Hemolytic Uremic Syndrome in Pregnancy and Postpartum

Abstract

Background

Pregnancy is associated with various forms of thrombotic microangiopathy, including hemolytic uremic syndrome. A previous small French study suggested that pregnancy-associated hemolytic uremic syndrome was to be included in the spectrum of atypical hemolytic uremic syndrome linked to complement alternative pathway dysregulation.

Design, setting, participants, & measurements

We sought to retrospectively analyze the presentation, outcome, and frequency of complement alternative pathway gene variants in a larger international (France, United Kingdom, Italy) cohort of patients with pregnancy-associated hemolytic uremic syndrome.

Results

Eighty-seven patients with pregnancy-associated hemolytic uremic syndrome were included. Hemolytic uremic syndrome occurred mainly during the first pregnancy (58%) and in the postpartum period (76%). At diagnosis, 56 (71%) patients required dialysis. Fifty-six (78%) patients underwent plasma exchanges, 21 (41%) received plasma infusions, and four (5%) received eculizumab. During follow-up (mean duration of 7.2 years), 41 (53%) patients reached ESRD, 15 (19%) had CKD, and 18 (28%) patients experienced hemolytic uremic syndrome relapse. Twenty-four patients (27%) received a kidney transplant and a recurrence of hemolytic uremic syndrome occurred in 13 (54%) patients. Variants in complement genes were detected in 49 (56%) patients, mainly in the CFH (30%) and CFI genes (9%).

Conclusions

Pregnancy-associated hemolytic uremic syndrome and atypical hemolytic uremic syndrome nonrelated to pregnancy have the same severity at onset and during follow-up and the same frequency of complement gene variants.

Introduction

Hemolytic uremic syndrome (HUS) is a rare and severe form of thrombotic microangiopathy associated with a poor renal prognosis. It is characterized by the association of mechanical hemolytic anemia, thrombocytopenia, and kidney failure (1). HUS arises from an insult to endothelial cells that in turn may result from distinct pathogenic mechanisms: verotoxin-induced endothelial cell activation and apoptosis in Shiga toxin–induced HUS, acquired or constitutional complement alternative pathway dysregulation leading to complement-induced endothelial cell damage in atypical HUS, or various, more or less well-defined patterns of endothelial cell lesions in the heterogeneous group of secondary HUS associated with autoimmune diseases, drugs, infections, and pregnancy. Pregnancy carries a high risk for various forms of thrombotic microangiopathy, including ADAMTS13 deficiency–associated thrombotic thrombocytopenic purpura (2), but also HUS (3). The diagnosis of pregnancy-associated HUS can be difficult with preeclampsia, HELLP (Hemolysis, Elevated Liver enzymes, and Low Platelet count) syndrome, and in some instances severe postpartum hemorrhage presenting with similar features (4,5). Moreover, until a few years ago, pregnancy-associated HUS was included in the group of secondary forms of thrombotic microangiopathy.

The identification of complement alternative pathway dysregulation as a major risk factor for atypical HUS (6,7) has led several authors to reconsider the pathogenesis of secondary forms of HUS and to assess whether, in these settings, complement alternative pathway dysregulation combined with specific precipitating events could lead to thrombotic microangiopathy. Previously, we tested this hypothesis in a small series of pregnancy-associated HUS and showed that 86% of patients with mainly postpartum HUS harbored complement gene variants and had an initial presentation and outcome similar to nonpregnancy-related atypical HUS (3). This first study was rather limited as it included only 21 patients from a single country. Thus, there remained some uncertainties and controversies regarding the pathogenesis and consequently the treatment of pregnancy-associated HUS, a crucial question since the clinical availability of the first complement inhibitor, eculizumab (4). We sought to analyze the presentation, outcome, and frequency of complement alternative pathway gene variants in a large international cohort of patients presenting with pregnancy-associated HUS.

Materials and Methods

Patients

This was a retrospective multicenter study conducted in France, the United Kingdom, and Italy—three countries in which a HUS registry has been established. We identified through computerized databases all women with pregnancy-related HUS included in these registries between 1983 and 2013 and for whom complement workup has been performed in the three national reference centers (Department of Biologic Immunology at Hôpital Européen Georges Pompidou, Paris, France; Mario Negri Institute, Bergamo, Italy; and The UK National atypical HUS Service in Newcastle).

Definitions

All patients with a diagnosis of pregnancy-associated HUS from the three registries were included in the study. Pregnancy-associated HUS was defined as HUS occurring during pregnancy, or in the postpartum period (up to 12 weeks after delivery). HUS was defined by the association of at least three of the following criteria: mechanical hemolytic anemia (hemoglobin <10 g/dl, lactate dehydrogenase over the upper limit of normal, undetectable haptoglobin, presence of schistocytes), thrombocytopenia (platelet count <150×10³/μl), AKI (serum creatinine >1.1 mg/dl or >25% increase from baseline value), or typical features of thrombotic microangiopathy in a kidney biopsy sample (fibrin/platelet thrombi, endothelial cell swelling and detachment from the basement membrane, double contours). Patients with preeclampsia (defined by hypertension >140/80 mmHg and proteinuria >300 mg/d [8] after 20 weeks of gestation) before the development of thrombotic microangiopathy features, HELLP syndrome (defined by aminotransferase>70 U/L, lactate dehydrogenase >600 U/L, and platelet count <100 × 10³/μl) (9), massive postpartum bleeding, or other identified causes of secondary HUS (lupus, antiphospholipid syndrome) were excluded. CKD was defined by a GFR estimated using the Modification of Diet in Renal Diseases formula <60 ml/min per 1.73 m².

Medical records of included patients were reviewed and relevant clinical and biologic features were collected.

C Analysis

C evaluation and genetic analysis were performed in each reference center in the usual management of the patients. Plasma concentrations of C3, C4, factor B, factor F, and factor I, and membrane cofactor protein expression on granulocytes were measured according to local practice (6,7,10,11). All coding sequences for complement factor H, complement factor I, membrane cofactor protein, C3, factor B, and thrombomodulin genes were sequenced as described previously (6,7,11). Variants were categorized as: (1) Pathogenic: novel (not found in the general population) or rare (minor allele frequency in the general population <0.1%) variant reported to cause disease in literature; functional data indicating the variant affects protein function or expression. (2) Likely pathogenic: novel or rare variants that change protein sequence or affect splicing and with highly deleterious effects by in silico predictions but without functional data; found in disease-related functional domains. (3) Uncertain significance: novel or rare variants that change protein sequence or affect splicing with no available functional data; uncertain deleterious effects by in silico prediction (12).

All patients gave informed consent for genetic analysis according to the Declaration of Helsinki.

Statistical Analyses

Data are presented as percentages or means±SD. The Wilcoxon test was performed for quantitative variables, and Fischer exact test for qualitative data. All analyses with P value <0.05 were considered statistically significant.

Results

Eighty-seven patients who presented with pregnancy-associated HUS between 1983 and 2013 were included in the study. The distribution by country and decade is presented in Supplemental Table 1. Pregnancy-associated HUS represented 16% (87 out of 547) of HUS cases occurring in women aged 18–45 years reported in the three national registries. The main characteristics of these patients are summarized in Table 1.

Table 1.

Characteristics of 87 patients with pregnancy-associated hemolytic uremic syndrome

Characteristics	Number (%)/Mean±SD
Number of patients	87
Age at HUS onset, yr	29±6.0
Number of previous pregnancies	0.7±1.2
Rank of pregnancy HUS was diagnosed in (n=83)
First	48 (58)
Second	23 (28)
Third	5 (6)
Fourth or subsequent	7 (8)
Preeclampsia during previous pregnancies (n=53)	5 (9)
Fetal loss during previous pregnancies (n=49)	10 (20)
Familial history of atypical HUS	14 (16)
Personal history of atypical HUS	7 (8)
Timing of HUSa
Postpartum	63 (76)
During pregnancy	20 (24)
Features at hemolytic and uremic syndrome onset
Serum creatinine, mg/dl	6.1±5.2
Dialysis	56 (71)
Platelet count ×103, per μl	97±99
Hemoglobin, g/dl (n=66)	7.8±1.9
Lactate dehydrogenase, U/L (n=56)	2225±1617
Neurologic involvement	7 (9)
Other extrarenal manifestationsb	4 (6)
Treatment
Number of patients who underwent plasma exchange (n=72)	56 (78)
Number of plasma exchange sessions performed per patient (n=41)	13±10
Number of patients who received plasma infusion (n=51)	21 (41)
Number of patients who received eculizumab	4 (5)
Steroids (n=60)	16 (27)
Otherc	3 (5)

The numbers of patients for whom data are available are reported in brackets. HUS, hemolytic uremic syndrome.

^aTiming of HUS is unknown for four patients.

^bPulmonary edema (n=2), pulmonary embolism (n=1).

^cIntravenous Igs (n=2), rituximab (n=1).

Patient Characteristics at Presentation

Mean age at the time of pregnancy-associated HUS was 29±6 years (Table 1). Fourteen patients (16%) had a familial history (at least one affected family member) of atypical HUS. Seven (8%) patients had a personal history of atypical HUS and had experienced one (n=6) or several (four in one patient) episodes of atypical HUS not related to pregnancy before presenting with pregnancy-associated HUS. One patient had a documented complement gene variant (C factor H) before pregnancy. Five patients had previously undergone plasma exchanges for atypical HUS, and two had stage 3 CKD related to atypical HUS. The risk for HUS occurrence was similar for the first pregnancy (48 out of 82, 58%) and for subsequent (second or subsequent) pregnancies (42%). HUS occurred mainly in the postpartum period (n=63, 76%; mean time of 14±12 days after delivery) (Figure 1), regardless of the rank of pregnancy. Twenty patients (24%) presented with HUS during pregnancy, mostly (n=18, 77%) in the third trimester. AKI was severe with 56 (71%) patients requiring dialysis at presentation. In contrast, thrombocytopenia was mild (mean platelet count 97±99×10³/μl) and even absent in 13 (15%) patients. All patients had aminotransferases levels <70 U/L and ADAMTS13 activity >10%. Extrarenal manifestations were noted in 11 patients (14%). Kidney biopsy performed in eight patients (mostly with normal platelet count) disclosed typical features of thrombotic microangiopathy.

Figure 1.

Pregnancy-associated HUS occured mainly in the postpartum period. HUS, hemolytic uremic syndrome.

Treatment

Fifty-six patients (56 of 72, 78%) underwent plasma exchanges as first-line therapy, 21 (21 out of 51, 41%) received plasma infusions, and 16 (16 out of 60, 27%) received corticosteroids (Table 1). Eculizumab was used in four cases starting in 2011, as a second-line therapy after plasma exchanges. Three patients had HUS in the postpartum period, one during pregnancy, but eculizumab was started after delivery in all cases. The four patients had severe AKI requiring dialysis but no extrarenal manifestations and received eculizumab 4 days, 5 days, 1 month, and 2 months after HUS diagnosis. Three of them had complement gene variants (an isolated complement factor H variant [n=2] and combined complement factor H/complement factor I variants [n=1]).

Outcome

Short-Term Outcome.

No maternal death was reported. Fetal or neonatal death occurred in 11 (14%) cases (Table 2). Twenty-five patients (32%) reached ESRD within 3 months of first manifestations of pregnancy-associated HUS and 15 (19%) developed CKD. The risk of ESRD (29 out of 56 [51%] versus seven out of 16 [47%], P=0.77) and CKD (11 out of 56 [20%] versus three out of 16 [20%], P>0.99) did not differ between patients treated with plasma exchanges and those who did not undergo plasma exchanges. Among the four patients treated with eculizumab, three (two with complement gene variants, one without) had a complete recovery of kidney function. The remaining patient treated with eculizumab 2 months after diagnosis remained dialysis-dependent.

Table 2.

Outcome of 87 patients with pregnancy-associated hemolytic uremic syndrome

Outcome	Number (%)/Mean±SD
Duration of follow-up, yr (n=78)	7.2±5.2
Patients who reached ESRDa	41 (53)
ESRD within 3 mo of pregnancy HUS (n=78)	25 (32)
Patients with an eGFR<60 ml/min per 1.73 m2 without ESRD	15 (19)
Patients with an HUS relapse	18 (28)
Relapse in the native kidneys	8 of 62b (13)
Number of relapses	1.6±1.4
Patients reaching ESRD after a relapse	6 of 8 (75)
Relapse in the renal graft	10 of 24 (42)

eGFR, GFR estimated using the MDRD formula; HUS, hemolytic uremic syndrome.

^aNine patients progressed to ESRD during follow-up without overt hemolytic and uremic syndrome relapse. The timing of ESRD is unknown in two patients.

^bNumber of patients who did not reach ESRD within 3 mo of pregnancy-associated HUS onset.

Long-Term Outcome.

Eight patients out of the 62 (13%) who did not reach ESRD within 3 months of pregnancy-associated HUS had an HUS relapse: in the postpartum period of a subsequent pregnancy (n=1), in the setting of malignant hypertension (n=1) or an infectious illness (n=1), or without any identified triggering event (n=5). In six (75%) relapsing patients, HUS relapse led to ESRD. At last follow-up (7.2± 5.2 years), 22 out of 78 (28%) patients had an eGFR>60 ml/min per 1.73 m², 15 (19%) had CKD, and 41 (53%) had progressed to ESRD. Twenty-four patients (24 out of 87, 27%) received a kidney transplantation and a recurrence of HUS occurred in ten patients (ten out of 24, 42%). Overall, 18 patients (18 out of 87, 21%) experienced HUS relapse either in their native kidneys or in a kidney allograft.

C Workup

Available results for complement component assays and complement gene sequencing are presented in Tables 3 and 4. C3 serum level was low in 29 of 74 (39%) of patients. Twenty-three women among the 29 (79%) with low serum C3 level had complement gene variants. Novel or rare variants in complement genes (missense, nonsense variants, ins-del, and exon-intron boundary variants affecting splicing) were detected in 49 of 87 (56%) patients, most frequently in the complement factor H (n=26, 30%) and complement factor I genes (n=8, 9%). Eight patients (9%) presented combined variants: complement factor H and complement factor I variants (n=2), complement factor H and membrane cofactor protein (n=2), complement factor H and C3 (n=2), complement factor I and membrane cofactor protein (n=1), or complement factor H and thrombomodulin (n=1). One patient had two heterozygous variants in complement factor I gene. Thirty-eight patients (44%) had no detected variant in complement genes. Detailed description of the 59 variants detected in complement genes is shown in Table 4. On the basis of published functional data, low protein level, and/or high pathogenicity predicted by in silico analysis, 40 variants were considered to be pathogenic, the remaining 19 variants being of uncertain significance.

Table 3.

Results of complement component assays and complement gene sequencing in patients with pregnancy-associated hemolytic uremic syndrome

Variable	Number (%)
C component assays
Low serum C3	29 of 74 (39)a
Low serum CFH	8 of 54 (15)b
Low serum FI	5 of 43 (12)c
Low serum FB	0 of 45 (0)
Low MCP expression on granulocytes	6 of 39 (15)d
C and THBD genes sequencing (n=87)
Number of patients with a variant detected	49 (56)
Isolated CFH variant	26 (31)
Isolated CFI variant	8 (9)
Isolated MCP variant	3 (3)
Isolated C3 variant	3 (3)
Isolated FB variant	0 (0)
Isolated THBD variant	1 (1)
Combined variants	8 (9)
No variant detected	38 (44)

CFH, complement factor H; CFI, complement factor I; MCP, membrane cofactor protein; FB, factor B; THBD, thrombomodulin.

^a60% of patients with CFH or C3 variants had low serum C3 level.

^b32% of patients with CFH variant had low serum CFH level.

^c44% of patients with CFI variants had low serum CFI level.

^d100% of patients with MCP variants had low MCP expression on granulocytes.

Table 4.

Characteristics of complement gene variants identified in patients with pregnancy-associated hemolytic uremic syndrome

Patient	Gene	Variation Classification	Amino Acid Changea	Population Frequencyb	Protein Level	Protein Function	References
1	CFH	Pathogenic	p. Arg1215Gly	Novel	↔	↓	(26,27)
2	C3	Pathogenic	p.Arg592Gln	0.000008241	↔	↓	(28)
3	MCP	Pathogenic	p.Met44Leu	0.0004454	↓	N/A
4	CFH	Pathogenic	p.Arg1210C	0.0001730	↔	↓	(26,27,29)
5	CFH	Pathogenic	p.Thr645ArgfsX20	Novel	↓	N/A
6	CFH	Pathogenic	p.Trp71	Novel	↓	N/A
7	CFH	Pathogenic	c.3468dupA	Novel	↓	N/A	(30)
	CFI	VUS	Pro553Ser	0.000008243	↔	N/A	(10,31)
8	CFH	Pathogenic	p.Ser1191Leu	Novel	↔	↓	(26,32)
9	CFH	Pathogenic	p.Val1197Ala	Novel	↔	↓	(29,32,33)
10	CFH	Pathogenic	p.Arg1215×		↔	N/A
11	CFH	Pathogenic	CFH/CFHR3 hybrid	Novel	↔	↓	(34)
12	CFH	Pathogenic	c.3486delA		↓	N/A
13	CFI	Pathogenic	p.W145×	Novel	↓	N/A	(35)
14	CFI	Pathogenic	p.Ile416Leu	0.001113	↓	N/A	(6,10,36,37)
15	CFI	VUS	p.Ile578Thr	0.00002477	↔	N/A	(10)
16	CFH	Pathogenic	p.G1194D	0.00003295	N/A	N/A	(38)
	MCP	Pathogenic	p.F242C	Novel	↓	N/A	(39,40)
17	CFI	Pathogenic	p.A240G	0.0002720	↓	N/A	(7,31,36)
18	CFH	VUS	p.P279L	Novel	N/A	N/A
19	THBD	VUS	p.D163N	Novel	N/A	N/A
20	CFH	Pathogenic	K474Nfs6×
21	MCP	Pathogenic	IVS2+2T>G	0.00003311	↓	N/A	(41)
	MCP	Pathogenic	p.Y189D	Novel	↓	N/A	(41)
	CFI	VUS	p.D44N	0.00001648
22	CFI	VUS	p.D519N	0.00001658	↔	↓	(42)
23	CFH	VUS	p.N516K	0.0004046
24	CFH	Pathogenic	p.R78G	Novel	↔	↓	(38,43)
25	CFH	VUS	p.R303Q	0.00001650
26	C3	Pathogenic	p.Arg161Trp	0.000008240	↔	↓	(6,44)
27	C3	VUS	p.Ile1095Ser	Novel	NA	NA	(6)
28	CFH	Pathogenic	p.Arg53Cys	0.00001652	↔	↓	(45)
29	CFH	Pathogenic	p.Ala161Ser	0.00004124	N/A	N/A
30	CFH	Pathogenic	p.Gly397Arg	Novel	N/A	N/A
31	CFH	Pathogenic	p.Cys431Tyr	0.000008262	N/A	N/A
32	CFH	Pathogenic	p.His893Arg	Novel	N/A	N/A
33	CFH	Pathogenic	p.Val1197Ala (Hom)	Novel	↔	↓	(29,32,46)
34	CFH	Pathogenic	p.Val1197Ala	Novel	↔	↓	(13–15,18,19)
35	CFH	Pathogenic	p.Arg1210Cys	0.0001730	↔	↓	(26,27,29,47,48)
	MCP	Pathogenic	p.Tyr29Stop	Novel	↓	N/A
36	CFH	Pathogenic	p.Gln81Pro	Novel	N/A	N/A
37	CFH	VUS	p.Lys1186Thr	Novel	N/A	N/A
	CFI	VUS	p.Ile340Thr	0.00004120	↔	↓	(42)
38	C3	Pathogenic	p.Lys155Gln	0.003362	N/A	N/A
	CFH	Pathogenic	p.Lys584Stop	Novel	N/A	N/A
39	C3	Pathogenic	p.Arg161Trp	0.000008240	↔	↓	(6,44)
	CFH	VUS	p.Arg341His	0.00001653	N/A	N/A
40	CFH	Pathogenic	p.Cys864Ser	Novel	N/A	N/A
	THBD	Pathogenic	p.Ala43Thr	0.003430	N/A	N/A
41	CFI	Pathogenic	p.Arg474Stop	0.00004956	↓	N/A	(6,36,49)
42	CFI	Pathogenic	p.Gly119Arg	0.0005290	↓	N/A	(6,31,36)
43	CFH	Pathogenic	p.Gly218Glu	Novel	↓	N/A	(6,7)
44	CFI	Pathogenic	p.Gly119Arg	0.0005290	↓	N/A	(6,31,36)
	CFI	VUS	p.Gly424Asp	Novel	↔	N/A	(36)
45	MCP	Pathogenic	IVS3+2T>G	0.00003121	N/A	N/A
46	MCP	Pathogenic	p.Tyr248Stop	Novel	↓	N/A
47	CFH	VUS	p.Lys82Arg;	0.00001649	N/A	N/A
48	CFH	Pathogenic	p.Tyr1016 (Hom)	0.000008239	↓	N/A
49	CFH	Pathogenic	p.Arg161Trp	0.000008240	↔	↓	(6,44,50)

CFH, complement factor H; ↔, unchanged; ↓, decreased; MCP, membrane cofactor protein; N/A, not available; CFI, complement factor I; VUS, variant of undetermined significance; Hom, homozygous; THBD, thrombomodulin.

^aAll heterozygous unless specified.

^bFrom the Exome Aggregation Consortium (ExAc) database (http://exac.broadinstitute.org/).

Patients’ Characteristics and Outcome Depending on the Presence or Absence of complement Genes Variants

Frequency of complement gene variants in patients presenting with HUS during their first pregnancy and in those presenting with HUS during subsequent (second or subsequent) pregnancies did not significantly differ (23 out of 48 [47%] versus 23 out of 35 [65%], P=0.11) (Figure 1). Patients with complement gene variants required dialysis at presentation more frequently than those with no detected variant (35 out of 89 [81%] versus 21 out of 38 [58%], P=0.02). The frequency of neurologic involvement was similar in patients with or without complement gene variants (five out of 49 [12%] versus two out of 38 [6%], P=0.38) (Table 5). The long-term outcome of HUS was more severe in patients with documented complement gene variants compared with patients with no identified variant (Table 5). Patients with variants progressed to ESRD more frequently than patients with no variant detected (29 out of 49 [64%] versus 12 out of 38 [36%], P=0.01). The risk of relapse was also significantly greater in patients with variants compared with those without (13 out of 49 [38%] versus five out of 38 [16%], P=0.04). Among patients with mutations, there was no difference according to the type of genetic abnormality (Supplemental Table 2).

Table 5.

Main characteristics of 87 patients with pregnancy-associated hemolytic uremic syndrome with (n=49) or without complement gene variants (n=38)

Characteristics	Complement Gene Variant Detected (n=49)	No Complement Gene Variant Detected (n=38)	P Value
At presentation
Age, yr	28±6	30±6	0.06
Personal history of HUS	4 (8%)	3 (9%)	1
Onset in the postpartum	39 (79%)	28 (72%)	0.49
Need for dialysis	35 (81%)	21 (58%)	0.02
Neurologic involvement	5 (12%)	2 (6%)	0.38
Plasma exchange	30 (79%)	26 (77%)	0.80
During follow-up, n=74
Duration of follow-up, yr	6.2±3.6	6.7±4.1	0.75
Relapse	13 (38%)	5 (16%)	0.04
CKD	9 (21%)	6 (18%)	0.81
ESRD	29 (64%)	12 (36%)	0.01

Data are mean ± SD for continuous variables, and N (%) for categorical variables. HUS, hemolytic uremic syndrome.

Patients’ Characteristics and Outcome Depending on the Timing of HUS (during Pregnancy or in the Postpartum Period)

Compared with patients with HUS in the postpartum period, patients presenting with HUS during pregnancy had more frequently a personal history of CKD or HUS (25% versus 5%, P=0.02) and tended to require less frequently dialysis in the acute phase (56% versus 76%, P=0.08), even though this difference was not statistically significant (Table 6). However, the frequency of complement gene variants and the risk of HUS relapse, CKD, and ESRD did not differ between the two groups.

Table 6.

Main characteristics of 83 patients who presented with hemolytic uremic syndrome during pregnancy (n=20) or in the postpartum period (n=63)

Characteristic	HUS during Pregnancy (n=20)	HUS Postpartum (n=63)	P Value
Medical history
Personal history of HUS	4 (20%)	3 (5%)	0.03
At onset
Age, yr	29±6	29±6	0.68
Need for dialysis	10 (56%)	45 (76%)	0.08
Neurologic involvement	0 (0%)	7 (12%)	0.12
Plasma exchange	10 (59%)	44 (83%)	0.04
C gene variants detected	10 (56%)	28 (78%)	0.09
Outcome
Duration of follow-up, yr	6.9±3.2	7.1±5.1	0.80
HUS relapse	4 (20%)	14 (30%)	0.33
CKD	4 (22%)	11 (19%)	0.79
ESRD	8 (44%)	32 (55%)	0.59

Data are mean ± SD for continuous variables, and N (%) for categorical variables. HUS, hemolytic uremic syndrome.

Discussion

This is the largest to date cohort study assessing the presentation, the outcome, and the frequency of complement complement gene variants in pregnancy-associated HUS. It was performed using three distinct registries in three national reference centers for HUS. Cases were diagnosed over a rather long period of time spanning four decades, but until the very recent availability of eculizumab the care of HUS did not significantly vary.

The results indicate that pregnancy-associated HUS has a severity at onset (two thirds of patients requiring dialysis) and in the long-term (more than half of the patients reaching ESRD), and a frequency (56%) and distribution of complement gene variants similar to those of atypical HUS cases from the same three national registries (6,7,13). Pregnancy-associated HUS is thus an atypical HUS triggered by pregnancy, in keeping with the findings of a previous small study performed in one country (3). It remains unknown why a significant proportion of women presented with HUS in the second or subsequent pregnancies and not in earlier pregnancies, and overall the precise mechanism by which pregnancy precipitates HUS remains ill-defined. During normal pregnancy, complement activation occurs in the placenta at the interface between the mother and the fetus (14) and one would expect pregnancy-associated HUS to occur mainly during pregnancy. Nevertheless, complement alternative pathway regulation in the placenta depends predominantly on CD59 and Decay Accelerating Factor (15,16), two membrane-bound proteins that negatively control the complement alternative pathway. Neither CD59 nor Decay Accelerating Factor has been to date implicated in the pathogenesis of atypical HUS, and their overexpression in the placenta may compensate for the deficiency in other regulatory proteins implicated in the pathogenesis of atypical HUS. In contrast, during the postpartum period, the protection conferred by the overexpression of Decay Accelerating Factor and CD59 is lost with placental delivery. Postpartum bleeding and infection may also trigger excessive complement activation in predisposed women and ultimately lead to HUS.

Three-quarters of the cases included in this study occurred in the postpartum period of uneventful pregnancies, and in this setting the diagnosis of HUS is rather straightforward. For the remaining one-quarter of cases occurring during pregnancy (mainly during the third trimester) the differential diagnosis is more complex, because several disorders may mimic HUS, including preeclampsia, HELLP syndrome, or severe bleeding (4). We have carefully excluded patients with these complications of pregnancy from the study. Besides, one-quarter of patients with HUS during pregnancy had a history of CKD/HUS, and the existence of renal vascular damage before gestation may explain the occurrence of HUS early during pregnancy and before delivery. Furthermore, patients with HUS during pregnancy and those with HUS in the postpartum period had a similar high frequency (50.0%–59%) of complement gene variants. In contrast, only a minority (8%–10%) of patients with preeclampsia and HELLP syndrome harbor complement gene variants, mostly of unknown significance or nonpathogenic (17). Finally, patients with HUS during pregnancy and those with HUS in the postpartum period shared the same severe renal outcome (risk of ESRD of 44%–55%), in sharp contrast to the usual complete recovery of kidney function in patients with preeclampsia and HELLP syndrome. Thus, the present cohort reflects the presentation, genetic risk factors, and outcome of HUS occurring during pregnancy or in the postpartum period.

Two-thirds of complement gene variants detected in patients with pregnancy-associated HUS were considered to be pathogenic, on the basis of quantitative defects and proven or predicted functional abnormalities of the encoded protein. These findings are similar to those of complement gene variants detected in atypical HUS (12). The presence of an identified complement gene variant was associated with a more severe outcome of pregnancy-associated HUS, with an increased risk of HUS relapse and of progression to ESRD. However, there was no difference in terms of presentation and outcome between patients with different types of complement gene variants.

This study raises several issues regarding the management of pregnancy-associated HUS. Firstly, plasma exchanges did not improve the renal outcome of pregnancy-associated HUS, and the risk of ESRD remained similarly high (around 50%) in patients who underwent plasma exchanges and in those who did not. Secondly, the availability of the anti-C5 antibody eculizumab has dramatically changed the treatment and outcome of atypical HUS (18,19). However, pregnant women were excluded from prospective trials with eculizumab and no specific data were reported for the few patients with postpartum HUS. In the present series, only four patients with pregnancy-associated HUS received eculizumab, and a limited number of cases reports of the use of the drug in this setting are available in the literature (20–23). In all of these cases except one, eculizumab proved efficacious in controlling HUS and improving kidney function. On the basis of the experience with eculizumab in patients with paroxysmal nocturnal hemoglobinuria (24), the use of eculizumab during gestation seems safe, even though the drug was detected in one-third of cord blood samples but was not present in maternal breast milk. Careful monitoring of complement blockade is mandatory because pregnancy may require an increase in the dosage and/or the frequency of eculizumab infusions due to the increase in the distribution volume or C5 synthesis (25). On the basis of all of these available data, treatment of pregnancy-related HUS should be similar to the treatment of nonpregnancy-related atypical HUS and relies on eculizumab use—as first-line treatment if the clinical presentation (postpartum of an uneventful pregnancy) and/or personal or familial history is highly suggestive of HUS, or as second-line treatment after plasma exchange if a diagnostic workup is necessary. Nevertheless, a further specific assessment of eculizumab use in pregnant women with HUS is warranted.

Secondly, the present data will help clinicians counsel patients with a history of atypical HUS or healthy carriers of complement gene variants who wish to start a pregnancy. In our study the risk of HUS was highest during the first pregnancy, but remained relatively high during subsequent pregnancies and may even increase as eculizumab rescues patients from ESRD and thus preserves their ability to conduct a pregnancy. However, the availability of eculizumab has dramatically improved the outcome of atypical HUS, and nowadays most clinicians do not advise against pregnancy in a patient with a history of atypical HUS or a healthy carrier of complement gene variant. Patients should make their decision about pregnancy after being informed of the risk of atypical HUS relapse (estimated, however, based mainly on studies in atypical HUS nonrelated to pregnancies) but also of the risks inherent to potential clinical or subclinical chronic kidney damage due to previous episodes of atypical HUS, including hypertensive complications of pregnancy that may occur despite treatment with eculizumab (25). Pregnancy in these patients requires close collaborative monitoring by nephrologists and obstetricians in level 3 maternity hospitals, starting in the first weeks of pregnancy and up to 3 months after delivery, in order to rapidly detect and thus treat the early manifestations of HUS.

In conclusion, pregnancy-related atypical HUS is a very severe disease, with a poor prognosis for maternal kidney function if left without specific efficacious treatment. It is associated with variants in complement genes in 56% of patients, and pregnancy seems an important trigger for the disease in these at-risk patients. Prospective studies are needed in order to specifically assess the efficacy and safety of eculizumab in pregnancy-associated HUS.

Disclosures

D.K., Y.D., F.P., C.L., and V.F.-B. received honoraria for consultancy from Alexion Pharmaceuticals. D.K. is scientific advisor and board member of Gyroscope Therapeutics. M.N. has received honoraria from Alexion Pharmaceuticals for giving lectures and participating in advisory boards, and research grants from Omeros and Chemocentrix. None of these activities has had any influence on the results or interpretation in this article. G.R. has consultancy agreements with AbbVie, Alexion Pharmaceuticals, Bayer Healthcare, Reata Pharmaceuticals, Novartis Pharma, AstraZeneca, Otsuka Pharmaceutical Europe, and Concert Pharmaceuticals (please note that no personal remuneration is accepted, compensations are paid to his institution for research and educational activities). The other authors declare no conflict of interest.