C3 glomerulonephritis and dense deposit disease share a similar disease course in a large United States cohort of patients with C3 glomerulopathy

clinical investigation

www.kidney-international.org

C3 glomerulonephritis and dense deposit disease share a similar disease course in a large United States cohort of patients with C3 glomerulopathy Andrew S. Bomback1, Dominick Santoriello2, Rupali S. Avasare3, Renu Regunathan-Shenk1, Pietro A. Canetta1, Wooin Ahn1, Jai Radhakrishnan1, Maddalena Marasa1, Paul E. Rosenstiel2, Leal C. Herlitz4, Glen S. Markowitz2, Vivette D. D’Agati2 and Gerald B. Appel1 1

Department of Medicine, Division of Nephrology, Columbia University College of Physicians and Surgeons, New York, New York, USA; Department of Pathology and Cell Biology, Division of Renal Pathology, Columbia University College of Physicians and Surgeons, New York, New York, USA; 3Department of Medicine, Division of Nephrology, Oregon Health and Science University, Portland, Oregon, USA; and 4Department of Anatomic Pathology, Cleveland Clinic, Cleveland, Ohio, USA 2

C3 glomerulonephritis (C3GN) and dense deposit disease comprise the two classes of C3 glomerulopathy. Studies from Europe and Asia have aided our understanding of this recently defined disorder, but whether these data apply to a diverse United States patient population remains unclear. We, therefore, reviewed clinical and histopathological data, including generation of a C3 Glomerulopathy Histologic Index to score biopsy activity and chronicity, to determine predictors of progression to end-stage renal disease (ESRD) and advanced chronic kidney disease (CKD) in 111 patients (approximately 35% non-white) with C3 glomerulopathy: 87 with C3GN and 24 with dense deposit disease. Complement-associated gene variants and autoantibodies were detected in 24% and 35% of screened patients, respectively. Our C3 Glomerulopathy Histologic Index denoted higher activity in patients with C3GN and higher chronicity in patients with dense deposit disease. Over an average of 72 months of follow-up, remission occurred in 38% of patients with C3GN and 25% of patients with dense deposit disease. Progression to late-stage CKD and ESRD was common, with no differences between C3GN (39%) and dense deposit disease (42%). In multivariable models, the strongest predictors for progression were estimated glomerular filtration rate at diagnosis (clinical variables model) and tubular atrophy/interstitial fibrosis (histopathology variables model). Using our C3 Glomerulopathy Histologic Index, both total activity and total chronicity scores emerged as the strongest predictors of progression. Thus, in a large, diverse American cohort of patients with C3 glomerulopathy, there is a high rate of progression to CKD and ESRD with no differences between C3GN and dense deposit disease. Kidney International (2018) 93, 977–985; https://doi.org/10.1016/ j.kint.2017.10.022

Correspondence: Andrew S. Bomback, Columbia University College of Physicians and Surgeons, Division of Nephrology, 622 West 168th Street, PH 4-124, New York, New York 10032. E-mail: [email protected] Received 8 August 2017; revised 5 October 2017; accepted 20 October 2017; published online 6 January 2018 Kidney International (2018) 93, 977–985

KEYWORDS: alternative complement pathway; C3 glomerulopathy; dense deposit disease; glomerulonephritis Copyright ª 2017, International Society of Nephrology. Published by Elsevier Inc. All rights reserved.

T

he term “C3 glomerulopathy” (C3G) denotes a glomerulonephritis with isolated or dominant C3 staining that implies an etiology rooted in dysregulation of the alternative complement pathway.1 This umbrella term encompasses both dense deposit disease (DDD, formerly known as membranoproliferative glomerulonephritis [MPGN] type 2) and C3 glomerulonephritis (C3GN, formerly known as MPGN type 1 or type 3 with isolated C3 staining). A genetic or acquired defect arising from rare genetic variants or autoantibodies that modulate activation of the C3 convertase of the alternative complement pathway can lead to a transformation from low-grade physiologic activity to unrestrained hyperactivity. This loss of alternative pathway control can result in a glomerulonephritis that on immunofluorescence stains only or dominantly for C3, consistent with mediation by deposition of complement proteins rather than immune complexes. The natural history of C3G, a rare disease with a reported incidence between 1 and 3 per million, is slowly being defined, and the heterogeneity among those diagnosed with C3G is striking. The clinical presentation can vary from asymptomatic hematuria and proteinuria with preserved renal function to the full nephrotic syndrome or a rapidly progressive glomerulonephritis.2,3 The most common histologic finding on light microcopy is membranoproliferative glomerulonephritis, although similar C3-dominant staining and deposits (presumably of complement components) in the mesangial, subepithelial, subendothelial, and intramembranous regions of the glomerulus have also been demonstrated in patients with mesangial proliferative or diffuse endocapillary proliferative patterns.4–6 Progression to end-stage renal disease (ESRD) has been reported in up to 50% of patients within the first 5 years of diagnosis, with DDD patients reportedly progressing at twice the rate of C3GN patients.2,3,7 977

clinical investigation

To date, the largest cohorts of C3G have emerged from centers in Japan,8 the United Kingdom,3 Spain,9 Turkey,10 Italy,11 and France.2 These reports have helped shape our current understanding of the natural history of C3G, but their generalizability to an ethnically diverse patient population in the United States remains unclear. In addition, these prior cohorts have often focused on either pediatric or adult populations, without taking into account the long disease course of many patients. Here, we present data on 111 pediatric and adult patients with C3G evaluated at the Center for Glomerular Diseases at Columbia University. This represents not only the largest C3G patient cohort in the United States but also the largest single-center experience reported worldwide. RESULTS

The cohort consisted of 111 patients, 87 with C3GN and 24 with DDD (Table 1). Patients with C3GN (mean age 28.3 years) were significantly younger than patients with DDD (mean age 40.0 years), with 36.8% of the C3GN subcohort considered pediatric (under 18 years) at the time of diagnosis versus 12.5% of the DDD subcohort. While whites comprised the majority (65%) of patients, Hispanic, Asian, and AfricanAmerican patients accounted for 19%, 10%, and 5% of the cohort, respectively. The most common clinical presentation was hematuria and proteinuria with preserved kidney function, although about one-quarter of the patients presented with significant chronic kidney disease (CKD) at the time of diagnosis, as seen in the median and interquartile range of creatinine values at the time of diagnosis across the cohort (1.3 mg/dl, 0.8–2.0 mg/dl). Creatinine values appeared similar between patients with C3GN and DDD, but due to the significant age difference in the subcohorts, the mean estimated glomerular filtration rate (eGFR), by both MDRD and CKD-EPI estimating equations, was higher in C3GN patients than DDD patients. The average proteinuria was 3.9 g/d for the entire cohort, but patients with C3GN were more likely than patients with DDD to present with full nephrotic syndrome. The prevalence of low complement levels was equal in C3GN and DDD patients, and we did not detect a difference in the rates of complement-associated gene variants or autoantibodies in the subgroup of patients (n ¼ 51) for whom such testing was available (Supplementary Table S1). Pediatric patients demonstrated a significantly higher prevalence of low C3 levels (83.3%) compared with adult patients (56.5%, P ¼ 0.01 for comparison) and demonstrated about twice the rate of detectable genetic variants and/or autoantibodies (24% vs. 13%), although this did not meet statistical significance (P ¼ 0.2 for comparison). Rates of antecedent infections, defined as self-reported infectious illness (viral or bacterial) in the 3 months prior to C3G diagnosis, did not differ between adult (17%) and pediatric (21%) patients. We did, however, find more than twice the prevalence of monoclonal paraproteins (57% vs. 25%) in DDD patients compared with C3GN patients in those tested for dysproteinemias. The 978

AS Bomback et al.: C3 glomerulopathy in the United States

Table 1 | Baseline clinical data of C3 glomerulopathy cohort Characteristic

C3GN (n [ 87)

DDD (n [ 24)

Age at diagnosis (yr) 28.3 40.0 Age groups (%) <18 years 36.8 12.5 18–50 years 47.1 58.3 >50 years 16.1 29.2 Male/female (%) 63.2/36.8 66.7/33.3 Race/ethnicity (%) White 63.2 70.8 Hispanic 19.5 16.7 Asian 12.6 8.3 African-American 4.6 4.2 Presentation (%) Hematuria and proteinuria, 40.7 41.7 preserved eGFR Nephrotic syndrome 32.6 16.7 AKI/RPGN 8.1 0 CKD with hematuria 18.6 41.7 and proteinuria Antecedent infection (%) 18.4 16.7 Creatinine at diagnosis (mg/dl) 2.0 2.1 eGFR at diagnosis (ml/min per 1.73 m2) MDRD (4-variable) formula 72.6 55.3 CKD-EPI formula 75.7 58.8 Proteinuria at diagnosis 3777 4410 (mg/g or mg/d) Low C3 at diagnosis (%) 64.9 63.6 Low C4 at diagnosis (%) 13.9 13.6 Complement factor gene variant 21.4 (9 of 42) 33.3 (3 of 9) (C3, CFH, CFI, CFB, CFHR5, MCP) identified (%)b 38.1 (16 of 42) 22.2 (2 of 9) Complement auto-antibody (C3Nef, factor H Ab, factor B Ab) identified (%)b 25.0 (6 of 24) 57.1 (8 of 14) Paraprotein detected in serum and/or urine (%)c

P valuea 0.03 0.05

0.8 0.9

0.05

0.8 0.9

0.1 0.1 0.5 0.9 1.0 0.5

0.4

0.05

AKI, acute kidney injury; C3GN, C3 glomerulonephritis; CKD, chronic kidney disease; DDD, dense deposit disease; eGFR, estimated glomerular filtration rate; RPGN, rapidly progressive glomerulonephritis. Categorical variables are presented as percentage of total group; continuous variables are presented as mean value. a Testing for differences was performed using a 2-sample Wilcoxon rank-sum (Mann-Whitney) test for continuous variables and Fisher exact test for categorical variables. b Testing for complement factor gene variants and autoantibodies was performed on a subset of patients in the cohort (N ¼ 51). More detailed results are available in Supplementary Table S1. c Testing for paraproteins in serum and/or urine was performed on a subset of patients in the cohort (N ¼ 38).

median (interquartile range) age in these 14 patients with detectable paraproteins was 53 (43–64) years, and 4 such patients had autoantibodies directed at the alternative complement pathway (2 with C3 nephritic factor and 2 with anti-Factor H). Biopsy materials (glass slides, immunofluorescence images, and electron microscopy scans) were available in 66 of the 111 cases for central re-review at Columbia University Medical Center; in the remaining 45 cases, histopathology data were extracted from the original biopsy reports (Table 2). The most common pattern on light microscopy was MPGN. In both the entire cohort and the subcohort with original biopsy materials re-reviewed, patients with DDD were more Kidney International (2018) 93, 977–985

clinical investigation

AS Bomback et al.: C3 glomerulopathy in the United States

Table 2 | Histopathology of entire C3 glomerulopathy cohort (N [ 111) and subcohort with biopsy re-read at Columbia University Medical Center (N [ 66) (A) Entire cohort Light microscopy pattern (%) MPGN Mesangial proliferative GN Diffuse endocapillary proliferative GN Diffuse sclerosing GN Tubular atrophy/interstitial fibrosis (estimated %) Cellular or fibrocellular crescents, any identified (%) Immunofluorescence (%) C3 alone C3 dominant (with trace or 1þ Ig) (B) CUMC subcohort Light microscopy pattern (%) MPGN Mesangial proliferative GN Diffuse endocapillary proliferative GN Diffuse sclerosing GN Globally sclerotic glomeruli (calculated %) Segmentally sclerotic glomeruli (calculated %) Exudative features (%) Cellular or fibrocellular crescents, any identified (%) Tubular atrophy/interstitial fibrosis (estimated %) Immunofluorescence (%) C3 alone C3 dominant (with trace or 1þ Ig)

P valuea

C3GN (n [ 87)

DDD (n [ 24)

68.8 17.5 8.8 5.0 28.4

45.8 29.2 12.5 12.5 23.4

0.4

10.3

16.7

0.4

46.8 53.2

61.9 38.1

Component (A) Activity score, 0–21 Mesangial hypercellularity

0.2 Endocapillary proliferation

Membranoproliferative morphology

0.3 Leukocyte infiltration

C3GN DDD P (n [ 46) (n [ 20) valuea 0.1 60.9 17.4 13.0 8.7 20.3 13.9

35.0 35.0 15.0 15.0 32.5 27.1

0.08 0.04

23.9 17.4

15.0 10.0

0.4 0.4

23.5

31.8

0.2 0.2

39.1 60.9

55.0 45.0

C3GN, C3 glomerulonephritis; CUMC, Columbia University Medical Center; DDD, dense deposit disease; GN, glomerulonephritis; MPGN, membranoproliferative glomerulonephritis. Data presented are percentage of total groups, unless otherwise noted as mean estimated or calculated percentages of sampled glomeruli or tubulointerstitial area. a Testing for differences was performed using a 2-sample Wilcoxon rank-sum (Mann-Whitney) test for continuous variables and Fisher exact test for categorical variables.

likely to have a diffuse sclerosing pattern on light microscopy. There was a trend toward greater likelihood of C3-alone staining (vs. C3-dominant staining) on immunofluorescence in DDD patients versus C3GN patients in both the entire cohort and the subcohort, although this did not reach statistical significance. We subsequently used original biopsy material to develop a C3G histologic index for the degree of disease activity and chronicity (Table 3). A semiquantitative scale of 0 to 3 was used for 7 markers of activity, allowing for an activity score of 0 to 21. For a chronicity score, ranging from 0 to 10, we used a semiquantitative scale of 0 to 3 for glomerulosclerosis, tubular atrophy, and interstitial fibrosis, and a scale of 0 to 1 for severity of arteriosclerosis. There were no statistically significant differences detected in any of the parameters of activity between C3GN and DDD patients (Table 4), although there was a trend for higher total activity among patients with C3GN (average activity score 9.1 out of 21) versus patients Kidney International (2018) 93, 977–985

Table 3 | C3 glomerulopathy histopathology index

Crescent formation

Definition

Score

% glomeruli with >3 mesangial cells per mesangial area

0 ¼ none 1 ¼ 1%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ none 1 ¼ 1%–25% 2 ¼ 26%–50% 3 ¼ >50%

% glomeruli with an increased number of cells within glomerular capillary lumina, causing luminal narrowing % glomeruli with GBM duplication with or without endocapillary proliferation % glomeruli with glomerular capillary infiltration by $3 neutrophils and/or macrophages % glomeruli with cellular and/ or fibrocellular crescents

Fibrinoid necrosis

% glomeruli with presence of $2 of fibrin, karyorrhexis, and GBM rupture

Interstitial inflammation

% cortical tubulointerstitial area with inflammation in non-fibrotic cortex

(B) Chronicity score, 0–10 Glomerulosclerosis % glomeruli with global and segmental sclerosis

Tubular atrophy

% cortical tubulointerstitial area involved with tubular atrophy

Interstitial fibrosis

% cortical tubulointerstitial area involved with interstitial fibrosis

Arterio- and arteriolosclerosis

Intimal thickening $ thickness of media

0 ¼ none 1 ¼ 1%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ none 1 ¼ 1%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ none 1 ¼ 1%–10% 2 ¼ 11%–25% 3 ¼ >25% 0 ¼ none 1 ¼ 1%–10% 2 ¼ 11%–25% 3 ¼ >25% 0 ¼ <10% 1 ¼ 10%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ <10% 1 ¼ 10%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ <10% 1 ¼ 10%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ <10% 1 ¼ 10%–25% 2 ¼ 26%–50% 3 ¼ >50% 0 ¼ absent 1 ¼ present

GBM, glomerular basement membrane.

with DDD (7.6). Patients with DDD demonstrated greater chronicity scores in all parameters evaluated compared with patients with C3GN, with total chronicity score averaging 6.0 (out of 10) compared with 4.1 (P ¼ 0.02). The mean duration of follow-up was 69 months for C3GN patients and 83 months for DDD patients (P ¼ 0.4). More than 75% of patients were treated with blockers of the renin-angiotensin-aldosterone system (RAAS). Eighty of the 111 patients were treated with at least 1 course of immunosuppression (Table 5), and there was no difference in this regard between C3GN and DDD patients (74% vs. 67% treated with immunosuppression). However, patients with C3GN vs. DDD were more likely to be treated with mycophenolate mofetil (44% vs. 17%, P ¼ 0.02) and demonstrated a 979

clinical investigation

AS Bomback et al.: C3 glomerulopathy in the United States

Table 4 | C3 glomerulopathy (C3G) histopathology index values in the C3G cohort Component (A) Activity scoreb Mesangial hypercellularity (0–3) Endocapillary proliferation (0–3) Membranoproliferative morphology (0–3) Leukocyte infiltration (0–3) Crescent formation (0–3) Fibrinoid necrosis (0–3) Interstitial inflammation (0–3) Total Activity Score (0-21) (B) Chronicity scorec Global and segmental glomerulosclerosis (0–3) Tubular atrophy (0–3) Interstitial fibrosis (0–3) Arterio- and arteriolosclerosis (0–1) Total chronicity score (0–10)

Table 5 | Treatment and outcome data of C3 glomerulopathy cohort

C3GN (n [ 46)

DDD (n [ 20)

P valuea

2.7 1.7 1.9

2.3 1.3 1.5

0.06 0.2 0.2

1.6 0.3 0.2 0.8 9.1

1.3 0.6 0.1 0.6 7.6

0.3 0.3 0.7 0.2 0.1

1.4

2.2

0.02

1.3 1.3 0.2 4.1

1.7 1.7 0.5 6.0

0.09 0.09 0.03 0.02

C3GN, C3 glomerulonephritis; DDD, dense deposit disease. a Testing for differences was performed using a 2-sample Wilcoxon rank-sum (MannWhitney) test. b For mesangial hypercellularity, endocapillary proliferation, membranoproliferative morphology, and leukocyte infiltration, we used a scale of 0 ¼ none, 1 ¼ 1-25%, 2 ¼ 26-50%, and 3 ¼ >50% involvement of sampled glomeruli. For crescent formation and fibrinoid necrosis, we used a scale of 0 ¼ none, 1 ¼ 1-10%, 2 ¼ 11-25%, and 3 ¼ >25% involvement of sampled glomeruli. For interstitial inflammation, we used a scale of 0 ¼ <10%, 1 ¼ 10-25%, 2 ¼ 26-50%, and 3 ¼ >50% based on percentage of involved cortical tubulointerstitial area. c Glomerulosclerosis, tubular atrophy, and interstitial fibrosis were assigned a score of 0 to 3 based on the percentage involvement (0 ¼ <10%, 1 ¼ 10-25%, 2 ¼ 26-50%, 3 ¼ >50%). For vascular disease, we assigned 0 if intimal thickening was < thickness of media and 1 if intimal thickening was $ thickness of media.

nonsignificant trend toward greater utilization of corticosteroids (70% vs. 50%, P ¼ 0.09). Seven of the patients in the cohort received eculizumab at some point in their disease course, of which 4 progressed to ESRD and 2 progressed to CKD stage 5. Complete or partial remission occurred more often in C3GN patients (38%) vs. DDD patients (25%), although this was not statistically significant (P ¼ 0.2, Figure 1). Likewise, a higher rate of complete remission was seen in C3GN (15%) versus DDD (8%), but this was not statistically significant (P ¼ 0.4). Nineteen of the 42 C3G patients treated with mycophenolate mofetil (MMF) achieved complete or partial remission, of which 18 received concomitant treatment with corticosteroids. Among those patients treated with MMF, we did not detect any clear predictors of response to treatment. C3 levels were similar among responders and nonresponders, with approximately half demonstrating depressed C3 levels at initiation of treatment. Fifteen of the 42 MMF-treated patients demonstrated elevated levels of serum membrane attack complex (C5b-9): 10 nonresponders versus 5 responders. Testing for C3 nephritic factor was positive in 4 of 14 nonresponders and 4 of 9 responders. Only 3 patients in this MMF subcohort had detectable genetic variants. Progression to late-stage CKD and ESRD requiring dialysis and transplantation was common in the cohort (Table 5), with no detectable differences between those with C3GN (39.1%) and DDD (41.7%) for our primary outcome of 980

Parameter (A) Treatment RAAS blockade: ACE-I, ARB, MRB Steroids Mycophenolate mofetil Calcineurin inhibitor Eculizumab Rituximab Cyclophosphamide Other immunosuppression Any immunosuppression (B) Outcomes Duration of follow-up (months) Remission (complete þ partial) Remission (complete) Doubling of creatinine (eGFR > 15 ml/min per 1.73 m2) Progression to CKD 5 (eGFR < 15 ml/min per 1.73 m2) Pre-emptive transplantation ESRD Death Combined outcome (doubling of creatinine, progression to CKD stage 5, ESRD, transplantation, death)

C3GN (n [ 87)

DDD (n [ 24)

P valuea

75.6 70.1 43.7 8.1 4.6 8.1 4.6 3.5 73.6

79.2 50.0 16.7 0.0 12.5 0.0 8.3 12.5 66.7

0.8 0.09 0.02 0.3 0.2 0.3 0.6 0.1 0.5

69.1 37.9 14.9 9.2

83.2 25.0 8.3 8.3

0.4 0.2 0.4 0.7

9.2

0.0

–

9.2 11.5 0.0 39.1

8.3 20.8 4.2 41.7

1.0 0.3 – 0.7

ACE-I, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; C3GN, C3 glomerulonephritis; CKD, chronic kidney disease; DDD, dense deposit disease; eGFR, estimated glomerular filtration rate; ESRD, end-stage renal disease; MRB, mineralocorticoid receptor blocker; RAAS, renin-angiotensin system. Data presented are percentage of total groups with the exception of duration of follow-up, given as mean months. a Testing for differences was performed using Fisher exact test for treatment history and univariate Cox proportional hazards model for outcomes of interest.

doubling of creatinine, progression to CKD stage 5, ESRD, transplantation, or death (Figure 2). In univariate analyses, we did detect a trend toward faster progression to this combined outcome in older patients compared with younger patients (Figure 3), but this distinction was mitigated in a multivariate model that adjusted for sex, race and/or ethnicity, presenting renal function and proteinuria, complement levels, presence of complement-associated gene variants and/or autoantibodies, and utilization of immunosuppression (Table 6). In this clinical variables–only model, the sole predictor of progression to late-stage CKD, ESRD, or death was kidney function at the time of presentation. In the second multivariate model based on histopathology variables only, the strongest predictor for our combined outcome was degree of tubular atrophy and/or interstitial fibrosis (each 10% increase in fibrosis imparting approximately twice the risk for progression). In this model, we additionally found that the presence of C3-only staining on immunofluorescence, as opposed to C3-dominant staining with trace to 1þ Ig, was associated with higher risk of progression (hazard ratio 5.09, 95% confidence interval 1.49–17.35, P ¼ 0.009). There was also a trend, not reaching statistical significance, for higher risk associated with the presence of cellular or fibrocellular crescents. Kidney International (2018) 93, 977–985

clinical investigation

AS Bomback et al.: C3 glomerulopathy in the United States

Figure 1 | Rates of remission (complete plus partial) did not differ among patients with C3 glomerulonephritis (C3GN) and dense deposit disease (DDD). Survival analysis excludes 1 patient for whom time to remission was unable to be verified (N ¼ 110). C3G, C3 glomerulopathy.

To further evaluate the contribution of biopsy markers of active versus chronic disease on risk for progression, we created a third multivariate model using the scores from our C3G histologic index. Both the total activity and chronicity scores emerged as significant predictors of our primary outcome (Table 6), highlighting the usefulness of this scoring system. Each 1-unit increase in total activity score increased the risk for doubling of creatinine, progression to CKD stage 5, ESRD, transplantation, or death by 18% (P ¼ 0.02), while each 1-unit increase in total chronicity score increased the risk for these outcomes by 59% (P < 0.001). In univariate analyses examining individual components of the activity score (Supplementary Table S2), crescent formation was the

Figure 2 | Progression to a combined outcome of end-stage renal disease, advanced chronic kidney disease (doubling creatinine or chronic kidney disease stage 5), or death did not differ among patients with C3 glomerulonephritis (C3GN) and dense deposit disease (DDD). Survival analysis excludes 3 patients for whom time to outcome was unable to be verified (N ¼ 108). C3G, C3 glomerulopathy. Kidney International (2018) 93, 977–985

Figure 3 | In univariate analysis, progression to end-stage renal disease, advanced chronic kidney disease (doubling creatinine or chronic kidney disease stage 5), or death was faster among adult patients diagnosed with C3 glomerulopathy than among pediatric patients. Survival analysis excludes 3 patients for whom time to outcome was unable to be verified (N ¼ 108).

only parameter to correlate with outcomes. In contrast, all the components of the chronicity score either demonstrated statistically significant association with the outcomes or a clear trend toward such association. DISCUSSION

C3G, a form of complement-mediated glomerulonephritis caused by dysregulation of the alternative complement pathway, is a heterogeneous disease with respect to etiology, clinical presentation, and prognosis. This retrospective cohort study examined presentation and disease course in 111 patients with C3G—87 patients with C3GN and 24 with DDD—followed over an average of 6 years from the time of diagnosis. We found an overall remission rate of 35% and progression to advanced CKD, ESRD, or death in 40% of C3G patients, with no detectable difference among those with C3GN versus DDD. Markers of chronicity at the time of diagnosis, whether they were clinical (i.e., elevated creatinine, reduced GFR, advanced age) or histologic (i.e., interstitial fibrosis, glomerulosclerosis), were the strongest predictors of outcome. Other than C3-alone versus C3-dominant staining on immunofluorescence microscopy, we found no detectable effect of complement testing on either remission or progression. Approximately 1 in 4 patients in our cohort progressed to ESRD requiring dialysis or transplantation, and another 16% progressed to late-stage CKD. This overall poor prognosis is similar to outcomes reported in cohorts from Japan,8 the United Kingdom,3 Spain,9 Turkey,10 Italy,11 and France.2 The rate of complement factor gene variants identified in our cohort (24%) was similar to the detection rate in the French cohort of 85 C3G patients (26%)2 and the Italian cohort of 73 C3G patients (18%).11 As in these other cohorts, these rare variants were more likely to be detected in younger patients; 981

clinical investigation

AS Bomback et al.: C3 glomerulopathy in the United States

Table 6 | Predictors of progression to primary outcome in C3 glomerulopathy using 3 multivariate models Predictor (A) Clinical variables model Female Age Less than 18 years 18-50 years Over 50 years Race/ethnicity White Hispanic Asian African-American eGFR via CKD-EPI equation (per 10 ml/min per 1.73 m2) Proteinuria at diagnosis (per 1 g/d) Low C3 and/or C4 Detectable complement abnormality (gene variant and/or antibody) Use of immunosuppression (B) Histopathology variables model Light microscopy pattern MPGN Mesangial proliferative GN Diffuse endocapillary proliferative GN Diffuse sclerosing GN Globally sclerotic glomeruli (per 10% increase) Segmentally sclerotic glomeruli (per 10% increase) Exudative features Cellular or fibrocellular crescents, any identified Tubular atrophy/interstitial fibrosis (per 10% increase) C3-only staining on immunofluorescence DDD by electron microscopy (vs. C3GN) (C) C3G histologic index model Total activity score Total chronicity score

Hazard ratio (95% CI)

P value

1.44 (0.53–3.92)

0.5

1.00 (reference) 0.91 (0.28–3.03) 1.19 (0.30–4.81)

N/A 0.9 0.8

1.00 1.29 4.01 3.24 0.69

(reference) (0.48–3.47) (0.81–19.77) (0.33–32.00) (0.57–0.83)

N/A 0.6 0.09 0.3 <0.001

1.03 (0.93–1.14) 1.35 (0.54–3.37) 0.57 (0.15–2.10)

0.6 0.5 0.4

0.87 (0.33–2.29)

0.8

1.00 1.16 0.27 0.93 0.69

N/A 0.9 0.3 0.9 0.1

(reference) (0.20–6.68) (0.02–3.57) (0.08–10.15) (0.44–1.08)

1.28 (0.99–1.66)

0.06

1.26 (0.30–5.31) 3.86 (0.93–15.97)

0.8 0.06

2.04 (1.33–3.15)

0.001

5.09 (1.49–17.35) 1.30 (0.31–5.38)

0.009 0.7

1.18 (1.03–1.34) 1.59 (1.26–2.01)

0.02 <0.001

C3G, C3 glomerulopathy; C3GN, C3 glomerulonephritis; CI, confidence interval; DDD, dense deposit disease; eGFR, estimated glomerular filtration rate; GN, glomerulonephritis; MPGN, membranoproliferative glomerulonephritis; N/A, not applicable.

the median age in our cohort was 19.5 years (interquartile range 10–22 years) for those identified as genetic variant carriers. Our finding that chronicity is a stronger outcome predictor than activity is consistent with data from 70 C3G patients in the United Kingdom and Ireland, among whom adult onset (defined as age >16 years) and renal dysfunction (defined as serum creatinine >1.5 mg/dl) were the only 2 clinical variables at the time of diagnosis associated with progression to ESRD.3 In the 14 patients in our cohort with identified dysproteinemia, the median age of 53 years is similar to the average age reported in earlier series on C3G and dysproteinemias published by Zand et al. (55 years)12 and Chauvet et al. (63 years).13 In important ways, however, our cohort differs from the previously published data on C3G patients. These differences likely reflect both the larger size and the geographical location of our cohort. Regarding race and ethnicity, our cohort 982

demonstrates considerably more diversity compared with prior series, with Hispanics, Asians, and African-Americans constituting 19%, 10%, and 5% of the cohort, respectively. One of the major findings in our analysis is that C3GN was not associated with better rates of remission or lower rates of progression than DDD, as has been seen in prior cohorts.2,14 Indeed, these similar outcomes in the DDD and C3GN subgroups prevailed despite the earlier age of diagnosis, better baseline renal function, and fewer markers of chronicity on biopsy found in our C3GN patients. Hispanic and Asian patients were more likely to have C3GN than DDD in our cohort. While ethnicity overall did not significantly predict risk of progression in multivariable analysis, confidence intervals were wide, and the point estimates of the hazard ratios for both these subgroups suggested the possibility of a substantial effect on risk progression. Conceivably, the C3GN variant in nonwhite patients is a more aggressive lesion than what has been previously reported, but whether this is related to genetic or socioeconomic factors is unclear. These differences should ideally be explored in larger, multicenter cohorts. Additionally, in contrast with prior C3G series in which patients with DDD had lower C3 levels and higher prevalence of C3nef antibodies compared with C3GN,2,7,11,14 this distinction was not apparent in our cohort. Presumably, lower circulating C3 and the presence of C3nef imply greater activity at the proximal (vs. terminal) components of the alternative pathway, imparting a more aggressive and potentially treatment-resistant glomerular lesion. Therefore, in our patients, the similar rates of low C3 and C3nef in both C3G subclasses suggest, at least in a large American cohort, that proximal complement defects are equally prevalent. Another major distinction for our cohort, as compared with prior published series, is our use of a C3G histologic index expressly created for this disease entity. This index, if validated by independent cohorts, could emerge as a useful tool for both clinical and research purposes, akin to the activity and chronicity scores used for prognosis and longitudinal management of patients with lupus nephritis. Interestingly, while the total activity score emerged in multivariate models as a predictor of poor outcome, only 1 of its 7 components (crescent formation) showed clear correlation on univariate analysis. This discrepancy suggests that the cumulative effect of disease activity, measured over a variety of indicators, may be the best way for clinicians and pathologists to assess the aggressiveness of a C3G lesion. Not surprisingly, given that traditional clinical and histopathologic parameters of chronicity were the strongest predictors of progression in this cohort, the total chronicity score appears to be a stronger predictor of outcome than the total activity score in our patients. This aligns with the experience of chronicity scores in other form of glomerulonephritis, such as IgA nephropathy, in which a T2 score (denoting >50% tubular atrophy and interstitial fibrosis) has emerged as the most reliable indicator of the Oxford Classification scheme.15 It is important to note that the C3GN patients in our cohort had higher activity but lower chronicity scores than the DDD Kidney International (2018) 93, 977–985

AS Bomback et al.: C3 glomerulopathy in the United States

patients, yet both groups had similar rates of progression to advanced CKD and ESRD, highlighting the importance of assessing, quantitatively, both activity and chronicity using our C3G histologic index. Further refinement of this scoring system ideally can be accomplished via collaborative, prospective study groups using histopathology and clinical data across a wide geographic spectrum of C3G. The strengths of this retrospective study include the number of subjects who constitute the cohort, representing not only the largest series of North American C3G patients but also the largest total C3G patient series to appear in the literature, and the long duration of follow-up for these individuals. With a rare disease such as C3G, it is difficult to detect statistically significant differences in subpopulations (e.g., C3GN vs. DDD) without the creation of a large patient sample, and it may be difficult to amass hard outcomes (e.g., ESRD) for analysis without an extended period of follow-up time. This cohort of >100 patients with mean follow-up time of 72 months minimizes these potential shortcomings. While we did examine remission as an outcome, using the change in proteinuria as a biomarker of disease, we were also powered to assess the effect of a number of predictor variables—including demographic, laboratory, histopathology, and treatment—on outcomes of ESRD and progression to advanced CKD. Nonetheless, our series has unavoidable limitations inherent to retrospective studies. Some of these data are reliant upon patient recall and thoroughness of documentation in the medical record, including antecedent infections, reported in approximately 18% of our cases, and immunosuppressive treatment courses. Importantly, the lack of randomized treatment may have confounded the outcome analysis, and our finding that immunosuppressive therapies were not associated with improved outcomes may have been influenced by the tendency to offer such therapies to patients considered to be at highest risk for progression. Testing for complement-associated gene variants and autoantibodies was performed only in a subgroup of patients (n ¼ 51), generally those with more recent diagnoses, as this testing was not routinely available in the past. The results from this limited genetic testing, furthermore, do not comply with recent international guidelines suggesting that, when identified, variants should be classified as benign, likely benign, variant of uncertain significance, likely pathogenic, or pathogenic.16,17 Our finding that such abnormalities, when detected, were not associated with renal outcomes may not be applicable to all patients in this cohort nor to future C3G patients with access to complement-targeting drugs. Additionally, important granular differences in outcomes associated with specific complement pathway defects may have been missed by analyzing them collectively rather than according to their known pathogenicity. Finally, testing for paraproteins was limited to only 38 patients, whose average age was 45 years. Our finding of dysproteinemias in DDD patients at more than twice the prevalence than in C3G patients may only be valid in the middle-aged and older patient population. Kidney International (2018) 93, 977–985

clinical investigation

We advocate that dysproteinemias be thoroughly investigated in all patients with C3G, regardless of subclassification (C3GN vs. DDD) or patient age, given that dysproteinemiaassociated C3G appears to be a more treatment-responsive version of the disease.13 In conclusion, in this large and diverse cohort of United States patients with C3G, we report a high rate of progression to advanced and end-stage kidney disease with no significant differences in outcomes between C3GN and DDD subgroups. Markers of chronicity, both clinical and histologic, were the strongest predictors of outcomes among these patients, while markers of disease activity and complement dysfunction were not as useful for prognostication. As we enter an era of novel, complement-targeting drugs, our data suggest that early detection and implementation of disease-specific therapy may be the best strategy to mitigate the apparently severe natural history of this rare disease. METHODS Subjects In this retrospective cohort study, we reviewed the charts of all patients with biopsy-diagnosed C3G seen for routine care at the Center for Glomerular Diseases at Columbia University Medical Center from January 1, 1997, to December 31, 2016. Because C3G was formally recognized as a distinct clinical-pathologic entity in 2010,1 those patients with a kidney biopsy performed prior to 2010 underwent an independent re-review of their biopsies by 2 pathologists (DS and LCH) to confirm the diagnosis according to 2013 consensus guidelines, which require C3 staining on immunofluorescence at least 2 orders of magnitude greater than any Ig.18 Data extraction from patient charts was performed primarily by 2 study investigators (RSA and RR-S). Discrepancies or missing data were adjudicated through a second-level review by the patient’s primary nephrologist at the Center for Glomerular Disease and/or the patient’s referring nephrologist, if applicable. The study was approved by the institutional review board of Columbia University Medical Center. Outcomes The outcomes of interest were doubling of serum creatinine from baseline, progression to CKD stage 5 (estimated GFR < 15 ml/min per 1.73 m2 by the CKD-EPI equation19), ESRD requiring dialysis or transplantation, and death. As the study’s primary outcome, we utilized a combined outcome of the above endpoints. Doubling of serum creatinine included only those patients who did not reach CKD stage 5 or ESRD. ESRD was defined as initiation of renal replacement therapy via either dialysis (often followed by transplantation) or preemptive transplantation. We also examined disease remission, defined as (i) a reduction to <300 mg/d proteinuria with stable or improved serum creatinine (complete remission) or (ii) a $50% reduction in proteinuria from baseline to 300 to 3500 mg/d with stable or improved serum creatinine (partial remission). Patients were followed up until their last clinical encounter or death. Predictors We obtained data regarding patients’ clinical courses in order to examine potential predictors of outcomes. We considered the time of first diagnostic biopsy to be baseline for all patients. Baseline clinical 983

clinical investigation

data collected included age, gender, race and/or ethnicity, serum creatinine, degree of proteinuria (quantified by either 24-hour collection or random urine protein-to-creatinine ratio), and level of serum complement components. A subgroup of patients underwent testing for genetic variants and acquired autoantibodies associated with dysregulation of the alternative complement pathway at the University of Iowa’s Molecular Otolaryngology and Renal Research Laboratories, as described elsewhere,7,14 and these findings were included in relevant analyses as predictor variables. In addition, a subgroup of patients was tested for presence of monoclonal gammopathies via serum and urine protein electrophoresis with immunofixation electrophoresis. Treatment data were collected throughout the duration of each patient’s clinical course. We also evaluated histopathologic predictors of outcomes via central re-review of biopsy materials by 2 pathologists (DS and LCH) in the Renal Pathology Laboratory at Columbia University Medical Center. When original case material was not available, extraction from the biopsy report was performed to collect relevant predictor variables, including 4 patterns on light microscopy (MPGN, characterized by diffuse and global double-contoured glomerular capillary walls with mesangial cell interposition and mesangial expansion by increased mesangial cell number and matrix; mesangial proliferative GN, characterized by diffuse mesangial proliferation in the absence of endocapillary proliferative or membranoproliferative features; diffuse endocapillary proliferative GN, characterized by endocapillary hypercellularity and leukocyte infiltration causing luminal occlusion; and diffuse sclerosing GN, characterized by segmental and/or global glomerulosclerosis involving >90% of glomeruli), degree of global and segmental glomerulosclerosis, presence of exudative features (defined as the presence of endocapillary proliferation by infiltrating neutrophils), presence of any cellular or fibrocellular crescents, degree of tubular atrophy and interstitial fibrosis, presence or absence of any Ig staining on immunofluorescence microscopy, and location and density of electron-dense deposits on electron microscopy for identification as C3GN versus DDD following the 2013 consensus guidelines.18 C3G histologic index In addition to the above extraction of conventional histopathologic predictors, we used original biopsy material to develop a novel C3G histologic index for the degree of disease activity and chronicity. For disease activity, we used 7 parameters based on previously published scoring systems for glomerulonephritides,20–22 each graded 0 to 3 for a total possible score of 21: (i) mesangial hypercellularity, defined as >3 mesangial cells per mesangial area; (ii) endocapillary proliferation, defined as an increased number of cells within glomerular capillary lumina, causing luminal narrowing or obliteration; (iii) membranoproliferative morphology, defined as glomerular basement membrane (GBM) duplication with or without endocapillary proliferation; (iv) leukocyte infiltration, defined as glomerular capillary infiltration by $3 neutrophils and/or macrophages (identified by their relatively abundant cytoplasm and characteristic “kidney bean-shaped” nuclei vs. lymphocytes, which have scant cytoplasm and round nuclei); (v) cellular and/or fibrocellular crescent formation, defined as extracapillary cell proliferation of more than 2 cell layers with >50% of the lesion occupied by cells; (vi) fibrinoid necrosis, defined as the presence of 2 or more features including fibrin, karyorrhexis, and GBM rupture; and (vii) interstitial inflammation, defined as inflammation in nonfibrotic cortex. For mesangial hypercellularity, endocapillary proliferation, membranoproliferative morphology, 984

AS Bomback et al.: C3 glomerulopathy in the United States

and leukocyte infiltration, we used a scale of 0 ¼ none, 1 ¼ 1% to 25%, 2 ¼ 26% to 50%, and 3 ¼ >50% involvement of sampled glomeruli. For crescent formation and fibrinoid necrosis, we used a scale of 0 ¼ none, 1 ¼ 1% to 10%, 2 ¼ 11% to 25%, and 3 ¼ >25% involvement of sampled glomeruli. For interstitial inflammation, we scored according to percentage of cortical tubulointerstitial area involved (0 ¼ <10%, 1 ¼ 10% to 25%, 2 ¼ 26% to 50%, 3 ¼ >50%). For disease chronicity, based on the recently proposed scale for standardized grading of chronic changes,23 we considered 4 features weighted towards a total possible score of 10: (i) global and segmental glomerulosclerosis; (ii) tubular atrophy; (iii) interstitial fibrosis; and (iv) arterio- and arteriolosclerosis. Glomerulosclerosis (global plus segmental), tubular atrophy, and interstitial fibrosis were assigned a score of 0 to 3 based on the percentage of glomeruli or cortical tubulointerstitial area involved (0 ¼ <10%, 1 ¼ 10% to 25%, 2 ¼ 26% to 50%, 3 ¼ >50%). For vascular disease, we assigned a score of 0 if intimal thickening was < thickness of media and 1 if intimal thickening was $ thickness of media. Statistical analyses Baseline clinical and histopathology characteristics for all subjects were assessed as mean (SD) and median (interquartile range) for continuous variables and percentage for categorical variables. These parameters were compared between patients with C3GN versus patients with DDD using a 2-sample Wilcoxon rank-sum (MannWhitney) test for continuous variables and Fisher exact test for categorical variables. Differences in utilization of treatment were compared via Fisher exact test. For the outcomes of interest— remission (complete þ partial), doubling of creatinine, progression to CKD stage 5, ESRD, death, and a combined outcome of these endpoints—we used Cox proportional hazards models over the entire duration of follow-up. The time of first diagnostic biopsy was considered as the start of follow-up. We utilized a univariate model of C3GN versus DDD for initial analyses, and then we created 3 multivariate models to further evaluate all predictor variables. Our first multivariate model (clinical variables–only model) included the following predictor variables: age, gender, race and/or ethnicity, creatinine and proteinuria at time of biopsy, C3 and C4 levels, presence of a complement-associated gene variant or autoantibody, and use of immunosuppressive therapy. Our second model (histopathology variables–only model) included the following predictor variables found on biopsy: pattern on light microscopy, percentage of globally and segmentally sclerotic glomeruli, presence of exudative features, presence of crescents (cellular or fibrocellular), degree of tubular atrophy and/or interstitial fibrosis, C3-only staining on immunofluorescence (vs. presence of trace to 1þ Ig), and location and density of deposits on electron microscopy (i.e., C3GN vs. DDD pattern). Our third model used activity and chronicity scores from the C3G histologic index. Analyses and plots were performed using STATA version 12.1 (StataCorp, College Station, TX). DISCLOSURE

ASB has received consulting honoraria from and/or served on advisory boards for Chemocentryx, Achillion, and Omeros. GBA has received consulting honoraria from and/or served on advisory boards for Alexion, Chemocentryx, Achillion, and Omeros, and is involved in research studies for Achillion and Omeros. LCH has received consulting honoraria from Chemocentryx. All the other authors declared no competing interests. Kidney International (2018) 93, 977–985

AS Bomback et al.: C3 glomerulopathy in the United States

ACKNOWLEDGMENTS

We are grateful to the Molecular Otolaryngology and Renal Research Laboratories at the University of Iowa, led by Dr. Richard Smith, for their assistance with complement-associated genetic and antibody testing for a subgroup of the patients in this cohort. We also thank our team of research coordinators at the Center for Glomerular Diseases—Michael DiVecchia, Byum Hee Kil, and Stacy Piva—for assistance with biosample collection in these patients. SUPPLEMENTARY MATERIAL Table S1. Prevalence of genetic variants and autoantibodies associated with dysregulation of the alternative complement pathway in the subgroup of patients (N ¼ 51) who underwent screening. Table S2. Progression to primary outcome as predicted by individual parameters of the C3G histologic index. Supplementary material is linked to the online version of the paper at www.kidney-international.org. REFERENCES 1. Fakhouri F, Fremeaux-Bacchi V, Noel LH, et al. C3 glomerulopathy: a new classification. Nat Rev Nephrol. 2010;6:494–499. 2. Servais A, Noël LH, Roumenina LT, et al. Acquired and genetic complement abnormalities play a critical role in dense deposit disease and other C3 glomerulopathies. Kidney Int. 2012;82:454–464. 3. Medjeral-Thomas NR, O’Shaughnessy MM, O’Regan JA, et al. C3 glomerulopathy: clinicopathologic features and predictors of outcome. Clin J Am Soc Nephrol. 2014;9:46–53. 4. Bomback AS, Appel GB. Pathogenesis of the C3 glomerulopathies and reclassification of MPGN. Nat Rev Nephrol. 2012;8:634–642. 5. D’Agati VD, Bomback AS. C3 glomerulopathy: what’s in a name? Kidney Int. 2012;82:379–381. 6. Hou J, Markowitz GS, Bomback AS, et al. Toward a working definition of C3 glomerulopathy by immunofluorescence. Kidney Int. 2014;85: 450–456. 7. Zhang Y, Meyer NC, Wang K, et al. Causes of alternative pathway dysregulation in dense deposit disease. Clin J Am Soc Nephrol. 2012;7: 265–274. 8. Okuda Y, Ishikura K, Hamada R, et al. Membranoproliferative glomerulonephritis and C3 glomerulonephritis: frequency, clinical features, and outcome in children. Nephrology (Carlton). 2015;20: 286–292.

Kidney International (2018) 93, 977–985

clinical investigation

9. Rabasco C, Cavero T, Román E, et al. Effectiveness of mycophenolate mofetil in C3 glomerulonephritis. Kidney Int. 2015;88:1153–1160. 10. Caliskan Y, Torun ES, Tiryaki TO, et al. Immunosuppressive treatment in C3 glomerulopathy: is it really effective? Am J Nephrol. 2017;46:96–107. 11. Iatropoulos P, Noris M, Mele C, et al. Complement gene variants determine the risk of immunoglobulin-associated MPGN and C3 glomerulopathy and predict long-term renal outcome. Mol Immunol. 2016;71:131–142. 12. Zand L, Kattah A, Fervenza FC, et al. C3 glomerulonephritis associated with monoclonal gammopathy: a case series. Am J Kidney Dis. 2013;62: 506–514. 13. Chauvet S, Frémeaux-Bacchi V, Petitprez F, et al. Treatment of B-cell disorder improves renal outcome of patients with monoclonal gammopathy-associated C3 glomerulopathy. Blood. 2017;129: 1437–1447. 14. Sethi S, Fervenza FC, Zhang Y, et al. C3 glomerulonephritis: clinicopathological findings, complement abnormalities, glomerular proteomic profile, treatment, and follow-up. Kidney Int. 2012;82:465–473. 15. Coppo R, Troyanov S, Bellur S, et al. Validation of the Oxford classification of IgA nephropathy in cohorts with different presentations and treatments. Kidney Int. 2014;86:828–836. 16. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17:405–424. 17. Goodship TH, Cook HT, Fakhouri F, et al. Atypical hemolytic uremic syndrome and C3 glomerulopathy: conclusions from a “Kidney Disease: Improving Global Outcomes” (KDIGO) Controversies Conference. Kidney Int. 2017;91:539–551. 18. Pickering MC, D’Agati VD, Nester CM, et al. C3 glomerulopathy: consensus report. Kidney Int. 2013;84:1079–1089. 19. Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. 20. Weening JJ, D’Agati VD, Schwartz MM, et al. The classification of glomerulonephritis in systemic lupus erythematosus revisited. J Am Soc Nephrol. 2014;15:241–250. 21. Working Group of the International IgA Nephropathy Network and the Renal Pathology Society1, Roberts IS, Cook HT, et al. The Oxford classification of IgA nephropathy: pathology definitions, correlations, and reproducibility. Kidney Int. 2009;76:546–556. 22. Sethi S, Haas M, Markowitz GS, et al. Mayo Clinic/Renal Pathology Society Consensus Report on Pathologic Classification, Diagnosis, and Reporting of GN. J Am Soc Nephrol. 2016;27:1278–1287. 23. Sethi S, D’Agati VD, Nast CC, et al. A proposal for standardized grading of chronic changes in native kidney biopsy specimens. Kidney Int. 2017;91: 787–789.

985