Abstract
Objective
To explore the predictive factors and establish a nomogram model for predicting relapse risk in primary membranous nephropathy (PMN).
Methods
The clinical, laboratory, pathological and follow-up data of patients with biopsy-proven membranous nephropathy were collected in the Affiliated Hospital of Qingdao University. A total of 400 PMN patients who achieved remission were assigned to the development group (n = 280) and validation group (n = 120) randomly. Cox regression analysis was performed in the development cohort to determine the predictive factors of relapse in PMN patients, a nomogram model was established based on the multivariate Cox regression analysis and validated in the validation group. C-index and calibration plots were used to evaluate the discrimination and calibration performance of the model respectively.
Result
Hyperuricemia (HR = 2.938, 95% CI 1.875–4.605, p < 0.001), high C-reactive protein (CRP) (HR = 1.147, 95% CI 1.086–1.211, p < 0.001), and treatment with calcineurin inhibitors with or without glucocorticoids (HR = 2.845, 95%CI 1.361–5.946, p = 0.005) were independent risk factors, while complete remission (HR = 0.420, 95%CI 0.270–0.655, p < 0.001) was a protective factor for relapse of PMN according to multivariate Cox regression analysis, then a nomogram model for predicting relapse of PMN was established combining the above indicators. The C-indices of this model were 0.777 (95%CI 0.729–0.825) and 0.778 (95%CI 0.704–0.853) in the development group and validation group respectively. The calibration plots showed that the predicted relapse probabilities of the model were consistent with the actual probabilities at 1, 2 and 3 years, which indicated favorable performance of this model in predicting the relapse probability of PMN.
Conclusions
Hyperuricemia, remission status, CRP and therapeutic regimen were predictive factors for relapse of PMN. A novel nomogram model with good discrimination and calibration was constructed to predict relapse risk in patients with PMN early.
Keywords: Primary membranous nephropathy, relapse, predictive factors, nomogram model
Introduction
Primary membranous nephropathy (PMN) is an autoimmune disease characterized by the thickening of the glomerular capillary walls due to the deposition of the immune complex and that is one of the common causes of adult nephrotic syndrome, and its prevalence has been increasing significantly in recent years [1,2]. Approximately one-third of untreated patients achieve spontaneous remission, one-third progress to end-stage renal disease (ESRD) over 10 years, and the others develop progressive chronic kidney disease (CKD). Most patients can achieve clinical remission after treatment, but relapse after remission is an important issue to be concerned about. Lower proteinuria level is significantly associated with a lower risk of reduction in renal function [3]. Persistent proteinuria and recurrence of PMN are independent risk factors for the development of CKD and progression to ESRD [4]. Patients who relapse may need to be treated with immunosuppressants with or without glucocorticoids again [5]. As we all know, immunosuppressants and glucocorticoids can lead to infection, gastrointestinal bleeding, elevated blood glucose and hypertension, and other complications. Therefore, it is essential to reduce the relapse rate of PMN. In recent years, the discovery of anti-phospholipase A2 receptor antibodies (PLA2R Abs) and other autoantibodies could predict response to therapy, relapse and the risk of progression to ESRD [6–8]. But some patients with PMN are PLA2R Abs negative, and the test of this item is not performed in all medical organizations. Meanwhile, other MN-related antigens such as thrombospondin type 1 domain-containing 7 A and other antigen-related antibodies have not been applied in clinical practice, so the application of PLA2R Abs to monitor MN relapse is still insufficient. When PLA2R Abs re-emergences or increases, it indicates that the patient may have clinically relapsed, rather than predicting relapse in advance to take intervention measures to prevent relapse. In recent years, nomogram prediction models have been used to predict the probability of outcome events by combining multiple factors to comprehensively determine the prognosis of diseases [9,10]. In this study, we aimed to identify the predictors of relapse in PMN patients, and construct a nomogram model identifying patients with high relapse risk early and guiding management to decrease relapse risk of PMN.
Methods
Study population
A total of 983 adult patients diagnosed with membranous nephropathy (MN) by renal biopsy in the Affiliated Hospital of Qingdao University from January 2013 to May 2021 were enrolled and followed up. The exclusion criteria included: (1) Secondary membranous nephropathy caused by autoimmune diseases (lupus nephritis, purpura nephritis, etc.), hepatitis, malignancy, drugs and other systemic diseases; (2) MN combined with another primary or secondary glomerulonephritis (diabetic nephropathy, hypertensive nephropathy, etc.); (3) Patients treated with immunosuppressive agents or corticosteroids before renal biopsy; (4) Patients were not followed up or followed up irregularly. After rigorous screening, 395 MN patients were excluded. Patients with PMN were followed up for more than 6 months after achieving remission, the follow-up endpoint was 30 November 2021 or relapse. Finally, 400 patients were enrolled in our study, they were divided into development group and validation group according to a ratio of 7:3 randomly (Figure 1). This study was conducted following the Declaration of Helsinki and was approved by the Medical Ethics Committee of the Affiliated Hospital of Qingdao University (the ethics approval number is QYFY WZLL 27505).
Figure 1.
Flow diagram of the whole study.
Date collection and therapy regimen
Patients’ demographic details, laboratory test results, pathological information and medication data were collected in detail by referring to their electronic medical records. The urine and venous blood samples of all participants were collected from the closest test before the renal biopsy. CTX + GCs referred to the regimen of cyclophosphamide combined with glucocorticoids, CNIs ± GCs referred to the application of calcineurin inhibitors with or without glucocorticoids. However, some patients were treatment-resistant to CTX + GCs or CNIs ± GCs, we changed the treatment regimen, therefore we defined CTX/CNIs ± GCs as using CTX + GCs and CNIs ± GCs alternatively in any order. Others: patients who were treated with mycophenolate mofitil, tripterygium wilfordii, and leflunomide with or without glucocorticoids, etc. All patients received optimal supportive care, including using ACEI/ARB, diuresis, anticoagulation, lipid-lowering therapy, etc.
Clinical outcomes
Complete remission (CR) was defined as 24h urinary protein excretion <0.3 g/24h, accompanied by serum albumin ≥40 g/L with a normal renal function. Partial remission (PR) was defined by 24h urinary protein excretion <3.5 g/24h plus a 50% or greater reduction from peak values, accompanied by an improvement or normalization of the serum albumin concentration and renal function remained stable (<15% decline in eGFR). No remission (NR) was defined as 24h urinary protein excretion ≥3.5 g/24 h or urinary protein reduction <50% from peak values, or eGFR reduction ≥15%. Relapse was defined as 24h urinary protein excretion >1 g/24h and greater than the highest level of albuminuria in the remission stage at least 1 time, or 24h urinary protein excretion greater than 3.5 g [3,11,12]. All the above indicators were confirmed by two values at least 1 week apart.
Statistical analysis and construction of the nomogram model
Continuous variables were expressed as median and interquartile range, rank data were represented by count and percentage, and the Mann–Whitney U test was performed to compare the two groups. For comparisons of categorical variables between groups, the Pearson chi-squared test or Fisher’s exact probability method was used, and the results were reported as count and percentage. Univariate Cox regression analysis was initially performed on each variable of PMN patients in the development group for primary selection. Multiple linear regression collinearity diagnostic tool was used to test the multicollinearity. Variables with p < 0.1 in the univariate Cox regression model and indicators with clinical significance were enrolled in the multivariate Cox regression model to predict the relapse probability of PMN patients. Finally, the multivariable Cox regression model was seriously chosen by using the backward LR method to select the significant predictors and prepare for the following nomogram model development. The discrimination and calibration of the models were evaluated by C-index and calibration plot respectively. P value <0.05 were considered statistically significant. Data analysis was conducted on the SPSS 26.0 and R-studio platform (version 4.0.4).
Results
A total of 983 adult patients with MN confirmed by biopsy were enrolled in our study, 395 patients were excluded according to the exclusion criteria, and 588 PMN patients were followed up at our centre. Of the 588 patients with PMN, 470 patients (79.93%) achieved remission after their first presentation. Four hundred PMN patients followed up for more than 6 months after achieving remission were included in the final study, 280 (70.00%) patients achieved CR and 120 (30.00%) PR. By the end of follow-up, there were 128 (32.00%) patients relapsed. The median follow-up time after remission was 16 (10–28) months. The cumulative relapse rates of 1, 2 and 3 years were 17.50%, 26.50% and 27.25%, respectively, and the median time of recurrence was 12 (8–19.75) months.
Cox regression analysis for predictors selection with PMN relapse
Research recipients were randomly allocated to the development (n = 280) and validation (n = 120) groups in accordance with a ratio of 7:3, ensuring that the data distribution of the development group and validation group was consistent. The basic characteristics of the recipients in the two groups were presented in Table 1. Compared to the non-relapsing groups in the development cohort, males and application of CNIs ± GCs made up higher proportions, the incidence of hyperuricemia was higher, while CR rate was lower. The levels of C-reactive protein (CRP), serum creatinine (Scr), urinary red blood cell count and serum IgE were higher, as shown in Table 2.
Table 1.
Comparison of characteristics between the development cohort and the validation cohort.
| Characteristic | Primary group (n = 400) | Development group (n = 280) | Validation group (n = 120) | P-value |
|---|---|---|---|---|
| Age (years) | 49.00 (39.00, 58.00) | 49.00 (38.25, 58.00) | 49.00 (40.00, 58.00) | 0.935 |
| Gender (male,n%) | 249 (62.25) | 184 (65.71) | 65 (54.17) | 0.029 |
| Hypertension,n (%) | 120 (30.00) | 83 (29.64) | 37 (30.83) | 0.812 |
| Diabetes mellitus,n (%) | 29 (7.25) | 16 (5.71) | 13 (10.83) | 0.199 |
| Hyperuricemia,n (%) | 166 (41.50) | 122 (43.57) | 44 (36.67) | 0.199 |
| Smoking,n (%) | 101 (25.25) | 75 (26.79) | 26 (21.67) | 0.280 |
| PLA2R Abs (RU/ml) | 15.68 (1.44, 89.02) | 16.60 (1.44, 93.79) | 11.25 (1.38, 83.70) | 0.793 |
| UTP (g/24h) | 4.02 (2.25, 6.35) | 4.18 (2.47, 6.62) | 3.60 (2.02, 6.30) | 0.134 |
| URBC(/ul) | 14.40 (3.63, 41.18) | 13.86 (3.96, 42.75) | 15.05 (3.30, 40.10) | 0.701 |
| Eosinophil count (*109/L) | 0.10 (0.05, 0.17) | 0.10 (0.06, 0.17) | 0.10 (0.05, 0.18) | 0.842 |
| Hemoglobin (g/L) | 138.00 (127.00, 149.00) | 139.50 (128.00, 149.00) | 135.00 (127.00, 147.00) | 0.107 |
| CRP (mg/L) | 1.66 (0.96, 2.30) | 1.66 (1.08, 2.42) | 1.60 (0.86, 1.81) | 0.068 |
| Total protein (g/L) | 47.90 (42.01, 53.63) | 47.80 (41.75, 52.81) | 48.16 (42.47, 55.26) | 0.254 |
| Albumin (g/L) | 25.32 (20.95, 29.83) | 25.05 (20.76, 29.83) | 26.10 (21.83, 29.83) | 0.246 |
| Blood urea nitrogen (mmol/L) | 5.16 (4.20, 6.15) | 5.12 (4.23, 6.26) | 5.17 (4.16, 6.05) | 0.815 |
| Serum creatinine (μmol/L) | 65.00 (52.00, 77.08) | 66.00 (52.25, 77.90) | 63.00 (52.00, 75.23) | 0.481 |
| eGFR (ml/min/1.73m2) | 105.46 (97.41, 114.13) | 105.82 (97.68, 115.39) | 104.92 (96.25, 113.27) | 0.442 |
| Triglyceride (mmol/L) | 1.89 (1.30, 2.66) | 1.92 (1.30, 2.71) | 1.87 (1.27, 2.61) | 0.659 |
| Total cholesterol (mmol/L) | 7.26 (5.98, 9.12) | 7.14 (6.00, 9.29) | 7.36 (5.88, 8.90) | 0.805 |
| Low-density lipoprotein cholesterol (mmol/L) | 4.62 (3.51, 6.00) | 4.60 (3.60, 6.09) | 4.73 (3.27, 5.82) | 0.449 |
| Serum-IgA (g/L) | 2.08 (1.60, 2.75) | 2.13 (1.59, 2.80) | 2.02 (1.62, 2.62) | 0.648 |
| Serum-IgM (g/L) | 0.98 (0.71, 1.36) | 0.99 (0.73, 1.33) | 0.90 (0.68, 1.41) | 0.838 |
| Serum-IgG (g/L) | 5.93 (4.41, 7.87) | 5.85 (4.46, 7.63) | 6.21 (4.30, 8.39) | 0.208 |
| Serum-IgE (IU/ml) | 79.62 (33.24, 161.78) | 81.78 (34.62, 163.90) | 70.70 (29.80, 154.13) | 0.312 |
| Serum-C3 (g/L) | 1.14 (0.99, 1.28) | 1.15 (0.99, 1.29) | 1.14 (0.99, 1.28) | 0.935 |
| Serum-C4 (g/L) | 0.27 (0.22, 0.32) | 0.27 (0.22, 0.32) | 0.26 (0.22, 0.31) | 0.329 |
| Pathological feature | ||||
| Churg’s stage | 0.266 | |||
| I, n (%) | 46 (11.50) | 29 (10.36) | 17 (14.17) | |
| II, n (%) | 346 (86.50) | 245 (87.50) | 101 (84.17) | |
| III, n (%) | 8 (2.00) | 6 (2.15) | 2 (1.67) | |
| IgG deposit | 0.096 | |||
| -, n (%) | 1 (0.25) | 1 (0.36) | ||
| 1+, n (%) | 20 (5.00) | 13 (4.64) | 7 (5.83) | |
| 2+, n (%) | 176 (44.00) | 116 (41.43) | 60 (50.00) | |
| 3+, n (%) | 203 (50.75) | 150 (53.57) | 53 (44.17) | |
| C3 deposit | 0.252 | |||
| -, n (%) | 129 (32.25) | 87 (31.07) | 42 (35.00) | |
| 1+, n (%) | 222 (55.50) | 155 (55.36) | 67 (55.83) | |
| 2+, n (%) | 49 (12.25) | 38 (13.57) | 11 (9.17) | |
| C1q deposit | 0.509 | |||
| -, n (%) | 276 (69.00) | 196 (70.00) | 80 (66.67) | |
| 1+, n (%) | 124 (31.00) | 84 (30.00) | 40 (33.33) | |
| PLA2R Ag | 0.576 | |||
| -, n (%) | 49 (12.25) | 30 (10.71) | 19 (15.83) | |
| 1+, n (%) | 109 (27.25) | 80 (28.57) | 29 (24.17) | |
| 2+, n (%) | 240 (60.00) | 168 (60.00) | 72 (60.00) | |
| 3+, n (%) | 2 (0.50) | 2 (0.71) | ||
| Glomerular IgG deposits | ||||
| IgG1, n (%) | 342 (85.50) | 236 (84.29) | 106 (88.33) | 0.292 |
| IgG2, n (%) | 302 (75.50) | 213 (76.07) | 89 (74.17) | 0.685 |
| IgG3, n (%) | 142 (35.50) | 101 (36.07) | 41 (34.17) | 0.715 |
| IgG4, n (%) | 373 (93.25) | 263 (93.93) | 110 (91.67) | 0.409 |
| ACEI/ARB | 377 (94.25) | 258 (94.85) | 119 (92.97) | 0.450 |
| Therapeutic regimen | 0.272 | |||
| CTX + GCs, n (%) | 93 (23.25) | 58 (20.71) | 35 (29.17) | |
| CNIs ± GCs, n (%) | 164 (41.00) | 120 (42.86) | 44 (36.67) | |
| CTX/CNIs ± GCs, n (%) | 57 (14.25) | 42 (15.00) | 15 (12.50) | |
| Others, n (%) | 86 (21.50) | 60 (21.43) | 26 (21.67) | |
| Complete remission, n (%) | 280 (70.00) | 194 (69.29) | 86 (71.67) | 0.634 |
| Follow-up time (months) | 16.00 (10.00, 28.00) | 16.00 (10.00, 27.75) | 17.00 (11.00, 28.75) | 0.412 |
The quantitative variable was displayed as median and interquartile ranges; categorical variable and rank variable were displayed as quantities and percentages.
BMI: body mass index; MAP: mean arterial pressure; PLA2R Abs: anti-phospholipase A2 receptor antibodies; UTP: 24-h urinary protein quantity; URBC: urinary red blood cell counts; CRP: C-reactive protein; GFR: estimated glomerular filtration rate; PLA2R Ag: glomerular phospholipase A2 receptor antigen; CTX + GCs: cyclophosphamide-glucocorticoids; CNIs ± GCs: calcineurin inhibitors with or without glucocorticoids; CTX/CNIs ± GCs: alternatively treat with CTX + GCs and CNIs ± GCs in any order.
Table 2.
Characteristics of patients between non-relapse group and relapse group in development group.
| Characteristic | Non-relapse group (n = 185) | Relapse group (n = 95) | P-value |
|---|---|---|---|
| Age (years) | 50.00 (42.00, 58.00) | 48.00 (36.00, 58.00) | 0.174 |
| Gender (male) | 112 (60.54) | 72 (75.79) | 0.011 |
| Hypertension, n (%) | 59 (31.89) | 24 (25.26) | 0.250 |
| Diabetes mellitus, n (%) | 11 (5.95) | 5 (5.26) | 0.816 |
| Hyperuricemia, n (%) | 57 (30.81) | 65 (68.42) | <0.001 |
| Smoking, n (%) | 44 (23.78) | 31 (32.63) | 0.113 |
| PLA2R Abs (RU/ml) | 18.19 (1.52, 98.47) | 5.95 (1.43, 84.61) | 0.390 |
| UTP (g/24h) | 4.16 (2.43, 6.75) | 4.28 (2.70, 6.18) | 0.815 |
| URBC (/ul) | 11.88 (2.64, 35.51) | 23.10 (7.26, 53.60) | 0.003 |
| Eosinophil count (*109/L) | 0.12 (0.07, 0.19) | 0.09 (0.04, 0.15) | 0.012 |
| Hemoglobin (g/L) | 139.00 (127.0, 150.00) | 140.00 (129.00, 149.00) | 0.858 |
| CRP (mg/L) | 1.66 (0.93, 1.66) | 2.42 (1.30, 3.42) | <0.001 |
| Total protein (g/L) | 47.90 (41.75, 53.73) | 46.82 (41.20, 51.82) | 0.571 |
| Albumin (g/L) | 25.60 (21.08, 29.96) | 24.30 (20.06, 28.72) | 0.285 |
| Serum creatinine (μmol/L) | 63.00 (50.20, 74.50) | 70.90 (59.00, 82.00) | 0.001 |
| eGFR (ml/min/1.73m2) | 106.55 (97.56, 116.86) | 104.67 (97.71, 113.37) | 0.359 |
| Triglyceride (mmol/L) | 1.91 (1.34, 2.66) | 1.93 (1.20, 2.82) | 0.449 |
| Total cholesterol (mmol/L) | 7.25 (5.90, 9.31) | 7.10 (6.17, 9.29) | 0.859 |
| Low density lipoprotein cholesterol (mmol/L) | 4.61 (3.33, 6.16) | 4.55 (3.72, 5.90) | 0.869 |
| Serum-IgA (g/L) | 2.19 (1.60, 2.92) | 2.00 (1.56, 2.55) | 0.247 |
| Serum-IgM (g/L) | 0.97 (0.74, 1.29) | 1.00 (0.72, 1.52) | 0.225 |
| Serum-IgG (g/L) | 5.78 (4.54, 7.26) | 6.23 (3.73, 8.06) | 0.696 |
| Serum-IgE (IU/ml) | 66.38 (31.99, 152.77) | 110.20 (41.60, 167.32) | 0.023 |
| Serum-C3 (g/L) | 1.14 (0.98, 1.28) | 1.16 (0.99, 1.33) | 0.488 |
| Serum-C4 (g/L) | 0.27 (0.23, 0.33) | 0.268 (0.22, 0.32) | 0.731 |
| Pathological feature | |||
| Churg’s stage | 0.126 | ||
| I, n (%) | 16 (8.65) | 13 (13.68) | |
| II, n (%) | 164 (88.65) | 81 (85.26) | |
| III, n (%) | 5 (2.70) | 1 (1.05) | |
| C3 deposit | 0.447 | ||
| -, n (%) | 61 (32.97) | 26 (27.37) | |
| 1+, n (%) | 99 (53.51) | 56 (58.95) | |
| 2+, n (%) | 25 (13.51) | 13 (13.68) | |
| C1q deposit | 0.130 | ||
| -, n (%) | 124 (67.03) | 72 (75.79) | |
| 1+, n (%) | 61 (32.97) | 23 (24.21) | |
| PLA2R Ag | 0.187 | ||
| -, n (%) | 23 (12.43) | 7 (7.37) | |
| 1+, n (%) | 54 (29.19) | 26 (27.37) | |
| 2+, n (%) | 107 (57.84) | 61 (64.21) | |
| 3+, n (%) | 1 (0.54) | 1 (1.05) | |
| Glomerular IgG deposits | |||
| IgG1, n (%) | 155 (83.78) | 81 (85.26) | 0.747 |
| IgG2, n (%) | 130 (70.27) | 83 (87.37) | 0.001 |
| IgG3, n (%) | 63 (34.05) | 38 (40.00) | 0.327 |
| IgG4, n (%) | 175 (94.59) | 88 (92.63) | 0.515 |
| ACEI/ARB | 175 (94.59) | 89 (93.68) | 0.756 |
| Therapeutic regimen | 0.256 | ||
| CTX + GCs, n (%) | 49 (26.49) | 9 (9.47) | |
| CNIs ± GCs, n (%) | 65 (35.14) | 55 (57.89) | <0.001* |
| CTX/CNIs ± GCs, n (%) | 32 (17.30) | 10 (10.53) | 0.297* |
| Others, n (%) | 39 (21.08) | 21 (22.11) | 0.015* |
| Complete remission, n (%) | 136 (73.51) | 58 (61.05) | 0.032 |
| Follow-up time (months) | 20.00 (12.00, 31.25) | 12.00 (8.00, 22.00) | <0.001 |
* Results of pairwise comparisons were performed with CTX + GCs as the reference category. Adjusted significance level α’=α/3, α’=0.05/3 = 0.017.
Indicators that may affect renal recurrence were included in univariate Cox regression analysis, and then factors including gender, hyperuricemia, urinary red blood cell count, CRP, Scr, eGFR, treatment regimen and CR or not were selected for further analysis (p < 0.1). We performed collinearity diagnosis for factors with p < 0.1 in the univariate Cox regression model, demographic data (such as age) and several indicators with clinical significance (such as PLA2R Abs). We found multicollinearity between age, Scr and eGFR, so we excluded age and Scr. Other variables were enrolled in the multivariate Cox regression model to predict the relapse probability of PMN patients, which confirmed that hyperuricemia (HR = 2.938, 95%CI 1.875–4.605, p < 0.001), high CRP (HR = 1.147, 95%CI 1.086–1.211, p < 0.001) and the use of CNIs ± GCs (HR = 2.845, 95%CI 1.361–5.946, p = 0.005) were independent risk factors, while CR (HR = 0.420, 95%CI 0.270–0.655, p < 0.001) was a protective factor affecting PMN relapse. The results of Cox regression analysis were shown in Table 3.
Table 3.
Univariate and Multivariate Cox regression analysis for predictors of relapse.
| Characteristic | Univariate Cox regression analysis |
Multivariate Cox regression analysis |
||
|---|---|---|---|---|
| HR (95%CI) | P-value | HR (95%CI) | P-value | |
| Age | 0.991 (0.975, 1.008) | 0.298 | – | – |
| Gender (male) | 1.561 (0.975, 2.498) | 0.064 | 1.447 (0.865, 2.419) | 0.159 |
| Hyperuricemia | 3.441 (2.220, 5.334) | <0.001 | 2.938 (1.875, 4.605) | <0.001 |
| PLA2R Abs | 1.000 (0.999, 1.001) | 0.903 | 1.000 (0.999, 1.001) | 0.414 |
| URBC | 1.003 (1.001, 1.005) | 0.002 | 1.002 (0.999, 1.004) | 0.146 |
| UTP | 0.999 (0.945, 1.056) | 0.971 | 0.972 (0.908, 1.040) | 0.411 |
| CRP | 1.164 (1.110, 1.221) | <0.001 | 1.147 (1.086, 1.211) | <0.001 |
| Serum creatinine | 1.027 (1.014, 1.039) | <0.001 | – | |
| eGFR | 0.986 (0.972, 1.001) | 0.070 | 0.997 (0.975, 1.020) | 0.814 |
| Eos | 0.216 (0.023, 1.994) | 0.176 | 0.229 (0.019, 2.809) | 0.249 |
| IgE | 1.000 (0.999, 1.001) | 0.412 | 1.001 (0.999, 1.002) | 0.343 |
| Pathological feature | ||||
| PLA2R Ag | 0.246 | 0.322 | ||
| - | 1 | 1 | ||
| 1+ | 0.515 (0.063, 4.219) | 0.536 | 1.865 (0.752, 4.632) | 0.178 |
| 2+ | 1.247 (0.168, 9.244) | 0.829 | 1.234 (0.532, 2.860) | 0.624 |
| 3+ | 1.049 (0.145, 7.582) | 0.962 | 0.723 (0.087, 6.030) | 0.764 |
| Therapeutic regimen | 0.000 | 0.013 | ||
| CTX + GCs | 1 | 1 | ||
| CNIs ± GCs | 4.838 (2.373, 9.866) | <0.001 | 2.845 (1.361, 5.946) | 0.015 |
| CTX/CNIs ± GCs | 1.552 (0.616, 3.915) | 0.351 | 1.269 (0.500, 3.222) | 0.617 |
| Others | 3.243 (1.482, 7.095) | 0.003 | 2.182 (0.967, 4.923) | 0.060 |
| Complete remission | 0.359 (0.233, 0.553) | <0.001 | 0.420 (0.270, 0.655) | <0.001 |
Construction and validation of the nomogram model
Factors with prognostic significance in the multivariate Cox regression analysis were utilized to build the predictive model, and a nomogram was constructed to visualize the model finally, as shown in Figure 2. Then the nomogram was evaluated and verified in the development group and validation group. The C-indices of the nomogram were 0.777 (95%CI 0.729–0.825) and 0.778 (95%CI 0.704–0.853) in the development group and validation group, respectively. The calibration curves showed that the predicted probabilities and actual probabilities of this model were in good consistency at 1 year, 2 years and 3 years both in the development group and the validation group, as shown in Figure 3. To sum up, these results confirmed that the nomogram had favourable discrimination and calibration predicting the relapse probability of PMN patients.
Figure 2.
Nomogram for the prediction for IMN reccurence. HUA: Hyperuricemia; CRP: C-reactive protein; CR: complete remisssion.
Figure 3.
Calibration plots in the development and validation group of the nomogram. Calibration at 1 year in the development group (A) and validation group(B), Calibration at 2 years in the development group (C) and validation group(D), Calibration at 3 years in the development group (E) and validation group (F).
Discussion
PMN is a common primary glomerulonephritis in our country, and the incidence has been increasing in recent years. Most patients can achieve clinical remission whether supportive therapy or immunosuppressive therapy, but relapse after remission is still an important issue to be faced with. The results of this study showed that the cumulative relapse rates were 17.50%, 26.50% and 27.25% at 1, 2 and 3 years after remission respectively, and the median recurrence time was 12 (8–19.75) months. A multicenter study in Japan showed that the cumulative relapse rates of 1, 2 and 3 years were 8%, 22% and 30% respectively, and the median time of relapse was 1.59 (1.03–2.59) years [12]. Compared to their study, patients in our study had an earlier recurrence time, and a higher relapse rate in the first year after remission, but the cumulative relapse rates at 2 years and 3 years were similar.
Our study showed that hyperuricemia, CRP, remission status, and treatment regimen were predictors for relapse of PMN. Firstly, PMN patients with hyperuricemia had a higher risk of relapse in our study. Uric acid can induce systemic and local inflammatory responses, oxidative stress, and activation of the local renin–angiotensin system, leading to endothelial dysfunction and glomerular hypertension [13], which result in the emergence of proteinuria. Hyperuricemia is an independent risk factor for renal tubular atrophy and interstitial fibrosis [14,15], and high baseline blood uric acid level is an independent predictor of poor renal outcome in PMN patients [15]. Therefore, it is of great significance to pay attention to the management of uric acid in clinical practice, the treatment of lowering uric acid should be given at the same time while treating the primary disease. Our study showed that baseline CRP levels in adult PMN with relapse are higher, CRP levels in children with relapsing nephrotic syndrome were higher than those in non-relapsing patients before treatment [16]. However, there are fewer studies in adults. CRP can reflect the micro-inflammatory state in the body, leading to the damage of glomerular microvessels, causing kidney injury and the excretion of proteinuria [17]. Further studies are required to determine the relationship between baseline CRP and relapse of PMN. Studies have shown that patients with CR have a lower risk of recurrence than those with PR [18, 19], our result was consistent with the findings of some previous studies that CR was a protective factor for relapse of PMN. Similar to previous studies, our research showed that the relapse rate of using CNIs ± GCs was higher than those of using CTX + GCs [20]. A randomized controlled trial showed that 40% of patients treated with tacrolimus relapsed, while only 7% of patients treated with CTX relapsed after remission at 24 months [21]. For a long time, CTX based regimens have been the standard of treatment because they have been proven to be able to prevent the occurrence of ESRD and more effective in high-risk patients [22,23]. However, there are many disadvantages of CTX including relatively high incidence of adverse events, such as bone marrow suppression, liver function injury, infertility and malignant tumors, and so on [24,25]. CNIs have immunosuppressive effects, which can directly act on kidney podocytes, thereby reducing proteinuria and inducing a higher remission rate [26], but high recurrence rate and nephrotoxicity are concerns for long-term treatment [27,28]. Therefore, various considerations should be taken into the selection of therapeutic schedule, and individualized cautious management should be carried out in the process of drug reduction and withdrawal.
In addition, a recent study showed that hematuria increased the risk of recurrence of MN, that a high level of initial hematuria was related to a higher risk of relapse and aggravation of hematuria significantly increased the risk of short-term recurrence [29]. However, in our study, univariate Cox regression analysis showed that urinary red blood cell count was correlated with recurrence, but the difference was not statistically significant after adjusting for confounders by multivariate analysis. Hematuria of glomerular origin was a marker of inflammation and pathological damage in nephridial tissue, which may be caused by hemoglobin or iron released from red blood cells, resulting in direct tubular cell injury, oxidative stress and pro-inflammatory cytokine production [30]. More researchers are needed to investigate the relationship between hematuria and PMN relapse. PLA2R Ab is a MN-specific antibody found in about 70% of adult MN patients [31]. The discovery of PLA2R Ab has promoted the paradigm shift change in diagnosis, outcome prediction and treatment monitoring for MN [32]. A high level of PLA2R Abs is associated with poor treatment response and adverse clinical outcome [33,34]. However, a meta-analysis showed that serum PLA2R Abs titer at baseline was not associated with relapse, but the relapse rate of the glomerular PLA2R Ag positive group was higher than that of negative group, which may indicate that glomerular PLA2R Ag deposition is associated with disease relapse rather than circulating antibodies [35]. Similarly, Cox regression analysis showed that the titer of PLA2R Abs was not a risk factor for relapse of MN, therefore PLA2R Abs was not incorporated into our prediction model. Further studies are needed to confirm the relationship between baseline PLA2R Ab titer and the recurrence of PMN.
In recent years, several nomogram models have been constructed to predict the recurrence of the disease early [36,37]. At present, there is no nomogram model for predicting the relapse probability of PMN. We established a nomogram model combining the four factors by analyzing the factors influencing the relapse of PMN based on the Cox regression analysis. By applying the individual clinical indicators of the patients, we can intuitively get the score of each influencing factor of the patient and calculate the total score through the nomogram, so as to get the recurrence probability of the patient and identify patients with high-risk. The discrimination of this nomogram model was good with the C-indices of 0.777 in the development group and 0.778 in the validation group respectively. The calibration plots showed that the probabilities of PMN recurrence predicted by this model were consistent with the actual probabilities of 1, 2 and 3 years.
This study has several limitations: First, it was a single-center retrospective study, and we only conducted internal validation of the model. Prospective multi-center studies for external validation are still needed before clinical application. Second, the sample size of this study was relatively small, and some patients refused renal biopsy for objective reasons resulting in the loss of some PMN patients, which led to research bias. In addition, MN is a chronic progressive disease, however, the follow-up time of our study was relatively short and only the recent risk of relapse could be observed. Further follow-up is required to investigate the risk of long-term relapse. Finally, the application of rituximab has been included in the guidelines, but there were only 3 patients who used rituximab before May 2021 in our center, so it is impossible to assess the effect of rituximab on relapse compared with other regimens. Further studies are still required to compare the effect of rituximab and other new treatments on recurrence in our center.
Conclusion
In summary, our study showed that hyperuricemia, CRP and treatment regimen were predictive indicators influencing relapse of PMN, we constructed a nomogram model with excellent discrimination and accurate calibration for predicting the risk of PMN relapse. The relapse probability of each patient can be calculated by using the common clinical data according to the nomogram plot to identify high-risk patients. Therefore we can formulate individualized treatment and management plans to reduce the relapse rate and the risk of progression to CKD and ESRD.
Funding Statement
This work was supported by the Project of Science and Technology of Qingdao People’s Livelihood under Grant number 19-6-1-18-nsh.
Author contributions
Min Li: design of research, collection and analysis of data, and drafting of the manuscript. Huifang Wang and Xiaoying Lai: drafting and revision of the manuscript. Zixuan Fu and Chunhui Jiang: collection and collation of the data. Xuemei Liu: conception and design of the study and critical revision of the paper. All authors contributed to the article and approved the submitted version.
Ethics statement
This study was conducted according to the principles of the Declaration of Helsinki; informed consent was obtained from all subjects. This study was approved by the Medical Ethics Committee of the Affiliated Hospital of Qingdao University (the ethics approval number is QYFY WZLL 27505).
Disclosure statement of interest
The authors report there are no competing interests to declare.
References
- 1.Yokoyama H, Yamamoto R, Imai E, et al. Better remission rates in elderly Japanese patients with primary membranous nephropathy in nationwide real-world practice: the Japan nephrotic syndrome cohort study (JNSCS). Clin Exp Nephrol. 2020;24(10):1–10. [DOI] [PubMed] [Google Scholar]
- 2.Hu R, Quan S, Wang Y, et al. Spectrum of biopsy proven renal diseases in Central China: a 10-year retrospective study based on 34,630 cases. Sci Rep. 2020;10(1):10994. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Yamaguchi M, Ando M, Katsuno T, et al. Urinary protein and renal prognosis in idiopathic membranous nephropathy: a multicenter retrospective cohort study in Japan . Ren Fail. 2018;40(1):435–441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Huh H, Lee H, Lee JP, et al. Factors affecting the long-term outcomes of idiopathic membranous nephropathy. BMC Nephrol. 2017;18(1):104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.KDIGO 2021 clinical practice guideline for the management of glomerular diseases. Kidney Int. 2021;100(4S):S1–S276. [DOI] [PubMed] [Google Scholar]
- 6.De Vriese AS, Glassock RJ, Nath KA, et al. A proposal for a serology-based approach to membranous nephropathy. J Am Soc Nephrol. 2017;28(2):421–430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tesar V, Hruskova Z.. Autoantibodies in the diagnosis, monitoring, and treatment of membranous nephropathy. Front Immunol. 2021;12:593288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Stefan G, Stancu S, Zugravu A, et al. Negative anti-phospholipase A2 receptor antibody status at three months predicts remission in primary membranous nephropathy. Ren Fail. 2022;44(1):258–268. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ji L, Gao D, Hao Y, et al. Low-dose glucocorticoids withdrawn in systemic lupus erythematosus: a desirable and attainable goal. Rheumatology. 2022;62(1):181–189. [DOI] [PubMed] [Google Scholar]
- 10.Chen T, Zhou Y, Zhu J, et al. Prediction model of renal function recovery for primary membranous nephropathy with acute kidney injury. BMC Nephrol. 2022;23(1):247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chapter 7: idiopathic membranous nephropathy. Kidney Int Supp. 2012;2(2):186–197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Yamamoto R, Imai E, Maruyama S, et al. Incidence of remission and relapse of proteinuria, end-stage kidney disease, mortality, and major outcomes in primary nephrotic syndrome: the Japan nephrotic syndrome cohort study (JNSCS). Clin Exp Nephrol. 2020;24(6):526–540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Oh J, Won HY, Kang SM.. Uric acid and cardiovascular risk. N Engl J Med. 2009;360(5):539. [PubMed] [Google Scholar]
- 14.Liu B, Zhao L, Yang Q, et al. Hyperuricemia and hypertriglyceridemia indicate tubular atrophy/interstitial fibrosis in patients with IgA nephropathy and membranous nephropathy. Int Urol Nephrol. 2021;53(11):2321–2332. [DOI] [PubMed] [Google Scholar]
- 15.Zhang J, Pan M, Zhang J, et al. Serum uric acid is an independent predictor of renal outcomes in patients with idiopathic membranous nephropathy. Int Urol Nephrol. 2019;51(10):1797–1804. [DOI] [PubMed] [Google Scholar]
- 16.Wasilewska A, Zoch-Zwierz W, Tobolczyk J, et al. High-sensitivity C-reactive protein (hs-CRP) level in children with nephrotic syndrome. Pediatr Nephrol. 2007;22(3):403–408. [DOI] [PubMed] [Google Scholar]
- 17.Festa A, D’Agostino R, Howard G, et al. Inflammation and microalbuminuria in nondiabetic and type 2 diabetic subjects: the insulin resistance atherosclerosis study. Kidney Int. 2000;58(4):1703–1710. [DOI] [PubMed] [Google Scholar]
- 18.Lee T, Chung Y, Poulton CJ, et al. Serum albumin at partial remission predicts outcomes in membranous nephropathy. Kidney Int Rep. 2020;5(5):706–717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Thompson A, Cattran DC, Blank M, et al. Complete and partial remission as surrogate end points in membranous nephropathy. J Am Soc Nephrol. 2015;26(12):2930–2937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Qiu TT, Zhang C, Zhao HW, et al. Calcineurin inhibitors versus cyclophosphamide for idiopathic membranous nephropathy: a systematic review and meta-analysis of 21 clinical trials. Autoimmun Rev. 2017;16(2):136–145. [DOI] [PubMed] [Google Scholar]
- 21.Ramachandran R, Yadav AK, Kumar V, et al. Two-Year follow-up study of membranous nephropathy treated with tacrolimus and corticosteroids versus cyclical corticosteroids and cyclophosphamide. Kidney Int Rep. 2017;2(4):610–616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.van de Logt AE, Dahan K, Rousseau A, et al. Immunological remission in PLA2R-antibody-associated membranous nephropathy: cyclophosphamide versus rituximab. Kidney Int. 2018;93(4):1016–1017. [DOI] [PubMed] [Google Scholar]
- 23.van den Brand JA, van Dijk PR, Hofstra JM, et al. Long-term outcomes in idiopathic membranous nephropathy using a restrictive treatment strategy. J Am Soc Nephrol. 2014;25(1):150–158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ramachandran R, Hn HK, Kumar V, et al. Tacrolimus combined with corticosteroids versus modified ponticelli regimen in treatment of idiopathic membranous nephropathy: randomized control trial. Nephrology (Carlton). 2016;21(2):139–146. [DOI] [PubMed] [Google Scholar]
- 25.Khan S, Bolton WK.. Balancing cancer risk and efficacy of using cyclophosphamide to treat idiopathic membranous nephropathy. Clin J Am Soc Nephrol. 2014;9(6):1001–1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Faul C, Donnelly M, Merscher-Gomez S, et al. The actin cytoskeleton of kidney podocytes is a direct target of the antiproteinuric effect of cyclosporine A. Nat Med. 2008;14(9):931–938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cattran DC, Brenchley PE.. Membranous nephropathy: integrating basic science into improved clinical management. Kidney Int. 2017;91(3):566–574. [DOI] [PubMed] [Google Scholar]
- 28.Fervenza FC, Appel GB, Barbour SJ, et al. Rituximab or cyclosporine in the treatment of membranous nephropathy. N Engl J Med. 2019;381(1):36–46. [DOI] [PubMed] [Google Scholar]
- 29.He P, Yu X, Zha Y, et al. Microhematuria enhances the risks of relapse and renal progression in primary membranous nephropathy. Front Med (Lausanne). 2021;8:704830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Moreno JA, Sevillano A, Gutierrez E, et al. Glomerular hematuria: cause or consequence of renal inflammation?. Int J Mol Sci. 2019;20(9):2205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Beck LJ, Bonegio RG, Lambeau G, et al. M-type phospholipase A2 receptor as target antigen in idiopathic membranous nephropathy. N Engl J Med. 2009;361(1):11–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ruggenenti P, Fervenza FC, Remuzzi G.. Treatment of membranous nephropathy: time for a paradigm shift. Nat Rev Nephrol. 2017;13(9):563–579. [DOI] [PubMed] [Google Scholar]
- 33.Ruggenenti P, Debiec H, Ruggiero B, et al. Anti-phospholipase A2 receptor antibody titer predicts post-rituximab outcome of membranous nephropathy. J Am Soc Nephrol. 2015;26(10):2545–2558. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Dahan K, Debiec H, Plaisier E, et al. Rituximab for severe membranous nephropathy: a 6-month trial with extended Follow-Up. J Am Soc Nephrol. 2017;28(1):348–358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Rao SJ, Shen Q, Wang HM, et al. The association of anti-PLA2R with clinical manifestations and outcomes in idiopathic membranous nephropathy: a meta-analysis. Int Urol Nephrol. 2020;52(11):2123–2133. [DOI] [PubMed] [Google Scholar]
- 36.Ruan ZB, Liang HX, Wang F, et al. Influencing factors of recurrence of nonvalvular atrial fibrillation after radiofrequency catheter ablation and construction of clinical nomogram prediction model. Int J Clin Pract. 2022;2022:8521735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Yuan Y, Che X, Ni Z, et al. Association of relapse with renal outcomes under the current therapy regimen for IgA nephropathy: a Multi-Center study. PLoS One. 2015;10(9):e137870. [DOI] [PMC free article] [PubMed] [Google Scholar]