Multicentre comparative study of neural epidermal growth factor-like 1 protein vs M-type phospholipase A2 receptor-associated membranous nephropathy

Abstract

BACKGROUND

The understanding of membranous nephropathy (MN) has significantly evolved with the discovery of novel antigenic targets. M-type phospholipase A2 receptor-associated MN (PLA2R-MN) is the most common antigenic variant, followed by neural epidermal growth factor-like 1 protein-associated MN (NELL1-MN). Few studies have directly compared the outcomes of these two variants.

AIM

To compare the outcomes of NELL1-MN with PLA2R-MN.

METHODS

In this retrospective cohort study, we compared the outcomes of NELL1-MN and PLA2R-MN using time-to-event analysis. Rate of remission - complete or partial - was the primary outcome of interest.

RESULTS

The 17 patients diagnosed with NELL1-MN during the study period were compared with 24 patients with PLA2R-MN diagnosed during the same month. Antecedent exposure to traditional indigenous medicines was more common in NELL1-MN than PLA2R-MN (70.6% vs 13%, P < 0.001). 11.8% with NELL1-MN had rheumatoid arthritis while none with PLA2R-MN had the same (P = 0.08). NELL1-MN showed a significantly higher remission-rate than PLA2R-MN (11.7 per 100-patient-months vs 5 per 100-patient-months, P = 0.01) and a shorter time-to-remission than PLA2R-MN (median: 7 months vs 13 months, P = 0.001). In multivariate Cox-regression, NELL1-MN was independently associated with a higher likelihood of remission compared with PLA2R-MN (hazard ratio = 2.65; 95% confidence interval: 1.25-5.64; P = 0.01). Spontaneous remission-rate was higher for NELL1-MN than PLA2R-MN (3.4 per 100-patient-months vs 0.3 per 100-patient-months, P = 0.01). Resistant nephrotic syndrome was more common among PLA2R-MN than NELL1-MN (2.5 per 100-patient-months vs 0 per 100-patient-months, P = 0.04).

CONCLUSION

NELL1-MN has a favourable prognosis compared to PLA2R-MN. Antigen-specific outcomes research could inform prognosis and guide treatment strategies in MN.

Key Words: Glomerulonephritis; Membranous nephroparhy; Nephrotic syndrome; Neural epidermal growth factor-like 1; M-type phospolipase A2 receptor

Core Tip: This retrospective cohort study is the first study that directly compared the outcomes of neural epidermal growth factor-like 1 protein-associated MN (NELL1-MN) with M-type phospholipase A2 receptor-associated MN (PLA2R-MN) using time-to-event analysis. NELL1-MN showed a favourable prognosis with a higher rate of remission, faster time-to-remission, more spontaneous remission and a lower rate of treatment-resistant nephrotic syndrome compared to PLA2R-MN. Additionally, this study also observed that NELL1-MN was more strongly associated with traditional indigenous medicine intake than PLA2R-MN. Antigen-specific outcomes research could inform prognosis and aid therapeutic decision-making.

INTRODUCTION

Membranous nephropathy (MN), the most common cause of nephrotic syndrome in adults, is characterised pathologically by immune complex deposition in the glomerular subepithelial space[1]. Traditionally, MN was classified as idiopathic or secondary depending on the presence or absence of an underlying disease condition[2]. The classification of MN has significantly evolved over the last two decades owing to the detection of novel target antigens. The histomorphology, disease association and clinical course of MN vary depending on the target antigen. Accordingly, an antigen-based classification of MN has been proposed by the Mayo Clinic group[3]. Among the growing list of identified antigenic targets, M-type phospholipase A2 receptor (PLA2R) and neural epidermal growth factor-like 1 protein (NELL1) predominate, accounting for approximately 55% and 10% of all MN cases, respectively[3].

PLA2R-associated MN (PLA2R-MN) was first described in 2009 and is characterised by global subepithelial deposits of predominantly immunoglobulin G4 antibodies directed against PLA2R antigen[4]. PLA2R-MN is most often idiopathic (70%)[4,5]; however, PLA2R antibody has been demonstrated in MN associated with hepatitis B virus (HBV), hepatitis C virus (HCV), and human immunodeficiency virus (HIV)[6] infections. The second most frequent antigen identified in MN is NELL1. NELL1-associated MN (NELL1-MN) is distinguished by focal subepithelial deposits of predominantly immunoglobulin G1 antibodies[7]. NELL1 is the target antigen in about 16% of idiopathic MN[7], and NELL1-MN is often associated with underlying conditions such as drugs[8], traditional indigenous medicines (TIM)[9], malignancy[10] and autoimmune disorders[11]. Given its rarity and recent recognition, information on NELL1-MN is incomplete and emerging. NELL1-MN generally has a favourable prognosis with a remission rate as high as 87%[12]. Few studies have directly compared the clinical outcomes of PLA2R-MN and NELL1-MN, particularly within developing country settings. We propose that antigen-specific outcomes research may inform antigen-guided therapeutic strategies, including decisions between supportive and immunosuppressive treatment approaches.

MATERIALS AND METHODS

Study design and setting

This retrospective cohort study was conducted across four tertiary care hospitals in Tamil Nadu, India, using an open cohort design. The case records of adult patients aged > 18 years who underwent kidney biopsy for nephrotic syndrome from January 2022 through June 2023 were screened for a diagnosis of NELL1-MN. The study was approved by the institutional ethics committee and conforms to the ethical principles of the Declaration of Helsinki.

Exposure of interest

Consecutive patients diagnosed with NELL1-MN during the study period were included in this study and compared with PLA2R-MN patients diagnosed during the same month. The distinction between NELL1-MN and PLA2R-MN was made by immunohistochemistry and corroborated by serologic testing for circulating PLA2R antibodies. Both primary and secondary MN were included. Patients were included only if they had at least six months of follow-up or were observed until remission, whichever occurred later.

Diagnostic workup and treatment protocol

Patients diagnosed with MN underwent a comprehensive evaluation for underlying disease conditions such as malignancy, viral infections - HBV, HCV and HIV, syphilis and autoimmune disorders as per the Kidney Disease: Improving Global Outcomes guidelines (2021)[13]. Following the same guidelines, patients were risk-stratified into low, moderate, high and very-high risk categories[13]. Given the increasing recognition of mercury toxicity and NELL1-MN, serum or urine mercury testing was offered to select patients with antecedent exposure to cosmetic creams or TIM[14-16]. Any patient who received TIM for at least 2 months within the 6 months prior to diagnosis of nephrotic syndrome was categorized as having significant TIM exposure. All patients received supportive treatment with the maximally tolerated dose of angiotensin converting enzyme inhibitor or angiotensin receptor blocker. Other supportive measures included sodium restriction, diuretics and targeted blood pressure management (< 120/80 mmHg)[17]. Patients deemed at high and very-high risk of disease progression were treated upfront with cyclical cyclophosphamide plus corticosteroids or rituximab or calcineurin inhibitor[13,17,18]. In secondary MN, treatment primarily focused on eliminating the causative factor and immunosuppressive therapy was commenced if nephrotic syndrome remained resistant to disease-directed therapy combined with three months of supportive treatment.

Data collection

Demographic profile, history, clinical findings and laboratory parameters were extracted from case records. History focused on the recent intake of TIM and prescribed medications like non-steroidal anti-inflammatory drugs. Case records were screened for rheumatoid arthritis, sarcoidosis, systemic lupus erythematous and psoriasis. Urine protein creatinine ratio (PCR), serum creatinine, estimated glomerular filtration rate (eGFR) and serum albumin at baseline were measured. Results of antinuclear antibody testing, cancer screening and viral serology (HBV, HCV, HIV) were noted.

Outcomes measured

The primary outcome of interest was the rate of remission - either partial or complete. Secondary outcomes included spontaneous remission rate and individual components of the primary outcome. Remission rates were measured using time-to-event analysis and reported as the number of events per 100-patient-months. Complete remission (CR) was defined as sustained normalisation of proteinuria, evident as urine PCR ≤ 0.3, stable serum creatinine and a serum albumin > 3 g/dL. Partial remission (PR) was defined as urine PCR > 0.3 and PCR ≤ 2, accompanied by at least 50% decline of proteinuria from baseline and no concomitant rise in serum creatinine. Spontaneous remission was remission (CR or PR) achieved with supportive therapy alone, obviating the need for immunosuppressive therapy. Resistant nephrotic syndrome was distinguished by failure to achieve CR or PR and/or progressive increase in serum creatinine despite immunosuppressive therapy.

Statistical analysis

Continuous variables were summarised using the mean and standard deviation, and the median and interquartile range. Categorical (binary) variables were reported as frequencies and proportions. Between-group comparisons were conducted using Pearson’s χ² test for proportions, the Wilcoxon rank-sum test for non-parametric continuous variables, and the t-test for normally distributed continuous variables. Time-to-event analysis was conducted with remission as the outcome event of interest (treated as the “failure” event within the survival-analysis framework). Loss to follow-up was censored. Remission probability (CR or PR) between NELL1-MN and PLA2R-MN was compared using Kaplan-Meier failure estimates and the log-rank test. Remission rates were expressed as events per 100-patient-months. Since we used an open cohort design where individual participants were allowed to have a variable length of follow-up and a censored outcome, time-to-event analysis best suited the comparison of outcomes between cohorts. Multivariable Cox-proportional-hazards model was used to compare remission probability between NELL1-MN and PLA2R-MN, adjusting for age, sex, baseline proteinuria, and baseline eGFR. Because the event of interest was remission - a favourable outcome - an adjusted hazard ratio (HR) > 1 indicates a beneficial association with the exposure, whereas a HR < 1 indicates a detrimental association; a HR of 1 denotes no association. A two-sided P-value < 0.05 was considered statistically significant. All statistical analyses were performed using the STATA version-17 software package.

RESULTS

During the study period, 17 patients were diagnosed with NELL1-MN, which were compared with 24 patients with PLA2R-MN who were diagnosed during the same month as NELL1-MN. A total of 41 MN cases were included. Median follow-up lasted for 16 months [interquartile range (IQR): 12-24]. The mean age of the sample was 42.6 ± 14.7 years, and 36.6% were men. A 14.6% had diabetes mellitus, and 34% had hypertension. The baseline median urine PCR was 8.3 (IQR: 4.7-12.5) and median eGFR was 107 mL/minute/1.73 m² (IQR: 95-119). The baseline characteristics of the two groups - NELL1-MN and PLA2R-MN - are compared in Table 1. Age, sex distribution, baseline urine PCR, eGFR, serum albumin and median follow-up were not significantly different between the groups. However, a significantly higher proportion of NELL1-MN patients had antecedent exposure to TIM compared to PLA2R-MN (70.6% vs 13%, P < 0.001). Two (11.8%) patients with NELL1-MN had rheumatoid arthritis, while none with PLA2R-MN had the same (P = 0.08). Both patients who had rheumatoid arthritis in association with NELL1-MN also had a history of exposure to TIM. Two (8.3%) with PLA2R-MN were diagnosed with malignancy, while none with NELL1-MN was diagnosed with malignancy during follow-up (P = 0.22). Neither of the groups had underlying HBV, HCV, HIV infections or systemic lupus erythematosus.

Table 1 Baseline characteristics of the cohort and treatment regimen followed, n (%).

Variable	Total (n = 41)	NELEL1-MN (n = 17)	PLA2R-MN (n = 24)	P value
Age (years), mean ± SD	42.6 ± 14.7	42.2 ± 16.9	42.8 ± 13.3	0.88
Male sex	15 (36.6)	4 (23.5)	11 (45.8)	0.14
Diabetes mellitus	6 (14.6)	1 (5.9)	5 (20.8)	0.18
Hypertension	14 (34.1)	6 (35.3)	8 (33.3)	0.90
TIM intake	15 (37.5)	12 (70.6)	3 (13.0)	< 0.001
Follow-up (months), median (IQR)	16 (12-24)	13 (12-19)	19.5 (12-24.5)	0.21
Disease association
Malignancy	2 (4.9)	0 (0)	2 (8.3)	0.22
Rheumatoid arthritis	2 (4.9)	2 (11.8)	0 (0)	0.08
Hypothyroidism	4 (10.5)	2 (13.3)	2 (8.7)	0.65
Psoriasis	1 (2.4)	1 (5.9)	0 (0)	0.23
HBV	0 (0)	0 (0)	0 (0)	-
HCV	0 (0)	0 (0)	0 (0)	-
HIV	0 (0)	0 (0)	0 (0)	-
SLE	0 (0)	0 (0)	0 (0)	-
Baseline laboratory parameters
Urine PCR (g/g), mean ± SD, median (IQR)	9.2 (5.2), 8.3 (4.7-12.5)	9.3 (5.9), 7.5 (6.9-11.7)	9.02 (4.8), 9.3 (4.4-13.2)	0.86, 0.91
eGFR (mL/minute/1.73 m2), mean ± SD, median (IQR)	106 (23.2), 107 (95-119)	112 (20.3), 117 (101-127)	102 (24.6), 102 (90-116)	0.17, 0.09
Serum albumin (g/dL), mean ± SD, median (IQR)	2.5 (0.7), 2.5 (1.9-3.1)	2.5 (0.9), 3 (1.7-3.4)	2.5 (0.6), 2.4 (2.1-3)	0.82, 0.67
Treatment followed
Supportive only	6 (16.6)	5 (29.4)	1 (4.1)	0.02
Cyclophosphamide + steroid	18 (43.9)	7 (41.2)	11 (45.8)	0.76
Rituximab	11 (26.3)	4 (23.5)	7 (29.2)	0.69
Calcineurin inhibitors	6 (14.7)	1 (5.9)	5 (20.3)	0.18

NELL1-MN: Neural epidermal growth factor-like 1 protein-associated membranous nephropathy; PLA2R-MN: M-type phospholipase A2 receptor-associated membranous nephropathy; IQR: Interquartile range; HBV: Hepatitis B virus; HCV: Hepatitis C virus; HIV: Human immunodeficiency virus; SLE: Systemic lupus erythematosus; PCR: Protein creatinine ratio; eGFR: Estimated glomerular filtration rate.

Serum mercury levels were measured in 9 of 17 patients with NELL1-MN, of whom four had toxic levels (> 7.5 μg/L). These patients received supportive care along with cessation of the offending agent as first-line therapy. The treatment regimen followed in both groups is summarized in Table 1. A greater proportion of patients with NELL1-MN achieved remission with supportive therapy alone compared with PLA2R-MN (29.4% vs 4.1%, P = 0.02). The proportion treated with rituximab and cyclophosphamide plus steroid was comparable between the groups. None of the patients in the cohort experienced death, doubling of serum creatinine or progression to end-stage kidney disease during follow-up. Among patients with PLA2R-MN, 2 (8.7%) experienced deep vein thrombosis of lower limbs, while none with NELL1-MN had venous thrombo-embolism (P = 0.21).

The primary and secondary outcomes are presented in Table 2. The incidence rate of the primary outcome - remission (CR or PR) - was 11.7 per 100-patient-months in NELL1-MN, which was significantly higher than 5.0 per 100-patient-months in PLA2R-MN (P = 0.01). All (100%) of NELL1-MN achieved remission during follow-up, while 66.6% of PLA2R-MN achieved remission (P = 0.008). The Kaplan-Meier remission estimates for the two groups, shown in Figure 1, revealed that the NELL1-MN group achieved remission faster than the PLA2R-MN group (P = 0.001). The median time to remission was 7 months for NELL1-MN and 13 months for PLA2R-MN (P = 0.001).

Figure 1 Kaplan-Meier remission probability for neural epidermal growth factor-like 1 protein-associated membranous nephropathy vs M-type phospholipase A2 receptor-associated membranous nephropathy. NELL1-MN: Neural epidermal growth factor-like 1 protein-associated membranous nephropathy; PLA2R-MN: M-type phospholipase A2 receptor-associated membranous nephropathy.

Table 2 Comparison of outcomes.

Outcome	Incidence rate (Total cohort)	Incidence rate (NELL1-MN)	Incidence rate (PLA2R-MN)	P value
Remission (Complete or partial)	7.2/100-pt-months	11.7/100-pt-months	5/100-pt-months	0.01
Complete remission	5.9/100-pt-months	10.3/100-pt-months	3.8/100-pt-months	0.01
Partial remission	1.3/100-pt-months	1.3/100-pt-months	1.2/100-pt-months	0.89
No remission	1.7/100-pt-months	0/100-pt-months	2.5/100-pt-months	0.04
Spontaneous remission	1.3/100-pt-months	3.4/100-pt-months	0.3/100-pt-months	0.01

NELL1-MN: Neural epidermal growth factor-like 1 protein-associated membranous nephropathy; PLA2R-MN: M-type phospholipase A2 receptor-associated membranous nephropathy; pt: Patient.

The NELL1-MN group had a significantly higher rate of CR than PLA2R-MN, while the PR rate was similar between groups (Table 2). Rate of spontaneous remission was higher among NELL1-MN compared to PLA2R-MN (3.4 per 100-patient-months compared to 0.3 per 100-patient-months, P = 0.01). A 29.4% of NELL1-MN achieved spontaneous remission compared to 4.1% of PLA2R-MN (P = 0.02). The incidence rate of resistant nephrotic syndrome (no remission) was significantly higher among PLA2R-MN compared with NELL1-MN (2.5 per 100-patient-months vs 0 per 100-patient-months, P = 0.04). Table 3 presents the univariate and multivariate associations between NELL1-MN, selected predictors, and remission. In multivariate Cox-regression analysis, NELL1-MN was independently associated with a higher likelihood of remission compared with PLA2R-MN (HR = 2.65; 95% confidence interval: 1.25-5.64; P = 0.01), after adjustment for age, sex, baseline urine PCR, and eGFR. Age, sex, urine PCR, and eGFR were not independently associated with remission in the multivariate model.

Table 3 Univariate and multivariate association of predictors with remission.

Predictor	Univariate HR1 (95%CI)	P value	Multivariate HR (95%CI)	P value
NELL1-MN2	2.98 (1.47-6.08)	0.003	2.65 (1.25-5.64)	0.01
Age	0.98 (0.96-1.01)	0.35	1.00 (0.97-1.04)	0.76
Male sex	0.54 (0.25-1.16)	0.12	0.66 (0.27-1.58)	0.35
Urine PCR	1.00 (0.94-1.08)	0.79	1.00 (0.93-1.07)	0.94
eGFR	1.01 (1.00-1.03)	0.04	1.01 (0.98-1.03)	0.39

¹Since the outcome is favourable (remission), hazard ratio (HR) > 1 indicates a beneficial association with exposure, HR < 1 indicates detrimental association and HR = 1 indicates no association.

²HR reported with M-type phospholipase A2 receptor-associated membranous nephropathy as comparator.

HR: Hazard ratio; CI: Confidence interval; NELL1-MN: Neural epidermal growth factor-like 1 protein-associated membranous nephropathy; PCR: Protein creatinine ratio; eGFR: Estimated glomerular filtration rate.

DISCUSSION

This study is the first comparative study of NELL1-MN and PLA2R-MN in which we have reported remission rates and multivariate HR for remission after adjusting for confounders. Although NELL1-MN and PLA2R-MN share overlapping histopathologic features, our findings demonstrate significant differences in their clinical course, response to immunosuppression, and disease association. In this study, NELL1-MN showed a favourable prognosis compared to PLA2R-MN, with 100% remission (CR or PR) and 29.4% spontaneous remission. This is in alignment with previous studies on NELL1-MN, which reported 85%-92% remission[12,19] and 26% spontaneous remission[12]. Additionally, our findings established a significantly shorter median time to remission for NELL1-MN (7 months) compared to PLA2R-MN (13 months). This provides a time-frame to anticipate remission and guide therapeutic decisions, especially choosing between watchful waiting and escalating/modifying treatment strategy.

A previous study compared NELL1-MN with unidentified antigen-associated MN and reported a remission rate of 81% for NELL1-MN and 68% for unidentified antigen-associated MN[20]. Our study is the first to compare NELL1-MN with PLA2R-MN and report adjusted HR for remission. An adjusted HR of 2.65 for remission observed in this study implies that the chance of remission was 2.65 times higher for NELL1-MN compared to PLA2R-MN. The higher rate of spontaneous remission in NELL1-MN as compared to PLA2R-MN (3.5 per 100-patient-months vs 0.3 per 100-patient-months) and the virtual absence of treatment-resistant nephrotic syndrome in NELL1-MN (0 per 100-patient-months vs 2.5 per 100-patient-months), once again exemplify the favourable outlook for patients with NELL1-MN. We hypothesise that the higher spontaneous remission in NELL1-MN observed in this study is attributable to its stronger association with TIM, withdrawal of which allowed remission. An Indian study observed that exposure to TIM was more associated with NELL1-MN (88%) than PLA2R-MN (9%)[9], and our findings reinforce this association (70.6% vs 13%).

Although this study was not powered to detect between-group differences in disease association, a notable finding was the association of NELL1-MN with rheumatoid arthritis (11.8% vs 0%, P = 0.08). The link between rheumatoid arthritis and NELL1-MN has been described in a few case series and observational studies[11,21]. A Japanese study found that 2 of 4 patients with NELL1-MN had underlying rheumatoid arthritis[21]. It is speculated that this association could be either direct or indirect, resulting from confounding by anti-rheumatic medications or TIM taken for rheumatoid arthritis. In the current study, both patients who had rheumatoid arthritis and NELL1-MN also had a history of exposure to TIM, implying a potential confounding by TIM intake. Other autoimmune disorders such as Hashimoto’s thyroiditis, cutaneous lupus and Sjogren’s syndrome have been linked to NELL1-MN in other observational studies[22].

In our study, malignancy was detectable in PLA2R-MN (8.3%) but not in NELL1-MN. A previous study observed that, though cancer was more common among NELL1-MN compared to PLA2R-MN, malignancy was not unique to NELL1-MN[10]. Among 111 patients with cancer-associated MN, 35 (31.5%) had PLA2R-MN and 30 (27%) had NELL1-MN, implying that the most common MN variant in malignancy was PLA2R-MN[10]. Our findings align with the 2021 Kidney Disease: Improving Global Outcomes recommendation for age-specific cancer screening in MN, irrespective of the antigenic variant[13].

This study was primarily intended to compare outcomes between the two groups and was not designed to delineate histological differences between them. This study had a modest sample size attributable to the rarity of NELL-MN and the relatively recent availability of NELL1 immunostaining. Despite an apparently small sample size, the study had enough statistical power (75%) to detect a difference in remission rates between the groups due to a large effect size (HR = 2.65). Likewise, the relatively recent availability of NELL1 immunostaining inherently contributed to short follow-up periods [median follow-up: 16 months for the whole cohort (Table 1)], limiting our ability to capture data on relapse rates. Although a strong association between NELL1-MN and antecedent exposure to TIM was observed, this study cannot establish a direct pathogenic role of TIM in NELL1-MN. However, the multicentre nature of the study, head-to-head comparative analysis and adjustment for confounders by regression techniques strengthen the observed results. Additionally, the use of time-to-event analysis enabled estimation of median time-to-remission and HR.

CONCLUSION

Synthesising these findings, it can be concluded that NELL1-MN has a benign and favourable prognosis compared to PLA2R-MN. Evaluation for underlying disease conditions in NELL1-MN should focus on TIM exposure and autoimmune disorders such as rheumatoid arthritis. Supportive therapy combined with elimination of suspected triggers such as TIM should constitute the first-line and core treatment strategy for NELL1-MN.