A Prospective Study on Complement Activation Distinguishes Focal Segmental Glomerulosclerosis from Minimal Change Disease

Abstract

Introduction

Minimal change disease (MCD) and focal segmental glomerulosclerosis (FSGS) are related podocytopathies with distinct kidney outcomes. Surprisingly, elevated urinary activation fragments have been found in FSGS despite little complement deposition on immunofluorescence (IF) staining. Whether complement activation distinguishes FSGS from MCD, participating in the development of segmental lesions, remains unknown.

Methods

We performed an observational study in patients with MCD and FSGS, and proteinuria ≥1 g/g of creatinine. We included both primary and secondary or unknown causes. We compared urinary fragments of terminal pathway activation, sC5b9, and C5a expressed as creatinine ratios, between MCD and FSGS.

Results

Patients with FSGS (n = 41) had a serum albumin of 31±10 g/l and proteinuria of 5.1 (2.6–9.1) g/g at sampling, whereas those with MCD (n = 15) had a lower serum albumin (22 ± 9 g/l; P = 0.002), and a proteinuria of 3.8 (1.9–7.7) g/g (P = 0.40). Urinary sC5b9 and C5a were 8.7 (1.7–52.3) and 1.26 (0.45–1.84) μg/mmol of creatinine, respectively in patients with FSGS; compared to 0.8 (0.0–1.5) and 0.06 (0.01–0.15) μg/mmol of creatinine in MCD (P < 0.001), respectively. We found no association between urinary complement fragments and age, estimated glomerular filtration rate (eGFR), or chronic kidney lesions. When analyzing samples with proteinuria ≥ 3 g/g, the c-statistics for urinary sC5b9 and C5a were 0.96 and 1.00, respectively, in differentiating FSGS from MCD.

Conclusion

We found no urinary complement activation fragments in MCD, in comparison to FSGS, despite similar levels of proteinuria. This suggests a role for complement activation in the pathogenesis of FSGS and provides an additional tool for distinguishing these 2 entities.

Graphical abstract

See Commentary on Page 520

Primary MCD and FSGS account for most nephrotic syndromes in young adults.¹ Pathology findings distinguish these 2 diseases, but unsampled FSGS can initially be misdiagnosed as MCD, especially when few glomeruli are sampled. Furthermore, children with idiopathic nephrotic syndrome are infrequently biopsied as they are presumed to be MCD. However, the distinction between these 2 entities is essential: FSGS is treated using diverse strategies and is a leading glomerular cause of kidney failure, as opposed to MCD.²

Multiple biomarkers have been studied to differentiate these 2 entities and unravel their respective pathogenesis. These include the soluble form of the urokinase plasminogen activator receptor, urinary CD80, and antinephrin antibodies, to name a few.^1,3 Although they provide important mechanistic insights, these biomarkers are insufficiently sensitive or specific for general clinical use.

Complement activation is central in the pathogenesis of many glomerular diseases. We recently studied the urinary soluble membrane attack complex (MAC), also known as sC5b9, in common autoimmune glomerulonephritis and found surprisingly elevated levels in FSGS, second only to membranous nephropathy.⁴ Thurman et al.⁵ also observed very elevated urinary sC5b9 in FSGS compared to other glomerulonephritis. In contrast, C3 staining by IF is usually faint in FSGS and limited to sclerotic areas, with no deposits found on electron microscopy (EM). To further elucidate the role of complement activation, we measured urinary sC5b9 in a cohort of patients with MCD and FSGS and studied clinicopathological associations. We hypothesized that urinary sC5b9 is higher in FSGS than MCD and that urinary sC5b9 levels are higher in primary FSGS than in secondary forms or FSGS of undetermined cause.

Methods

Study Design and Population

We performed a prospective observational study in patients with glomerular diseases recruited between 2006 and 2023 in 4 hospitals affiliated with the University of Montreal, Canada. Each center’s ethics committee approved this study, and all participants gave informed consent. We included all available individuals with a pathological diagnosis of MCD or FSGS and an available urinary sample with a proteinuria ≥1 g/g of creatinine. We did not exclude secondary causes, except for HIV, which can be associated with other conditions with complement activation. This work was carried out following the declaration of Helsinki.

Variables

In addition to demographics, we collected data on blood pressure, serum albumin and creatinine, proteinuria, and body mass index at the time of urinary sampling. We recorded the exposure to immunosuppressive agents at the time of sampling and during follow-up. Because samples were taken at times during therapy, we also recorded the highest proteinuria and lowest serum albumin. We reviewed all recorded medical history and blood tests, including genetic (available in only 5 patients) and radiologic evaluations for evidence of secondary causes.

Pathology Review

All cases with available glass slides were reviewed by an experienced nephropathologist, different from and blinded to the original assessment. MCD required ≥10 nonsclerotic glomeruli available for examination in the light microscopy, IF, immunohistochemistry, or EM sections combined, where glomeruli are histologically unremarkable except for mesangial hypercellularity or globally sclerotic obsolescence. FSGS required ≥5 glomeruli in the light microscopy, IF, immunohistochemistry, or EM sections combined, with at least 1 glomerulus with segmental sclerosis or consolidation by increased extracellular matrix obliteration of the capillary lumen.^6,7 We identified “probable FSGS” in those with glomeruli showing only adhesion(s) without evident segmental sclerosis or consolidation. We also assessed interstitial fibrosis and tubular atrophy, arteriosclerosis, and arteriolar hyalinosis using a semiquantitative scale from 0 to 3+, corresponding for interstitial fibrosis and tubular atrophy affecting <5%, 5% to 25%, 26% to 50%, and >51% of the parenchyma. We also reviewed EM findings and determined if diffuse (≥75%) or segmental (<75%) foot process effacement existed.

Definitions

Definite primary MCD and FSGS required a nephrotic syndrome without evident secondary cause that remitted following immunosuppression. Primary disease also had to show diffuse foot process effacement when the biopsy was performed during a nephrotic period and before immunosuppression. However, this criterion was not a requirement when the biopsy was done outside of a nephrotic period or under immunosuppressive therapy, because this occurred in some children who were biopsied distantly from their initial presentation. Cases that did not satisfy these conditions were labeled of undetermined cause or secondary when a plausible etiology existed.

We assessed remissions, either partial or complete, and relapses, as previously defined.² The eGFR was derived using the 2021 Chronic Kidney Disease-Epidemiology Collaboration formula in adults and the modified Schwartz formula in children. Kidney failure was defined by an irreversible eGFR <15 ml/min per 1.73 m².

Urinary Measurements

A urinary void was taken, and one part was sent to measure proteinuria with urinary creatinine, expressed in g/g of creatinine. The other part was aliquoted and stored at −80 °C until further processing. No protease inhibitor was used for storage. We used human EIA Kits (MicroVue, Quidel Corp., San Diego, CA) to study the urinary complement fragments C5a and sC5b-9 as evidence of terminal pathway activation. Given that the selective albuminuria unique to MCD could restrict circulating sC5b9 (1030 kilodaltons) from appearing in the urine,⁸ we also measured the much smaller and freely filtered C5a (10 kilodaltons) to confirm our findings. Urine samples were initially diluted 1:5. Results are expressed as creatinine ratios. When patients had provided repeated urinary measurements, we chose the one with the highest proteinuria. Finally, given that sC5b9 and C5a are both peptides, we also compared the sC5b9-to-protein ratios between FSGS and MCDS.

Statistical Analyses

Normally distributed variables are presented as means ± SDs and were compared using the Student t-test. Nonparametric variables are expressed as median with interquartile range, and differences were tested using Mann-Whitney U and Spearman’s rho (ρ). Urinary sC5b9 and C5a performances to differentiate MCD from FSGS were examined using receiver operating characteristic curves and the c-statistic. The optimal cutoff for each urinary biomarker was obtained by identifying the point maximizing the sensitivity and specificity product. To examine whether the associations between urinary sC5b9 and C5a with the diagnosis were independent of age and eGFR, we performed a logistic regression where we used urinary biomarkers dichotomized according to these cut-offs to respect the assumption of normality. Two-tailed P-values less than 0.05 were considered statistically significant. Analyses were performed using SPSS software (version 26, IBM, NY).

Results

Study Population

There were 64 patients with a proteinuria >1 g/g of creatinine. After 8 exclusions, 15 patients with MCD and 41 with FSGS remained (Figure 1), including 9 children. In 5 cases, there was a single adhesion without segmental sclerosis; these patients were categorized into the FSGS group. We included 8 patients despite unavailable biopsies to review, all with an initial pathology assessment describing FSGS unambiguously (Table 1). EM was either not performed or without glomeruli in 11 instances. All other data were present except for 2 without clinical follow-up.

Figure 1.

Patient selection. Bx, biopsy; cr, creatinine; FSGS, focal segmental glomerulosclerosis; MCD, minimal change disease; NOS, not otherwise specified.

Table 1.

Cohort demographics and clinicopathological findings

a	Sex, age, race, BMI	eGFR, albumin and proteinuria at sample	Worst albumin and proteinuria	Immunosuppression: at sample, at FU. Outcomes	Presumed Etiology	aGlom: total, sclerosed, normal, NOS±Adh, type	Chronicity 0-3+ IFTA, AS, AH	EM podocyte effacement
FSGS
1	F, 5, Cauc, 24	123, 27, 55.0	8, 55.0	yes, yes. PR	primary	31, 1, 29, 1, NOS	0, 0, 0	diffuse
2	F, 51, Cauc, <30	31, 17, 28.6	16, 28.6	yes, yes. CR, relapse	primary	35, 5, 29, 1, NOSc	1, 2, 2	diffuse
3	F, 51, A, 38	16, 15, 17.8	10, 29.0	no, no. NR, KF	pamidronate	21, 1, 8, 12, Collapsing	2, 1, 0	N/A
4	M, 22, A, <30	34, 5, 15.7	5, 18.4	no, yes. PR	primary	18, 0, 16, 2, NOSc	0, 0, 0	diffuse
5	M, 18, Cauc, <30	133, 21, 15.5	19, 15.5	no, yes, CR	primary	8, 0, 5, 3, NOS	1, N/A, 0	N/A
6	M, 71, Cauc, <30	31, 22, 14.5	12, 14.5	yes, yes. PR, relapse	primary	17, 6, 9, 2, NOS	2, 2, 1	diffuse
7	F, 25, Cauc, <30	126, 20, 13.6	15, 13.3	yes, yes. PR, relapse, KF	primary	18, 4, 13, 1, tip	1, 3, 3	N/A
8	F, 63, Cauc, <30	53, 35, 10.2	29, 10.2	no, no. NR, KF	uncertain	18, 2, 9, 7, NOS	1, 1, 1	no glom
9	M, 5, Cauc, 19	65, 15, 10.1	9, 10.1	yes, yes. PR	primary	18, 0, 17, 1b, NOS	0, 0, 0	segmental
10	M, 51, Cauc, 38	45, 15, 9.6	15, 20.0	no, N/A. N/A	primary	22, 5, 8, 9, NOSc	2, 1, 2	diffuse
11	F, 19, Cauc, <30	129, 26, 8.7	15, 8.7	no, yes. CR, RL	primary	16, 1, 14, 1b, NOS	1, 1, 0	diffuse
12	M, 32, A, <30	36, 28, 8.5	28, 17.8	yes, yes. PR, KF	primary	10, 7, 3, 0, NOS	2, 2, 2	segmental
13	F, 54, Cauc, <30	44, 29, 8.3	17, 21.4	yes, yes. CR	primary	10, 0, 8, 2, Collapsing	0, N/A, 1	diffuse
14	M, 61, M-east, <30	39, 37, 7.6	33, 10.6	no, no. PR, KF	uncertain	10, 5, 2, 3, NOS	3, 3, 2	segmental
15	F, 26, Cauc, 39	123, 23, 7.2	15, 9.0	no, yes. PR	primary	29, 0, 27, 2, Tip	0, 0, 0	diffuse
16	F, 26, M-east, <30	34, 38, 7.1	28, 9.3	no, yes. PR	primary	15, 10, 2, 3, NOS	3, 1, 1	no glom
17	M, 73, Cauc, 29	44, 34, 6.5	32, 6.5	no, no. NR	uncertain	38, 19, 13, 6, NOS	2, 3, 3	segmental
18	M, 72, Cauc, <30	35, 25, 6.5	22, 13.9	no, yes. PR, KF	primary	29, 2, 21, 6, NOS	1, 1, 1	diffuse
19	M, 57, Cauc, <30	36, 19, 6.5	16, 11.7	yes, yes. CR, RL	primary	25, 0, 22, 3, Tip	1, 0, 0	diffuse
20	M, 49, Cauc, <30	45, 40, 5.5	38, 5.5	no, no. NR	uncertain	9, 1, 7, 1, NOS	0, 1, 0	diffuse
21	M, 46, M-east, 25	42, 36, 5.1	34, 7.0	no, yes. PR, RL, KF	primary	13, 9, 1, 3, NOS	3, 1, 1	N/A
22	M, 67,S-Asian,<30	13, 38, 4.8	13, 23.2	yes, yes. PR, RL, pCKD	primary	9, 6, 2, 1, NOS	3, 3, 1	diffuse
23	M, 42, Cauc, 31	17, 34, 4.4	34, 12.0	no, no. KF	anabolic drug	26, 17, 4, 5, Collapsing	3, 3, 1	segmental
24	M, 64, Cauc, 31	29, 25, 3.8	25, 3.8	no, no. pCKD	uncertain	24, 14, 5, 5, NOSc	3, 3, 3	no glom
25	F, 63, Asian, <30	51, 37, 3.7	30, 7.9	yes, yes. PR, RL, KF	primary	6, 2, 3, 1, NOS	1, 2, 1	no glom
26	M, 40, Cauc, <30	43, 38, 3.7	37, 9.2	no, yes. PR, pCKD	COL4A4	8, 5, 2, 1, NOS	2, 1, 1	diffuse
27	M, 34, Cauc, 25	93, 34, 3.6	34, 7.0	no, yes. CR	primary	34, 1, 29, 4, Perihilar	0, 0, 0	segmental
28	M, 53, Cauc, <30	27, 39, 3.1	39, 3.1	no, no. pCKD	uncertain	39, 26, 9, 4, NOS	3, 2, 1	segmental
29	M, 63, Cauc, <30	11, 41, 2.8	40, 3.0	no, no. KF	uncertain	6, 4, 0, 2, NOS	3, N/A, 1	no glom
30	M, 61, Cauc, <30	91, 35, 2.6	20, 9.6	yes, yes. PR	primary	26, 1, 21, 1b, NOS	0, 1, 1	diffuse
31	F, 72, M-east, <30	13, 44, 2.6	36, 2.6	no, no. KF	uncertain	16, 10, 4, 2, NOSc	2, 2, 2	diffuse
32	M, 62, Cauc, <30	44, 40, 2.5	39, 6.6	no, no. KF	uncertain	11, 6, 2, 3, NOS	3, 1, 1	segmental
33	M, 72, Cauc, <30	58, 35, 2.3	16, 4.9	yes, yes. CR, RL	primary	5, 0, 4, 1b, NOS	0, N/A, 0	diffuse
34	M, 64, Cauc, 24	52, 33, 2.1	31, 2.4	no, N/A. N/A	childhood GN	25, 8, 15, 2, NOSc	1, 3, 2	no glom
35	M, 56, Cauc, <30	25, 35, 2.1	26, 7.8	yes, yes. PR, RL, KF	primary	21, 5, 11, 5, NOS	1, 1, 0	segmental
36	M, 47, Cauc, 25	40, 33, 2.0	29, 4.5	no, yes. PR	primary	29, 21, 5, 3, NOS	3, 0, 0	segmental
37	M, 43, Cauc, 27	34, 37, 1.9	37, 7.0	no, no. KF	uncertain	14, 8, 3, 3, tip	2, 2, 2	segmental
38	M, 70, Cauc, 31	35, 36, 1.4	28, 1.4	no, no. KF	uncertain	11, 4, 5, 2, NOSc	2, 3, 2	segmental
39	M, 17, Cauc, 20	155, 41, 1.4	10, 9.7	yes, yes. CR	primary	22, 0, 21, 1b, NOS	0, 0, 0	segmental
40	F, 53, M-east, <30	24, 37, 1.2	37, 1.6	yes, yes. pCKD	uncertain	26, 10, 8, 8, NOSc	2, 1, 2	segmental
41	M, 36, Cauc, <30	66, 49, 1.0	49, 2.3	no, no. NN	uncertain	7, 2, 4, 1, NOS	0, 0, 0	diffuse
Minimal change disease
1	F, 14, A, 22	124, 13, 22.6	13, 22.6	no, yes. CR	primary	19,0,19,0	0, 0, 0	diffuse
2	M, 11, A, <30	41, 9, 19.1	9, 19.1	yes, yes. CR, RL	primary	13, 1, 12, 0	0, 0, 0	segmental
3	M, 24, Cauc, <30	135, 22, 10.1	22, 13.0	yes, yes. CR, RL	primary	14, 0, 14, 0	0, 0, 0	no glom
4	F, 55, Asian, 18	88, 23, 7.7	21, 21.7	yes, yes. CR	primary	29, 1, 28, 0	0, 1, 1	diffuse
5	M, 44, A, 25	107, 16, 6.1	10, 7.4	yes, yes. CR, RL	primary	16, 2, 14, 0	0, 0, 0	diffuse
6	M, 25, M-east, <30	106, 39, 4.5	19, 6.2	yes, yes. CR, RL	primary	38, 0, 38, 0	0, 0, 0	diffuse
7	M, 13, Cauc, 19	112, 19, 4.3	8, 4.3	yes, yes. CR, RL	primary	11, 1, 10, 0	0, N/A, 0	diffuse
8	F, 41, Cauc, 45	117, 20, 3.8	17, 7.8	no, yes. CR	primary	10, 1, 9, 0	0, 1, 0	segmental
9	F, 28, Cauc, <30	122, 17, 3.6	6, 5.6	no, yes. CR	primary	32, 0, 32, 0	0, 0, 0	segmental
10	M, 8, Cauc, 18	107, 21, 3.5	9, 5.7	yes, yes. CR, RL	primary	19, 0, 19, 0	0, 0, 0	segmental
11	M, 40, A, 32	112, 32, 2.4	29, 3.1	yes, yes. CR, RL	primary	11, 0, 11, 0	0, N/A, 0	diffuse
12	F, 78, Cauc, <30	26, 24, 1.9	15, 6.6	yes, yes. CR	primary	13, 2, 11, 0	0, 1, 0	diffuse
13	M, 28, A, <30	129, 13, 1.8	11, 20.9	yes, yes. CR	primary	38, 0, 38, 0	0, 0, 0	diffuse
14	M, 75, Cauc, 23	65, 37, 1.2	37, 2.3	no, no. CR	uncertain	12, 2, 10, 0	0, 3, 3	diffuse
15	M, 17, Cauc, 22	98, 25, 1.1	8, 12.8	no, yes. CR	primary	14, 0, 14, 0	0, 0, 0	segmental

A, African; Adh, adhesion; AH, arteriolar hyalinosis; AS, arteriosclerosis; BMI, body mass index; Cauc, Caucasian; CR, complete remission; eGFR, estimated glomerular filtration rate; EM, electron microscopy; F, female; FSGS, focal segmental glomerulosclerosis; FU, follow-up; IFTA, interstitial fibrosis and tubular atrophy; KF, kidney failure; M, male; M-east, Middle Eastern; NOS, not otherwise specified; NR, no remission; pCKD, progressive chronic kidney disease; PR, partial remission; RL, relapse.

^aRefer to Figure 1 identification.

^bPatients with probable FSGS defined by glomeruli with an adhesion but without NOS lesion.

^cPatients where we could not locate the pathology slides for review, all of which had an unambiguous description of FSGS at first assessment. eGFR, albumin and proteinuria are expressed in ml/min per 1.73 m², g/l, and g/g of creatinine, respectively.

Patients with FSGS were 48 ± 19 years old, with 29% female, an eGFR of 53 ± 37 ml/min per 1.73 m², serum albumin of 31 ± 10 g/l, and proteinuria of 5.1 (2.6–9.1) g/g at sampling. Their minimal albumin and maximal proteinuria during follow-up were 25 ± 11 g/l and 9.0 (4.7–13.6) g/g. Compared to FSGS, patients with MCD were younger at the time of urinary sampling (33 ± 22 years; P = 0.02), with 33% female (P = 0.77), a greater eGFR (99 ± 32 ml/min per 1.73 m²; P < 0.001), lower albuminemia (22 ± 9 g/l; P = 0.002), and proteinuria of 3.8 (1.9, 7.7) g/g (P = 0.40). Their minimum albuminemia and maximal proteinuria during follow-up were 16 ± 9 g/l and 7.2 (5.3–19.6) g/g.

All patients with MCD obtained a complete remission with preserved kidney function. All were considered primary except for 1 unknown cause, which remitted spontaneously without immunosuppressive therapy (Table 1). In contrast, 15 of 41 patients with FSGS experienced kidney failure, and only 25 experienced partial remission, often followed by a relapse. The causes of FSGS varied, with 24 of 41 (59%) satisfying our definition of primary disease; all received immunosuppression at one point during their follow-up. Two FSGS cases were drug-induced, 1 was genetic, 1 was maladaptive from suspected distant glomerulonephritis, and 13 were of uncertain cause, of which only 1 received immunosuppression. There were 6 of 41 patients with FSGS with a body mass index above 30; however, obesity was the remaining plausible etiology in only 2.

Urinary sC5b9 and C5a

Patients with FSGS presented urinary sC5b9 levels of 8.7 (1.7–52.3) μg/mmol of creatinine, which was much higher than 0.8 (0.0–1.5) in MCD (P < 0.001; Figure 2). Using a receiver operating characteristic curve, the sC5b9 threshold of 2.3 μg/mmol of creatinine maximized sensitivity (73%) and specificity (93%) to distinguish FSGS from MCD with a c-statistic of 0.82 (95% confidence interval, 0.70–0.93) (Figure 3). When limiting samples to a proteinuria ≥3 g/g, the >2.3 μg/mmol sC5b9 threshold sensitivity increased to 93%, with a c-statistic of 0.96 (95% confidence interval, 0.89–1.00). One individual with steroid-dependent MCD had an elevated urinary sC5b9. One child with MCD had a urinary sC5b9 level of 36 μg/mmol during acute tubular necrosis (typical sediment with doubling of serum creatinine), which rapidly decreased to 0.6 μg/mmol after recovery despite a proteinuria of 4.3 g/g.

Figure 2.

Urinary sC5b9 and C5a creatinine ratios compared to proteinuria in focal segmental glomerulosclerosis and minimal change disease. FSGS, focal segmental glomerulosclerosis; MCD, minimal change disease. Proteinuria was associated with urinary sC5b9 (ρ = 0.85; P < 0.001) and C5a (ρ = 0.71; P < 0.001) in FSGS, but not in MCD (P > 0.05).

Figure 3.

Receiver operating characteristic curves for urinary sC5b9 and C5a creatinine ratios to distinguish FSGS from MCD. FSGS, focal segmental glomerulosclerosis; MCD, minimal change disease. The c-statistics for urinary sC5b9 and C5a to differentiate FSGS from MCD were 0.82 and 0.96, respectively (blue curves). When limiting these analyses to those with a proteinuria ≥3 g/g of creatinine, the c-statistics were 0.96 and 1.0, respectively (green curves).

We repeated these analyses using C5a, a terminal pathway activation peptide unlikely to be affected by the selective proteinuria in MCD due to its small size. Patients with FSGS presented with a urinary C5a of 1.26 (0.45–1.84) μg/mmol of creatinine, which was markedly higher than the levels of 0.06 (0.01–0.15) μg/mmol of creatinine in MCD (P < 0.001; Figure 2). A urinary C5a threshold >0.33 μg/mmol of creatinine was 88% sensitive and 100% specific for FSGS, with a c-statistic of 0.94 (95% confidence interval, 0.88–1.00). When limiting samples to a proteinuria ≥3 g/g, the >0.33 μg/mmol C5a threshold sensitivity increased to 100%, with a c-statistic of 1.0 (Figure 3).

The relationships between proteinuria and urinary sC5b9 and C5a in each patient are detailed in Supplementary Figure S1 and S2, which can be linked to the patient numbering and clinical characteristics in Table 1. Primary FSGS had urinary sC5b9 and C5a of 12.5 (4.5–82.9) and 1.45 (0.41–1.87) μg/mmol of creatinine, respectively, compared to 4.8 (0.7–29.6) and 1.05 (0.46–1.73) μg/mmol of creatinine in secondary or undetermined causes of FSGS (P = 0.09 and P = 0.87), respectively.

A significant correlation between proteinuria with urinary sC5b9 and C5a (themselves proteins) existed, whether expressed as a creatinine ratio or not. We therefore compared the urinary sC5b9-to-protein ratio between FSGS and MCD to account for proteinuria and found similar results (Figure 4).

Figure 4.

Urinary sC5b9 and C5a to proteinuria rations in focal segmental glomerulosclerosis and minimal change disease. FSGS, focal segmental glomerulosclerosis; MCD, minimal change disease. The median urinary sC5b9-to-protein ratio was 16 (5–53) for FSGS and 1 (0–6) for MCD (P < 0.001), which a c-statistic of 0.82 (95% confidence interval, 0.71–0.93). When restricting this to proteinuria ≥3 g/g of creatinine, the c-statistic rose to 0.96 (95% confidence interval, 0.88–1.00).

Finally, we found no statistical association between urinary complement activation fragments and age or eGFR (Supplementary Figure S3). Nevertheless, we performed a logistic regression between the urinary biomarkers and the renal pathology, adjusting for age and eGFR. The adjusted odds ratios of having FSGS with a urinary sC5b9 >2.3 or a urinary C5a >0.33 μg/mmol of creatinine were greater than 40 in both instances (data not shown).

Pathology Findings

The time from biopsy to urinary sampling was 2 (0–18) months. Among patients with FSGS with a biopsy within 3 months of sampling (n = 29), we found no association between urinary sC5b9 or C5a and interstitial fibrosis and tubular atrophy, or the proportions of glomeruli with segmental or global glomerulosclerosis (Figure 5). We also found no association between the level of proteinuria and interstitial fibrosis and tubular atrophy or the proportions of glomeruli with segmental or global glomerulosclerosis. Interestingly, in the 5 individuals with a single adhesion lacking an observable segmental sclerosis, 2 had nephrotic range proteinuria and elevated urinary sC5b9 and C5a, suggesting they may indeed be unsampled FSGS (Supplementary Figures S1 and S2).

Figure 5.

Lack of association between urinary membrane attack complex and the proportions of glomeruli with segmental sclerosis (a), global glomerulosclerosis (b), and tubular atrophy and interstitial fibrosis (c). Associations were tested for statistical significance using Spearman’s rho.

Discussion

This observational study found no urinary complement activation fragments in MCD as opposed to FSGS, despite similar levels of proteinuria, suggesting a role of complement activation in the development of segmental sclerosis. This finding is remarkable, given the little or absent C3 staining by IF in MCD and FSGS.

Significantly elevated levels of urinary sC5b-9 were observed in another FSGS cohort and were accompanied by increased plasma sC5b9.⁵ Whether this reflects systemic MAC being filtered into the urine or kidney-derived sC5b9 backleaking into the plasma is uncertain. Regardless, this biomarker can differentiate both diseases when nephrotic range proteinuria exists. One patient with MCD with an elevated urinary sC5b9 may have been an unsampled FSGS. Using the small C5a fragment, we confidently excluded that the selective proteinuria found in MCD accounted for absent urinary MAC. Although urinary sC5b9 is found in many autoimmune glomerulonephritis and diabetic nephropathy, it is uncertain if its absence eliminates all causes of nephrotic syndrome other than MCD and whether a renal biopsy can be avoided. Currently, only anti-PLA2R positive nephrotic syndrome with preserved renal function, or steroid-responsive nephrotic syndrome in children 1 to 12 years of age, allow physicians to confidently establish a diagnosis confidently without a biopsy.^9,10 We did not find an association between urinary sC5b9 or C5a and the proportion of glomeruli with segmental or global glomerulosclerosis. Therefore, urinary sC5b9 and C5a appear associated with activity.

It is uncertain if plasma and urine complement activation fragments offer the same information. Such biomarkers are found in the plasma of healthy individuals with reported normal values of 127 to 400 μg/l for sC5b9 and 1.9–13.1 μg/l for C5a.¹¹ Plasma levels are likely influenced by eGFR and age (also linked with eGFR) in addition to the underlying renal disease.^5,12 In contrast, we found no sC5b9 or C5a in the urine of patients with nondiabetic and nonproteinuric chronic kidney disease⁴; we can infer that urinary complement fragments are absent in healthy individuals.

It is uncertain how complement activation occurs in FSGS, particularly when IF allows only ≤1+ C3, C1q, and IgM with small segmental mesangial immune-type deposits by EM.^6,7 An association between urinary Ba and urinary C5a supports activation of the alternative complement pathway among patients with FSGS.⁴ An important study by Trachtman et al. found elevated plasma C4a in FSGS with self-reactive IgM, colocalized with C4d in some glomeruli, favoring the classical or lectin pathway activation.¹² Interestingly, patients with FSGS and MCD had similar levels of plasma C4a as opposed to Ba, contrasting our “all or nothing” finding using urinary samples. However, plasma C4a and Ba in FSGS and MCD overlapped partly with healthy controls. The same group also found markedly higher levels of urinary sC5b9 in patients with FSGS compared to anti-neutrophil cytoplasmic autoantibody vasculitis and lupus nephritis, where the pathogenic importance of complement activation is firmly established.⁵ These studies strongly support a role for complement activation in FSGS.

The diagnosis of FSGS requires focal and segmental consolidation of the tuft by increased extracellular matrix, obliterating the glomerular capillary lumen or lamina, often seen with an adhesion. It is proposed that podocyte depletion leads to the appearance of bare areas of basement membrane and the formation of an adhesion to Bowman’s capsule, which is considered the earliest feature of FSGS.¹³ We show that a single adhesion without identifiable segmental sclerosis, insufficient for a definite FSGS diagnosis, can have an elevated urinary MAC. Pathology assessment in FSGS can be difficult when few glomeruli are sampled, and urinary sC5b9 could increase diagnostic precision. Our results also support that the proportion of glomeruli with segmental sclerosis does not reflect disease activity, given the absence of association between proteinuria and urinary sC5b9 or C5a.

In addition to a potential diagnostic use to distinguish MCD from FSGS, we previously reported on the prognostic value of urinary sC5b9 in FSGS. Compared to proteinuria, the reduction in urinary sC5b9 with a remission was greater. When a relapse occurred, levels of sC5b9 increased significantly more than proteinuria, supporting urinary sC5b-9 as a more sensitive marker of immunologic activity.¹⁴

Some biomarkers have been put forward to differentiate these 2 entities. The proposed biomarkers for MCD were vascular permeability factor miming hemopexin,¹⁵ angiopoietin-like-4 secreted by the podocyte,¹⁶ and cytokines IL-8 and IL-13.¹⁷ For FSGS, the soluble form of the urokinase plasminogen activator receptor,¹⁸ apoA1b (an isoform of ApoA1),¹⁹ anti-CD40 antibody,²⁰ and cardiotrophin-like cytokine factor 1²¹ have been studied. Unfortunately, these biomarkers are present in other glomerulopathies and are considered insufficiently specific for clinical use. More recently, the anti-nephrin antibodies have been put forward as a biomarker for diagnosis and response to therapy in a subset of patients with MCD.³ In that series, 29% (18/62) of patients with MCD had circulating nephrin autoantibodies, which correlated with disease activity. Interestingly, the review of cases with MCD identified delicate punctate staining for IgG by routine IF in a subset of patients. Subsequent analyses found specific colocalization of nephrin with punctate IgG.³ Moreover, the nephrin autoantibodies were correlated to disease activity since the antibodies were decreased or absent in patients with partial or complete remission.

Another finding was the lack of increased urinary biomarkers in primary FSGS. Although patients with primary disease usually have more severe proteinuria, we did have some secondary cases with very elevated measurements, such as case number 3 with severe pamidronate toxicity. The proposed classification system of FSGS lesions includes primary (autoimmune), secondary (maladaptive, toxic, or viral), genetic, and uncertain etiology.²² Those of undetermined cause may still be autoimmune, although we currently cannot identify them.²³ Our definition of primary disease was strict, requiring a complete nephrotic syndrome that responds to immunosuppression, and milder forms of primary disease may have been considered of uncertain cause.

The small number of patients is a limitation of this study, and our results require validation. We also underline that acute kidney injury may influence urinary sC5b9. We used the sample with the highest level of proteinuria in each patient; we could not assess confidently and retrospectively whether a recent infection had occurred, triggering a flare, which could account for the elevated complement activation fragments.

In conclusion, we found no complement activation fragments in MCD as opposed to FSGS, despite similar levels of proteinuria, allowing us to distinguish the 2 entities and supporting the role of complement activation in the pathogenesis of segmental sclerosis. Our results also raise the question of whether complement inhibition should be tested in FSGS, given its poor outcome and limited treatment options.^2,24

Disclosure

VR has received consultant fees from Vertex Pharmaceuticals and honoraria from Janssen. A-LL has received honoraria from Alexion and has participated on an advisory board for Novartis and Alexion. SB presented for GlaxoSmithKline. GB was a consultant for Otsuka and Alnylam. All the other authors declared no conflicting interests.

Supplementary Material

Figure S1. Urinary sC5b9 in focal segmental glomerulosclerosis and minimal change disease with patient numbering.

Figure S2. Urinary C5a in focal segmental glomerulosclerosis and minimal change disease with patient numbering.

Figure S3. Associations between urinary complement activation biomarkers and age and eGFR.

STROBE Statement (PDF)