Complement Mutation-Associated De Novo Thrombotic Microangiopathy Following Kidney Transplantation


* Corresponding author: Véronique Frémeaux-Bacchi,


Mutations in one or more genes encoding complement-regulatory proteins predispose to atypical hemolytic uremic syndrome (aHUS) and its recurrence following kidney transplantation. We evaluated plasma complement level and performed a screening for mutations in genes encoding complement Factors H and I (CFH, CFI) and membrane cofactor protein (MCP) in 24 kidney transplant recipients experiencing de novo thrombotic microangiopathy (TMA). Six patients presented with low C3 and/or low Factor B levels suggestive complement alternative pathway. A mutation in the CFH or CFI gene was found in 7/24 patients (29%), two of whom had a mutation in both genes. On the contrary, no mutation was identified in a control kidney transplant patients group (n = 25) without TMA. Patients with or without mutations were similar with regard to clinical features. Eight out of 24 patients lost their graft within 1 year of posttransplantation including six patients with a CFH mutation or a decrease of C3 or CFB in plasma. To conclude, kidney transplant patients with de novo TMA exhibit an unexpectedly high frequency of CFH and CFI mutations. These results suggest that genetic abnormalities may represent risk factors for de novo TMA after kidney transplantation and raise the question of the best therapeutic strategy.


Thrombotic microangiopathies (TMA) are microvascular occlusive disorders characterized by hemolytic anemia caused by fragmentation of erythrocytes and thrombocytopenia. These, in turn, are due to increased platelet aggregation and thrombus formation, which eventually lead to disturbed microcirculation and reduced organ perfusion. When TMA appears after renal transplantation and affects the graft, it is useful to distinguish between recurrent hemolytic uremic syndrome (HUS), appearing in recipients with HUS as their primary renal disease, and de novo TMA. Primary HUS encompasses two distinct entities. Most cases of diarrhea-associated HUS are caused by Shiga-toxin-producing bacteria, particularly Shiga-toxin-producing Escherichia coli (STEC), O157:H7 that is directly responsible for endothelial injury. If permanent end-stage renal disease occurs in 10% of patients after the acute phase, the outcome of renal transplantation is characterized by the absence of recurrence of the disease (1–3). There are also atypical forms that occur less frequently (atypical HUS, also termed as non-shigatoxin-associated HUS; aHUS) that are unrelated to Shigatoxin infections and have been associated with mutations in the genes that encode complement components or regulators. In 50% of these patients, mutations have been identified in the genes encoding the fluid-phase complement inhibitor Factor H (CFH), the serine protease Factor I (CFI) or the surface-bound regulator membrane cofactor protein (MCP, CD46) (4–6). The recurrence rate after transplantation varies widely according to the genetic abnormalities that are present. In recent studies, recurrence was observed in more than 70% of recipients with a CFH mutation (7).

De novo TMA typically develops in the early posttransplant period, but it may also develop 2–6 years after transplantation. Estimates of the incidence of de novo TMA after kidney transplantation vary between 1% and 14% (8,9). It is unclear what risk factors are involved, but ischemia-reperfusion injury, acute rejection, viral infections and immunosuppressive drugs such as calcineurin inhibitors (CNIs, cyclosporine and tacrolimus), OKT3 or sirolimus have all been associated with the development of TMA after kidney transplantation (10–12). The mechanisms by which de novo TMA is induced, however, are poorly understood. Complement abnormalities have been shown to predispose to aHUS and recurrence after transplantation, so we investigated whether there is an association between protein complement regulatory mutations and risk of de novo TMA posttransplant. We screened for CFH, MCP and CFI mutations in 24 kidney transplant patients who developed de novo TMA.

Patients and Methods


Twenty-four kidney transplant recipients with a history of posttransplantation de novo TMA from five French transplant centers were retrospectively recruited. The biological criteria for TMA at diagnosis were thrombocytopenia (platelets <100 000/mm3) and/or microangiopathic hemolytic anemia (Hb <10 g/dL) with lactate dehydrogenase (LDH) >2-fold higher than normal values associated with a negative Coomb's test and/or acute renal failure defined as plasma creatinine >120% of baseline. We included patients with at least one biological criterion and histological criterion of TMA. The exclusion criteria were pediatric recipients, HUS, aHUS or TMA on native kidneys.

The following data were recorded for each patient: recipient's age and gender, initial nephropathy, duration of dialysis pretransplant, number of kidney transplants, presence of panel reactive antibodies (PRA), deceased or living donor, cold ischemia time, use of CNI therapy (cyclosporine or tacrolimus), occurrence of delayed graft function (DGF) defined as ≥1 dialysis session in the first week post transplant, delay between transplantation and diagnosis of de novo TMA, renal function as assessed by serum creatinine on the last visit or day prior to TMA diagnosis and at the time of diagnosis, incidence of graft loss.

Histological assessment

Histological criteria for thrombotic microangiopathy were: a ‘double contour’ aspect on silver stains, presence of intracapillary thrombosis, focal necrosis with or without crescents and mesangiolysis, which corresponds to loss of glomerulus architecture with apoptosis of mesangial cells. Acute arterial changes were assessed in terms of severity, ranging from mild (defined as swelling of the endothelium) to severe (defined as the presence of media fibrinoid necrosis and thrombosis of the lumen). Immunofluorescence microscopy was used to demonstrate the deposition of fibrin or fibrinogen in the glomeruli and the mesangium as well as within the vessel walls. Vascular lesions were evaluated by light microscopy and immunofluorescence study and graded according to Banff classification. C4d staining was performed by immunofluorescence on cryosections and by immunohistochemistry on paraffin sections in cases 13 and 9, respectively. Diffuse staining in peritubular capillaries was considered as positive.

Complement assays and genetic screening

Analyses were performed using EDTA plasma samples at the immunology laboratory of the Georges Pompidou European Hospital, Paris, France. Plasma concentrations of CFH and CFI were measured by ELISA while concentrations of C4, C3 and complement Factor B (CFB) were determined by nephelometry (Dade Behring, Deerfield, IL). Membrane expression of CD46 was analyzed on granulocytes from patients using phycoerythrin (PE)-conjugated antibodies (clone MEM 258, Serotec, Oxford, United Kingdom). All CFH, MCP and CFI exons were sequenced as previously described (13–15).

The same analysis was performed in a group of 25 kidney transplant recipients. The sole criterion used to select patients included in the control group was kidney transplant recipients who did not develop de novo TMA after renal transplantation. They received the same immunosuppression (induction with basiliximab or Thymglobuline® together with steroids, MMF and an anticalcineurine). The absence of histological de novo TMA was evaluated on a routine biopsy performed either at month 3 posttransplant or at 1-year posttransplant. The initial nephropathy was a nephroangiosclerosis (n = 8), a chronic glomerulonephritis (n = 10), an undetermined nephropathy (n = 4), an autosomal dominant polycystic kidney disease (n = 1), an Alport syndrom (n = 1) and a nephronophtis (n = 1).

Statistical analysis

Data were expressed as mean ± SD for continuous variables and as percentage of the population for categorical variables. Mean values were compared by nonparametric tests including the Mann–Whitney U-test for unpaired comparisons and the Wilcoxon signed-rank test for paired comparisons between the two groups. The chi-square test was used to analyze differences in categorical variables. P-values ≤0.05 were considered statistically significant. All analyses were performed using StatView 5.0.1 (SAS Institute, Inc., Cary, NC).


Clinical and biological patient characteristics

All participants gave informed consent prior to genetic testing. Twenty-four patients with de novo TMA were recruited to the study. The initial causes of nephropathy were chronic glomerulonephritis (n = 11), ureteral vesical reflux (n = 2), Alport nephropathy (n = 1), nephroangiosclerosis (n = 6) and undetermined renal disease (n = 4). Among patients with chronic glomerulonephritis, 4/11 (36%) presented with membranoproliferative glomerulonephritis (MPGN) type I with no detectable C3 nephritic factor (Table 1).

Table 1.  Patient's clinical data
PatientAgeGenderRenal diseaseTime on dialysisRT rankISTMA delay (days)Renal function (μmol/L)1STOP ISPE/FFPGraft loss 1 year after TMA
  1. 1Renal function at the diagnosis of TMA.

  2. Alp = Alport syndrome; CGN = chronic glomerular nephritis; CrGN = crescent glomerulonephritis; CsA = cyclosporine; FSGS = focal glomeruloscerosis; FK = tacrolimus; IgAN = IgA nephritis; IS = Immunosuppressive treatment; MPGN = membranoproliferative glomerulonephritis; NAS = nephroangiosclerosis; Rapa = sirolimus; RVU = reflux vesico ureteral; Und = undetermined.

  3. *No calcineurin inhibitor or sirolimus.

  4. IS = immunosuppressive; PE = plasma exchange; FFP = fresh frozen plasma.

 147FNAS 31FK10284YYY
 349MCrGN 41CsA32581YYY
 428FIgAN 71FK690290NYN
 545HNAS 61CsA90223YYN
 660MUnd 11FK120230NNN
 750MMPGN 71FK1850NYN
1032MUnd 72FK409243YNY
1124FLGM 41*404293NY
1230MNAS 61Rapa2850NYN
1336FRVU 71CsA2300519YYN
1448FRVU 21*15850NN
1658MUnd 11CsA15142YNN
1839MCGN 41CsA3850YYN
1948MNAS 61CsA4185YYN
2029MMPGN 21CsA10534YNY
2143FMPGN 32FK94204YYY
2252FNAS 11Rapa7170YNN
2336MNAS 21CsA35409YYN
2451FFSGS 41CsA45 90YYN

Nineteen patients received a first kidney transplant while five patients had undergone a previous kidney transplant, two of whom had lost their previous graft due to TMA. All patients received a deceased donor transplant. Mean age at transplantation was 46 ± 12 years. The cold ischemia time was 22 ± 6 h and 8/24 (33%) experienced delayed graft function. Six patients had panel reactive antibodies >0% and four patients experienced acute rejection before the onset of de novo TMA. The mean time between transplantation and onset of de novo TMA was 205 days (range 1 day to 6 years), with 16/24 patients (67%) developing de novo TMA during the first 3 months posttransplant. Of note, 19/24 patients were receiving a CNI-based regimen (cyclosporine, n = 10; tacrolimus, n = 9). All patients received steroids and mycophenolate mofetil.

Biological characteristics were significantly different between the last biological analysis before TMA diagnosis and the one at time of diagnosis: hemoglobin (8.5 g/dL vs 10 g/dL, p < 0.05), LDH (1083 IU/L vs. 549 IU/L, p < 0.01), platelet count (115 000/mm3 vs. 208 000/mm3, p < 0.01) and serum creatinine (441 vs. 284 μmol/L, p < 0.01).

On the renal biopsy performed at TMA diagnosis, four patients had isolated glomerulus lesions (P2, P7, P16, P21) four patients had isolated arterial lesions (P9, P12, P19, P21) and the other 16 patients had arterials and glomerulus lesions. Three patients had CNI toxicity (P4, P13, P21). C4d staining was available in 22 out of 24 cases. C4d staining was positive in three instances: one case of acute humoral rejection (P2), one case of chronic active humoral rejection (P13) and one case of isolated C4d positive staining (P17).

After diagnosis of TMA, CNI treatment was interrupted in 15/19 patients and 15/24 patients received plasma therapy (fresh-frozen plasma and/or plasma exchange) (Table 1).

Complement component assessment

Plasma complement levels at time of genotyping are shown in Table 2. Three out of the 24 patients had a decreased C3 concentration (P3, P8, P11). In one patient, a low C3 level was associated with a low level of Factor B (P3), a finding suggestive of mild alternative pathway activation. Three patients exhibited a decreased CFB with C3 in the normal range (P1, P4, P10). Eighteen patients presented with normal concentration of C3 and CFB and no detectable complement activation. CFH concentration was in the normal range in all patients. Three out of 24 patients presented with a decreased CFI concentration (P1, P4, P8). Expression of MCP was within normal range for all patients except one.

Table 2.  Plasma complement level and molecular genetic abnormalities
PatientsC3 (mg/L)C4 (mg/L)CFB (mg/L)CFH (mg/L)CFI (mg/L)MCP (MFI)Mutations
  1. NA = Not available.

 169220380510321201 N516KG144Dno
 359325786673961006 K1186TI322Tno
 71040 23213861273612noIVS12 + 5no
 8362 74NA39835643nonono
10674153 7557652684nonono
121280 266NA65376638nonono
141130 455NA69979747nonono
15817158143469441093 nonono
161040 279238643701064 nonono
17994164189592871938 nonono
181380 35323065882685nonono
201000 26514278092765nonono
21677370NA587691871 nonono
Normal Values(660–1250)(93–320)(90–320)  (338–682)(42–78)(600–1500) 

Molecular characterization of mutations

The CFH, CFI and MCP genes including all exons and their flanking regions were analyzed by direct sequencing. Two patients had two mutations (one each on the CFH and CFI genes), four patients had a mutation on the CFI gene and one patient had a single mutation on CFH gene (Table 3). All mutations were heterozygous and had not been detected in 100 normal individuals from the same ethnic background. No genetics abnormality in CFH, CFI and MCP genes was identified in the 25 kidney patients without TMA.

Table 3.  Genetics abnormalities in the 7 patients
PatientCause of ERSDMutationProtein domainMutation characteristic
  1. NAS = nephroangiosclerosis; CrGN = crescent glomerulonephritis; IgAN = IgA nephritis; MPGN = membranoproliferative glomerulonephritis; Und = undetermined.

  2. *Residue numbering according to the nomenclature recommendations; residue numbering without the 18 amino acid signal sequence corresponding to that previously used.

 1NASp.Asn516Lys (c.1548T>A)SCR9SCR9 is implicated binding of CFH to the C3c and heparin
 2CrGNp.Gln950His (c.2850G>T)SCR16SCR16 is implicated binding of CFH to C3b/C3d, as well as to endothelial cells. Q950H has been reported in one patient with aHUS (28)
 3CrGNp.Lys1186His (c.3557A>C)SCR 20SCR20 is highly implicated binding of CFH to C3b/C3d, as well as to endothelial cells
 4IgANp.Ser90Asn (c.269G>A); S72NFIMACAssociated with a reduced CFI concentration seem to result in quantitative defect
 1NASp.Gly162Asp (c.485G>A); G144DCD5Associated with a reduced CFI concentration seem to result in quantitative defect
 5NASp.Ile416Leu (c.426A>C); I398LSPAssociated with a normal CFI concentration; presumed functional deficiency has not yet been defined; reported in one patient with aHUS (4)
 6Undp.Ile306Val (c.916A>G); I288VLDLRA-2Associated with a normal CFI concentration; presumed functional deficiency has not yet been defined
 3CrGNp.Ile340Thr (c.1019T>C); I322TBetween Heavy and light chainI322T has been reported in one patient with aHUS (32) Complete loss C3b cofactor activity (33)
 7MPGNIVS12 + 5 (c.1536 + 5 G>T)Spice defectPreviously reported in patients with aHUS (4,6)

Several mutations were identified (Table 3). A nucleotide substitution leading to a change in an amino acid localized in the SCR 9 (Asn516Lys), in the SCR 16 (Gln950His) and in the SCR 20 (Lys1186His) on the CFH gene was identified in three patients (P1 to P3) who presented with a normal CFH plasma level. Patients 1 and 4, who presented with a mild decrease in CFI level, had a nucleotide substitution leading to a missense mutation located in exon 4 (G144D) and exon 3 (S72N) in CFI gene, respectively. The molecular abnormalities found in the CFI gene led to a missense mutation located in exon 9 (I398L) and in exon 5 (I288V) in patients 5 and 6, respectively, who presented with a normal concentration of CFI. Patient 7 presented with a heterozygous exon 12 donor splice site change (IVS12 + 5; CAGgtaagt to CAGgtaatt). This nucleotide change was found in 1/200 French healthy donors and was detected in cases of aHUS (4,6). No mutation was detected in the 14 exons of the MCP gene in any patient.

Relationship of patient characteristics and clinical outcomes with regard to mutations

Within the first year, 8/24 patients had lost their graft (33%) including 3/7 in a mutation (42%) and 5/17 without (29%). Among the 8 patients who lost their graft at 1 year, three patients had a CFH mutation (P1 to P3), two patients had a low level of C3 (P8, P11) and one patient had a low level of Factor B (P10) suggesting a mild alternative pathway activation. The two patients (P8, P21) who lost the previous graft due to TMA had a membrano-proliferative nephropathy. We did not find mutations in these two patients but one of them (P8) had a low level of C3.

There were no statistically significant differences in gender, age, immunosuppressive treatment, biological characteristics of TMA, histological presentation or CNI toxicity at time of TMA according to the presence or the absence of mutations.

Nineteen of the 24 patients, including the seven cases with genetic mutations, received a CNI, cyclosporine (n = 10) or tacrolimus (n = 9). TMA treatment (plasmatherapy and/or withdrawal of immunosuppressive drug) was similar between those with or without mutations. C4d staining was positive in three patients. One of the patients (P2) with CFH mutation transplanted for the third time developed an acute renal failure in the first week induced by TMA and acute humoral rejection. He lost his graft in the first 3 months. The second patient (P13) developed a TMA 6 years after the renal transplantation and lost her graft in few months because of TMA and chronic humoral rejection. The third patient (P17) developed TMA 15 days after renal transplantation and had an isolated C4d staining positive without other histological criteria for humoral or cellular rejection on biopsy. Without any treatment for the rejection or TMA, he kept his graft with good function.


Our results have demonstrated for the first time an association between mutations in genes coding for complement regulation proteins and de novo TMA after kidney transplantation.

Seven of the 24 patients (29%) with de novo TMA presented with a CFH or CFI mutation, similar to the proportion observed in a series of patients with aHUS (14–30%) (4,6,16). The presentation of de novo TMA after kidney transplantation is variable, with some patients exhibiting clinical and biological features of TMA, some showing only a progressive renal failure, and others having only histological lesions (9,17). In our study, 21 patients presented with severe clinical TMA comprising microangiopathic hemolytic anemia, thrombocytopenia and acute renal failure; the remaining three patients had only an acute renal failure without the typical clinical triad. All but three of the patients were receiving CNI or sirolimus immunosuppressive treatment and two patients presented also evidence for humoral rejection, which are believed to contribute to risk of de novo TMA (18,19). Against this background, we cannot exclude the possibility that genetic susceptibility to de novo TMA is restricted to the severe form.

Several diseases are associated with defective control of the alternative complement pathway, including aHUS, Membranoproliferative glomerulonephritis type I and type II (MPGN I; MPGN II) and recently described age-related macular degeneration (AMD) (20,21). Several studies have emphasized the role of the alternative pathway (AP) of the complement system in the pathogenesis of aHUS for which no clear triggering conditions have been identified. aHUS-associated mutations have been identified in the genes coding for the components of the amplification convertase CFB (CFB) as well as the complement regulators Factor H, membrane cofactor protein (MCP/CD46) and CFI (14,22–24). The clinical outcome of aHUS is unfavorable, typified by relapse and progression to end-stage renal failure in up to 50% of the cases. After kidney transplantation, recurrence is observed in 70% of patients with CFH or CFI mutations of whom more than 80% lose their graft (7). The prognosis of aHUS induced by MCP mutation is better, and only two cases of recurrence after renal transplantation have been described (13,25). Heterozygous mutations of CFH, CFI and MCP have been recently associated with primary glomerulonephritis with isolated C3 deposits and homozygous Factor H deficiency is associated with membranoproliferative glomerulonephritis type I and II (15,26). None of the 24 patients were investigated for complement abnormality before the transplantation and none of these diseases was clearly associated with genetic complement abnormalities. A certain degree of clinical overlap between malignant nephroangiosclerosis and aHUS has been described, but the relationship between these two entities remained undetermined. One observation documents the association of IgA nephropathy and CFH deficiency without evidence of a role of Factor H in pathogenesis of this disease (27). Thus, a link between the initial nephropathy and complement abnormalities could not be excluded. In our population, we identified three mutations of the CFH gene, located throughout the gene but within the C3b and anionic-heparin binding sites essential for regulation of the alternative pathway on cellular surfaces. One mutation located in SCR16 has previously been reported in patients with aHUS (28). Recent studies have shown that the mutated protein could not bind efficiently to endothelial cells, suggesting that dysfunction could lead to uninhibited complement activation on the surface of these cells (29,30). In our study, five patients were found to have a CFI mutation. CFI is a serine protease that cleaves C3b in the presence of cofactor proteins (31) and it must be present in sufficient amounts to limit activity of the complement amplification convertase C3bBb and prevent complement-mediated host cell damage. Three of these mutations have already been reported in patients with aHUS (I322T, I368L and IVS12 + 5) (4,6,32). Two of our patients presented with low CFI level suggesting a quantitative CFI deficiency. Recently, Kavanah et al. demonstrated that the I322T mutation results in secreted proteins that lack C3b cofactor activity (33). Interestingly, we identified no mutation in the MCP gene within our series of patients. MCP is expressed in endothelial cells of the kidney and thus the presence of MCP mutations of donor transplant should be screened. It is becoming clear that complement mutations may be associated with a large spectrum of functional consequences ranging from a complete defect to no detectable implications, and that this might play a role in the level of endothelial protection. As previously reported in patients with aHUS, we isolated also kidney transplant patients without genetic abnormality in CFH, CFI or MCP genes (4,6). Over the last years other susceptibility factors have been implicated in aHUS patients. Mutations in CFB, which increased the affinity between C3b and Factor B and thus stabilized the alternative C3 convertase have been described in two patients with aHUS (24). In addition, a common CFH haplotype with a deletion incorporating the genes encoding CFH-related proteins 1 and 3 (CFHR1 and CFHR3) has been shown to be associated with aHUS (34). Our patients were not screened for these mutations, and we assume that this study may likely underestimate the frequency of genetic abnormalities in this population. Three of the five patients without detected mutations, who lost graft from TMA at 1 year, had either an unexplained low C3 (P8, P11) or low CFB (P10), which supports this hypothesis.

Our study did not investigate protein activity of ADAMTS-13, or the presence of ADAMTS-13 inhibitors. Although, it may be difficult to differentiate complement-mediated HUS with ADAMST13-related TTP, deficiency in ADAMTS 13 activity has been rarely identified in patients experiencing posttransplant TMA. In our knowledge, two cases have been reported in the literature (35,36). Our results suggest an important role for complement in the protection of kidney cells after kidney transplantation although its role in TMA remains to be fully elucidated.

Treatment for de novo TMA has not been well defined. Usually, therapy consists of complete withdrawal of the CNI, a switch from cyclosporine to tacrolimus, or a switch to mTOR inhibition (sirolimus). However, not all patients respond, and withdrawing CNI treatment increases the risk of acute rejection (37). Addition of plasma exchange may salvage the graft in about 80% of cases (38). In this study, the management of TMA treatment was no different in the two groups and the physicians in charge of the patients were not aware of complement mutation information. In our series, 16/19 patients discontinued CNI and 15/24 patients received plasma therapy.

Interestingly, six out of eight patients who lost their graft within 1 year posttransplantation presented with a mutation in CFH gene or with an ‘unexplained’ complement alternative pathway. Outcomes differed from those seen in recurrent aHUS (39). In aHUS recurrence, graft survival is <50% at 1 year; in one series 10 of 15 patients had at least one graft failure and 12/24 grafts failed during the first year (4). Thus, all aHUS patients being considered for renal transplantation should undergo screening for mutations in complement regulators. In our population, graft survival was 67% at 1 year after TMA diagnosis and while all patients with CFH mutations lost their graft, there was no graft loss among patients with CFI mutations. The difference of outcome remains unexplained. The prognosis for patients with TMA seems less severe than for aHUS recurrence, but it is still poor.

No mutation was found neither in 25 kidney transplant recipients without de novo TMA nor in 100 healthy controls. In striking contrast, a mutation in the genes encoding CFH and/or CFI was found in 29% of the patients with de novo TMA. Although the number of patients was small, our results might suggest that genetic abnormality in CFH and CFI genes are risk factors for de novo TMA after kidney transplantation. The incomplete penetrance of HUS associated with mutations in CFH and CFI is described, suggesting that the phenotype of the disease may be modulated by other genetic or environmental factors (5). We hypothesize that transplantation, CNI or acute/chronic rejection might be the triggering event of the first episode of TMA. In this setting, endothelial activation triggered by the graft procedure or the immunosuppressive drug or acute/chronic rejection could be undesirably enhanced by excessive complement activation secondary to impaired regulation. We therefore suggest to screen for complement abnormalities including a genetic study kidney recipient at the time of the first episode of de novo TMA.

In summary, there is a genetic susceptibility to de novo TMA that is similar to that already identified for recurrent aHUS, and it is associated with inappropriate regulation of the alternative complement pathway. It is highly likely than other genetic mutations in various complement factors play a significant role, at least as cofactors, and these remain to be defined.


With thanks to Nelly Poulain, Florence Marliot and Jacques Blouin for expert technical support. We are particularly grateful to Dr D Nochy and Dr A Loupy for performing the C4d staining, Dr Noel and Dr Buob for kidney biopsy analysis and the clinicians who referred their patients for complement investigations. The study received no external funding.