3 (0.6 to 2.6)
Anti-SOD2 (S)
Positive Negative	57 (32%) 87 (49%)	6 (3%) 27 (15%)	2.9 (1.1 to 7.6)	24 (14%) 31 (18%)	39 (22%) 83 (47%)	1.6 (0.8 to 3.2)	29 (16%) 40 (23%)	34 (19%) 74 (42%)	1.6 (0.8 to 2.9)
Anti-αENO (E)
Positive Negative	67 (38%) 77 (44%)	11 (6%) 22 (12%)	1.7 (0.8 to 3.8)	29 (16%) 26 (15%)	49 (28%) 73 (41%)	1.7 (0.9 to 3.1)	36 (20%) 33 (19%)	42 (24%) 66 (37%)	1.7 (0.9 to 3.1)
P+/S+a
Positive Negative	45 (34%) 64 (49%)	3 (2%) 19 (14%)	4.4 (1.2 to 16)	19 (14%) 25 (19%)	29 (22%) 58 (44%)	1.5 (0.7 to 3.2)	21 (16%) 32 (24%)	27 (21%) 51 (39%)	1.2 (0.6 to 2.5)
P+/S+/E+a
Positive Negative	30 (23%) 79 (60%)	2 (2%) 20 (15%)	3.8 (0.8 to 17)	15 (11%) 29 (22%)	17 (13%) 70 (53%)	2.1 (0.9 to 4.8)	13 (10%) 40 (31%)	19 (15%) 59 (45%)	1 (0.4 to 2.3)
Anti-PLA2R1 (P)
High titer Low titer	65 (50%) 44 (34%)	6 (5%) 16 (12%)	3.9 (1.4 to 11)	26 (20%) 18 (14%)	45 (34%) 42 (32%)	1.3 (0.6 to 2.8)	33 (25%) 20 (15%)	41 (31%) 37 (28%)	1.5 (0.7 to 3)
Anti-SOD2 (S)
High titer Low titer	28 (44%) 29 (46%)	5 (8%) 1 (2%)	0.2 (0.02 to 1.7)	14 (22%) 10 (16%)	18 (29%) 21 (33%)	1.6 (0.6 to 4.6)	15 (24%) 14 (22%)	16 (25%) 18 (29%)	1.2 (0.4 to 3.2)
Anti-αENO (E)
High titer Low titer	33 (42%) 34 (44%)	5 (6%) 6 (8%)	1.2 (0.3 to 4)	15 (19%) 14 (18%)	23 (29%) 26 (33%)	1.2 (0.5 to 3)	15 (19%) 21 (27%)	27 (35%) 15 (19%)	0.4 (0.15 to 1)
High P+/S+b
High titer P/S+ High titer P/S−	17 (35%) 28 (58%)	1 (2%) 2 (4%)	1.2 (0.1 to 14)	8 (17%) 11 (23%)	11 (23%) 18 (37%)	1.2 (0.4 to 3.9)	6 (12%) 13 (27%)	10 (21%) 17 (35%)	1 (0.3 to 3.4)

The 2×2 contingency table reports the association with OR and 95% CI between antibody titer and clinical outcome. Data are calculated from the subset of patients with nephrotic syndrome excluding the eight anti–THSD7A positive patients (n=177). The upper section shows the association between positivity versus negativity of each autoantibody at diagnosis with clinical outcome of proteinuria (complete (≤0.3 g/d) or partial (<3.5 g/d) remission) and eGFR after 12 mo. The lower section shows the association of high versus low levels of each antibody with the same parameters of proteinuria and eGFR. The additive effect of positivity for more than one antibody is indicated as P⁺/S⁺ and P⁺/S⁺/E⁺. THSD7A, thrombospondin type-1 domain containing 7A; PLA2R1, phospholipase A2 receptor 1; αENO, α-enolase; S, SOD; OR, odds ratio; 95% CI, 95% confidence interval; curve analysis.

^aIn these cases, P⁺/S⁺ were compared with P⁺/S⁻, and P⁺/S⁺/E⁺ were compared with P⁺/S⁻/E⁻.

^bIn this case, high P/S (high P titer and S positivity) were compared with high P titer and S negativity.

We further compared the added value of anti-SOD2 and anti-αENO over anti-PLA2R1 to predict clinical outcome using NRI and AUC-ROC analysis (38). In NRI analysis, the association of two antibodies allowed the reclassification of patients who developed or maintained proteinuria <3.5 g/d after 12 months in 34% and 22% of cases negative for both PLA2R1 and SOD2 or αENO, respectively (Table 4). The same association allowed reclassification of eGFR >60 ml/min per 1.73 m² in 27% and 24% of patients at 12 months, respectively (Table 4). AUC-ROC analysis discriminated the outcome after 12 months in respect to the titer of a single antibody or its association. In this case, the association of more than one antibody slightly increased the discrimination for patients who developed eGFR <60 ml/min per 1.73 m² (Table 4).

Table 4.

Two-category net reclassification improvement and area under the receiver-operating characteristic curve analysis for predictors of clinical outcome after 12 mo in membranous nephropathy patients by adding intracellular autoantibodies to anti-phospholipase A2 receptor 1 positivity

Parameter	Proteinuria >0.3 g/d	Proteinuria >3.5 g/d	eGFR<60 ml/min per 1.73 m2
Events (n)	209	74	98
Nonevents (n)	68	203	179
Two-category NRI (n)
Anti-PLA2R1 plus anti-SOD2
NRIevents(%)	16	1	−2
NRInonevents(%)	−35	34	27
NRIoverall(%)	−19	35	25
P value	0.002	<0.001	<0.001
Anti-PLA2R1 plus anti-αENO
NRIevents(%)	13	1	13
NRInonevents(%)	−37	22	24
NRIoverall(%)	−24	23	37
P value	0.003	<0.001	<0.001
AUC-ROC (95% CI)
Anti-PLA2R1	0.66 (0.6 to 0.66) P=0.001	0.60 (0.50 to 0.7) P=0.02	0.56 (0.49 to 0.63) P=0.09
Anti-SOD2	0.60 (0.52 to 0.67) P=0.02	0.60 (0.51 to 0.7) P=0.02	0.56 (0.50 to 0.64) P=0.09
Anti-αENO	0.57 (0.5 to 0.65) P=0.09	0.60 (0.52 to 0.7) P=0.02	0.54 (0.50 to 0.61) P=0.1
Anti-PLA2R1 plus anti-SOD2	0.61 (0.5 to 0.73) P=0.03	0.65 (0.54 to 0.71) P=0.02	0.61 (0.53 to 0.71) P=0.05
Anti-PLA2R1 plus anti-αENO	0.68 (0.53 to 0.82) P=0.03	0.63 (0.51 to 0.76) P=0.04	0.66 (0.52 to 0.76) P=0.02

Data are calculated from all patients excluding the eight anti-THSD7A positive patients (n=277). For NRI, combined negativity of anti-PLA2R1 and anti-SOD2 antibodies allowed the reclassification of 34% of patients with proteinuria <3.5 g/d (nonevent) and 27% of cases with eGFR >60 ml/min per 1.73 m² (nonevent) after 12 mo. Similar results were found for combined negativity of anti-PLA2R1 and anti-αENO. In both cases, the reclassification improvement was in comparison with anti-PLA2R1 positivity alone. NRI, net reclassification improvement; PLA2R1, phospholipase A2 receptor 1; αENO, α-enolase; AUC-ROC, area under the receiver-operating characteristic curve analysis; 95% CI, 95% confidence interval; THSD7A, thrombospondin type-1 domain containing 7A.

Stratification of Patients Based on Combined Positivity for Multiple Autoantibodies and Association with Clinical Outcome

Based on combined positivity for autoantibodies against PLA2R1 and intracellular antigens at diagnosis and excluding patients who were THSD7A⁺, we stratified the remaining 277 patients into four groups (Table 1). In total, 118 patients were anti-PLA2R1⁺/intracellular⁺ (i.e., with additional single or multiple positivity for autoantibodies against intracellular antigens), 64 patients were anti-PLA2R1⁺/intracellular⁻, 64 patients were anti-PLA2R1⁻/intracellular⁺, and 31 patients were anti-PLA2R1⁻/intracellular⁻. At diagnosis, the four groups did not differ for age or sex and had the same percentage of histology classes (Table 1). All patients received comparable therapy (Table 1). Proteinuria at diagnosis was higher in the two groups of patients who were anti-PLA2R1⁺ versus those who were anti-PLA2R1⁻ (Figure 3A, Table 1). Proteinuria after 12 months was lower and eGFR higher in patients who were anti-PLA2R1⁻/intracellular⁻ versus others (Figure 3, A and B, Table 1).

Figure 3.

Patients with single or multiple positivity for autoantibodies have different clinical outcome of proteinuria and eGFR.(A) Proteinuria and (B) eGFR values (data are presented as median with interquartile range [IQR]) at diagnosis and after 12 months of follow-up for patients divided in four groups based on the concomitant positivity or not for anti-PLA2R1 (PLA2R1⁺ or PLA2R1⁻) and anti-intracellular antigen antibodies (intracellular⁺ or intracellular⁻). Proteinuria (A) and eGFR (B) values are presented at diagnosis (open bars) and after 12 months (gray bars) for all patients (overall), for patients with nephrotic syndrome (NS) or not (non NS) at diagnosis and for patients who had been treated with an immunosuppressive treatment (IS-treated with cyclophosphamide, cyclosporine A, or rituximab) after 12 months of follow-up (black bars).

Differences in outcomes were estimated using Kaplan–Meier curves and the log-rank (Mantel-Cox) test and their hazard ratio. A total of 45% of patients who were anti-PLA2R1⁻/intracellular⁻ reached complete remission of proteinuria after 12 months, as compared with 16%, 16%, and 28% for the other groups (Figure 4A). Patients anti-PLA2R1⁻/intracellular⁺ had a worse clinical outcome as compared with those anti-PLA2R1⁻/intracellular⁻ (Figure 4B). Odds ratios and 95% CIs were utilized to determine the association with reduced eGFR after 12 months. Patients who were anti-PLA2R1⁺/intracellular⁺ and anti-PLA2R1⁻/intracellular⁺ had about four-fold higher risk of developing eGFR <60 ml/min per 1.73 m² after 12 months, as compared with patients who were PLA2R1⁻/intracellular⁻ (Figure 4C). We finally analyzed whether the combined positivity for PLA2R1 epitope spreading and intracellular antigens may be associated with clinical outcome (Supplemental Table 5). The concomitance of spreading and anti-αENO positivity was associated with low eGFR at diagnosis and after 12 months (Supplemental Figure 4), whereas eGFR at diagnosis in nonspreaders was only barely modified by positivity for anti-αENO autoantibodies (not shown).

Figure 4.

Patients with positivity for multiple autoantibodies have worse clinical outcome (complete remission and eGFR) than those all negative. (A) Kaplan–Meier curve showing the proportion of patients in the different groups, based on the concomitant positivity or not for anti-PLA2R1 (P⁺/P⁻) and anti-intracellular antibodies (IC⁺/IC⁻) who reached the endpoint of complete remission (proteinuria <0.3 g/d) after 12 months of follow-up. The Kaplan–Meier curve of the difference was calculated with the log-rank (Mantel-Cox) test and the hazard ratios with confidence intervals for the different groups. (B) Kaplan–Meier curve showing the proportion of patients in the different groups, based on the concomitant negativity for anti-PLA2R1 (P⁻) and anti-intracellular antigen antibody negativity or positivity (IC⁺/IC⁻) who reached the endpoint of reduction of kidney function (eGFR<60 ml/min per 1.73 m²) after 12 months of follow-up. (C) Odds ratios and confidence intervals were calculated for the association between the different groups and the risk of not reducing kidney function (eGFR<60 ml/min per 1.73 m²) after 12 months of follow-up.

Discussion

This is the largest retrospective study cohort of patients with membranous nephropathy with the combined detection of autoantibodies against the two major membrane-bound antigens, PLA2R1 and THSD7A, and the three intracellular autoantigens, AR, SOD2, and αENO. As expected, patients who were PLA2R1⁺ represent the majority of the cohort (64%), while those who were THSD7A⁺ represent only 3%, in line with recent studies (32,33,39). Our results confirmed that high titers of anti-PLA2R1 autoantibodies are associated with poor outcomes in terms of remission of proteinuria and eGFR (40–45). We also confirmed that epitope spreading was associated with reduced eGFR after 12 months (19,37).

In this study, we add to the above points that the presence of circulating anti-SOD2 and anti-αENO autoantibodies was independently associated with poor outcomes in terms of partial and complete remission of proteinuria and eGFR. Furthermore, anti-SOD2, anti-αENO, and anti-AR autoantibodies are not correlated to anti-PLA2R1 autoantibodies, and similar percentages of patients were positive for one or more intracellular antigens, irrespective of positivity (n=118, 65%) or not (n=64, 67%) for PLA2R1. Further analysis by Kaplan–Meier curves showed worse clinical outcomes for patients positive for anti-PLA2R1 and anti-SOD2 or anti-αENO, whereas NRI and AUC-ROC suggested that intracellular autoantibodies might have an added value over anti-PLA2R1 for refinement of association with clinical outcome. We thus conclude that the positivity for intracellular autoantibodies has added value over anti-PLA2R1 positivity to predict adverse clinical outcome. It remains to be seen if this might be also true for other membrane-bound antigens such as THSD7A or the new extracellular autoantigens, in particular NELL-1.

Based on the concomitant presence or absence of anti-PLA2R1 and anti-intracellular antigen autoantibodies, we stratified patients into four groups. Extending what was reported above for anti-PLA2R1 positivity, another novel finding is that patients who are negative for the two categories of autoantibodies have reduced proteinuria at diagnosis and a better clinical outcome after 12 months of follow-up, in comparison to other patients. Thus, in anti–THSD7A negative patients, the simultaneous measurement of the two categories of autoantibodies (anti-PLA2R1, and anti-SOD2 and anti-αENO) at baseline might be an additive factor to better evaluate clinical outcomes. In this setting, positivity to intracellular antigens may represent per se a negative factor associated with poor clinical outcome. However, we acknowledge limitations in our study including the use of a retrospective cohort with patients of different ages, different clinical characteristics and treatments, and homemade assays to measure autoantibodies to intracellular autoantigens. Our results should thus be confirmed in prospective studies and/or randomized clinical trials.

The second main finding reported here is that positivity for autoantibodies targeting the intracellular autoantigens, especially anti-αENO autoantibodies, is maximal in PLA2R1⁺ patients with high anti-PLA2R1 titer or those defined as spreaders. The combined positivity associates with a reduction of eGFR, being worse in patients with a high anti-PLA2R1 titer (Table 1) and in spreaders (Supplemental Figure 4) positive for anti-αENO autoantibodies. The lower positivity of anti-αENO autoantibodies or other intracellular autoantigens in patients with low anti-PLA2R1 titer or defined as nonspreaders suggests the formation of autoantibodies against intracellular antigens represents a secondary phenomenon that occurs after the formation of autoantibodies against PLA2R1 (and possibly THSD7A) as primary triggers of membranous nephropathy. We thus propose that anti-SOD2 and anti-αENO autoantibodies may appear as a second wave of autoantibodies in the pathogenesis of membranous nephropathy. This hypothesis is also supported by the in vitro and in vivo data obtained in the mouse model of THSD7A-associated membranous nephropathy where SOD2 was induced after podocyte injury and in carbonic anhydrase II–associated membranous nephropathy (46,47).

Anti-SOD2 and anti-αENO autoantibodies are not unique to membranous nephropathy, and their occurrence has been described in lupus nephritis (27,48). However, the IgG subclass is different in membranous nephropathy and lupus nephritis, with IgG4 in membranous nephropathy but IgG2 in lupus nephritis. Therefore, the same autoantigen can trigger the formation of autoantibodies with different IgG isotypes depending on the type of disease. For lupus nephritis, it has been speculated that exposure of methyl-oxidized αENO and/or the DNA-oxidized αENO by neutrophils to Toll-like receptor 9 is a possible mechanism leading to the production of anti-αENO antibodies (48,49).

SOD2 is a key antioxidant enzyme playing a major role in protecting cells from oxidative injury (50). It is activated as the main antioxidant pathway after formation of autoimmune deposits in glomeruli. Formation of anti-SOD2 autoantibodies may modify the antioxidant protective effect of SOD2 and worsen the clinical outcome.

αENO is a glycolytic enzyme that catalyzes dehydration of 2-phospho-D-glycerate to phosphoenolpyruvate and produces pyruvate and ATP in the Embden-Meyerhof-Parnas pathway (51). αENO also has an extracellular localization where it functions as a plasminogen receptor (52). On binding to αENO, plasminogen is transformed into plasmin that degrades extracellular matrix components and activates metalloproteases (51,53). The presence of anti-αENO autoantibodies may reduce its anti-fibrotic role and favor the accumulation of ECM.

Overall, the results of this study show a multi-autoantibody composition in membranous nephropathy, with antibodies that may act in concert, leading to high proteinuria and low eGFR. Based on the possible functions of SOD2 and αENO after podocyte injury both in vitro and in vivo (46,47), we propose a multi-hit pathogenic mechanism of membranous nephropathy that first involves the formation of anti-PLA2R1 (or anti-THSD7A or autoantibodies to another similar “primary” autoantigen), that then progresses by a mechanism of both intramolecular (such as within PLA2R1) and intermolecular epitope spreading along which (1) the anti-PLA2R1 titer increases and (2) is exacerbated by a second wave of autoantibodies targeting intracellular autoantigens, leading for instance to the production of anti-SOD2 and/or anti-αENO autoantibodies. In patients negative for anti-PLA2R1 and anti-THSD7A autoantibodies, an autoantibody targeting a still unknown membrane-bound autoantigen would take a role similar to anti-PLA2R1 or anti-THSD7A antibodies, possibly represented by patients positive for NELL-1, exostosins, or other unknown antigens (29,30).

In conclusion, our study first confirms the relevance of detecting and measuring anti-PLA2R1 (and anti-THSD7A) autoantibodies as fundamental tools to diagnose and identify patients with membranous nephropathy and at risk of severe disease. The study further shows the added value of assessing autoantibodies targeting intracellular podocyte antigens, combined or not with positivity for the above membrane-bound autoantigens. Altogether, our findings pave the way for more personalized medicine in membranous nephropathy.

Disclosures

G. Dolla has a patent on “prognosis and monitoring of membranous nephropathy based on the analysis of PLA2R1 epitope profile and spreading” pending. G. Ghiggeri is inventor on a patent on the “use of anti-SOD2 and anti-ENO antibodies as biomarkers in membranous nephropathy” (EP 3097419). G. Lambeau has a patent on “diagnosis for membranous nephropathy,” with royalties paid to Euroimmun; a patent on “methods and kits for monitoring membranous nephropathy,” with royalties paid to Euroimmun; and a patent on “prognosis and monitoring of membranous nephropathy based on the analysis of PLA2R1 epitope profile and spreading” issued. M. Prunotto is an employee of Galapagos Ltd. B. Seitz-Polski has a patent on “methods and kits for monitoring membranous nephropathy” EP14306195.0 and a patent on “prognosis and monitoring of membranous nephropathy based on the analysis of PLA2R1 epitope profile and spreading” pending. All remaining authors have nothing to disclose.

Funding

The Institute Giannina Gaslini, Scientific Institute for Research, Hospitalization and Healthcare (trial sponsor) had provided logistic and financial support to the study through a grant from the Ministero della Salute (Ricerca Corrente). People working on the membranous nephropathy project belong to the Fondazione Malattie Renali del Bambino, from which we acknowledge financial support. V. Brglez reports receiving grants from Fondation de la Recherche Médicale, during the conduct of the study. G. Ghiggeri received a grant from Compagnia di San Paolo (ROL 9849). J. Justino reports receiving grants from Investments for the Future Laboratory of Excellence SIGNALIFE and Fondation de la Recherche Médicale during the conduct of the study. G. Lambeau reports receiving grants from the National Research Agency (France) during the conduct of the study. C. Zaghrini reports receiving grants from Investments for the Future Laboratory of Excellence SIGNALIFE, during the conduct of the study. This work was supported by grants from the Fondation Maladies Rares (LAM-RD_20170304, G. Lambeau), the National Research Agency (grants MNaims ANR-17-CE17-0012-01, G. Lambeau) and Investments for the Future Laboratory of Excellence SIGNALIFE, a network for innovation on signal transduction pathways in life sciences (ANR-11-LABX-0028-01) with allocated fellowships for C. Zaghrini and J. Justino) and the Fondation pour la Recherche Médicale (FRM ING20140129210, SPF20150934219, DEQ20180339193, and FDT201805005509 to G. Lambeau, V. Brglez, and J. Justino).

Supplemental Material

This article contains the following supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.02500220/-/DCSupplemental.

Supplemental Methods. Recombinant proteins. Antibodies. Dot-blot for anti–aldose reductase, anti-SOD2 and anti–α-enolase autoantibodies. Anti-phospholipase A2 receptor 1 epitopes.

Supplemental References.

Supplemental Table 1. Epidemiology, histological stage, clinical characteristics, and treatment of patients with membranous nephropathy versus controls (healthy donors) at diagnosis and after 12 months of follow-up.

Supplemental Table 2. Quantitative data for limits of positivity and distribution of serum levels for each autoantibody.

Supplemental Table 3. Detailed autoantibody positivity for intracellular antigens in nonspreaders and spreader patients who are phospholipase A2 receptor 1⁺.

Supplemental Table 4. Two-way contingency table showing the association of antibody levels with indexes of kidney outcome and the interaction between autoantibodies against phospholipase A2 receptor 1 and intracellular antigens for patients treated with cytotoxic drugs, cyclosporine A, or rituximab.

Supplemental Table 5. Clinical characteristics at diagnosis and after 12 months of follow-up for patients who are phospholipase A2 receptor 1⁺ stratified according to epitope profiles.

Supplemental Figure 1. Prognosis clinical factors associated with complete remission.

Supplemental Figure 2. Circulating levels of the various autoantibodies at diagnosis and during follow-up.

Supplemental Figure 3. Multi-autoantibody composition in the cohort of patients with membranous nephropathy.

Supplemental Figure 4. Kidney function (eGFR) in spreaders who were positive (n=63) or not (n=84) for anti–α-enolase autoantibodies.