Hereditary atypical hemolytic uremic syndrome (aHUS), a dramatic disease frequently leading to dialysis, is associated with germline mutations of the CFH, CD46, or CFI genes. After identification of the mutation in an affected aHUS patient, single-site gene testing of relatives is the preventive care perspective. However, clinical data for family counselling are scarce.
From the German-Speaking-Countries-aHUS-Registry, 33 index patients with mutations were approached for permission to offer relatives screening for their family-specific mutations and to obtain demographic and clinical data. Mutation screening was performed using direct sequencing. Age-adjusted penetrance of aHUS was calculated for each gene in index cases and in mutation-positive relatives.
Sixty-one relatives comprising 41 parents and 20 other relatives were enrolled and mutations detected in 31/61. In total, 40 research participants had germline mutations in CFH, 19 in CD46 and in 6 CFI. Penetrance at age 40 was markedly reduced in mutation-positive relatives compared to index patients overall with 10% versus 67% (P < 0.001); 6% vs. 67% (P < 0.001) in CFH mutation carriers and 21% vs. 70% (P= 0.003) in CD46 mutation carriers.
Age-adjusted penetrance for hereditary aHUS is important to understand the disease, and if replicated in the future, for genetic counselling.
Atypical Hemolytic Uremic Syndrome (aHUS) is characterized by acute renal failure, nonimmune-mediated hemolytic anemia, and low platelet count. Laboratory test results include elevation of serum creatinine and lactate dehydrogenase, decreased haptoglobin and hemoglobin as well as schistocytes in the blood smear, and negative Coombs test. Without adequate treatment, there is a high risk for end-stage renal failure necessitating life-long renal replacement therapy. Kidney allograft transplantation is characterized by a high rate of early deterioration of function due to relapse of HUS. The histological correlate of aHUS in renal biopsy is thrombotic microangiopathy. aHUS is the counterpart of classical HUS. Classical HUS is characterized by diarrhea induced from infections by shiga-like toxin-producing bacteria such as E. coli strain 0157:H7. Classical HUS occurs mainly in infants with intestinal infections causing >90% of HUS cases (Liu et al., 2001).
aHUS has been known to be familial for several decades. Molecular genetic advances have led to the identification of distinct genes underlying the etiology. The modern classification of aHUS is therefore based on results of molecular genetic research. To date, at least nine susceptibility genes, with three major ones, have been identified for aHUS. Complement factor H (CFH) was first reported in 1998 (Warwicker et al., 1998) followed by CD46 (also called MCP) in 2003 (Noris et al., 2003). In 2004, complement factor I (CFI) was recognized as a predisposition gene for aHUS as well (Fremeaux-Bacchi et al., 2004). CFH, CD46, and CFI are currently considered the three major predisposition genes. More recently, germline mutations in C3 were described in a small number of aHUS patients (Fremeaux-Bacchi et al., 2008). Other predisposition genes include CFHR1, 3, 4 and thrombomodullin (THBD) (Delvaeye et al., 2009; Moore et al., 2010) as well as complement factor B (CFB) (Dragon-Durey et al., 2005).
The identification of germline mutations in the susceptibility genes has important scientific and clinical relevance. First, it provides the basic information for the cause of the disease. All the susceptibility genes encode products within or interacting with the alternative complement activation pathway (Zipfel et al., 2010). Second, the identification of predisposing genes also makes molecular diagnosis possible for the patient. Finally, it allows for genotype-based predictive testing for at-risk relatives. The latter may be a powerful risk assessment and management tool for as yet unaffected relatives. Family counselling has become an important part of clinical management and care in aHUS. The dominant question is the risk for manifestation of the disease in relatives. Such available information remains scarce. Overall, the penetrance of aHUS in probands with germline mutations of susceptibility genes is regarded as about 50% (Kavanagh and Goodship, 2010). Documentation of detailed and age-related penetrance for any of the susceptibility genes is absent in the literature, to date. Further and more importantly, there are no studies available on penetrance of aHUS as age-related estimations. As with most hereditary disease, there exists age-related penetrance in mutation positive probands. These data, although correctly reported, may not represent an adequate basis for family counselling because these studies are based on mainly probands and mixed probands and relatives.
Historically, age-related penetrance figures for component phenotypic features of genetic disorders were assumed to be similar between mutation positive probands and their mutation positive relatives. However, it is important to recognize the potential differences in clinical manifestations and the first-to-present phenomenon in index cases compared to their relatives. While aHUS carries 100% penetrance based on index case studies, the penetrance in relatives has not been systematically studied. To address our hypothesis that penetrance between probands and relatives may be different, we studied mutation positive probands (index cases) presenting with aHUS and members of their respective families found to have germline mutations in the respective susceptibility genes.
Patients and Methods
The Registry for aHUS in German-Speaking Countries established in 1998 was used for the basis of this study (Sullivan et al., 2010). We recontacted all patients who were identified with a germline mutation of any aHUS susceptibility gene and asked if their relatives were willing to participate. The cooperating nephrology centers performed measurements of serum creatinine at the time of registration. Participants with abnormal renal function or previous episodes suggestive of HUS were asked for medical records documenting these episodes.
Molecular Genetic Analyses
All index cases were scanned for germline mutations in all three known major aHUS predisposition genes CFH (RefSeqGene NG_007259.1), CD46 (RefSeqGene NG_009296.1), and CFI (RefSeqGene NG_007569.1) using PCR amplification and direct sequencing of all exons (Sullivan et al., 2010). All relatives were analyzed for the mutations that had been identified in the respective index patient using sequencing of the exon in which the mutation of the index cases was located (family-specific mutation).
DNA variants predicting truncated proteins were regarded as pathogenic. Missense DNA variants were classified as pathogenic if not found in 100 normal controls (200 chromosomes), who derive from healthy anonymized blood donors to the blood bank serving the University Medical Center in Freiburg. Missense variants were also checked by in silico analyses using the SIFT, PolyPhen, SNAP, and MutationTaster prediction based on conservation programs.
All index cases with germline mutation and all relatives found to carry the family-specific mutation were included for evaluation of penetrance. Age-related penetrance was estimated using the Kaplan–Meier method and compared between index cases and relatives using the log-rank test. Penetrance was also compared between research participants with truncating mutations and those with missense variants using the log-rank test.
The study was approved by the Ethics Committee of the University Medical Center of the Albert-Ludwigs-University of Freiburg, Germany. All participants provided written consent.
The index-patient/proband cohort comprised 33 unrelated index patients (probands) with aHUS in whom germline mutations have been identified. Sixty-one relatives of 27 of the 33 probands (members of 27 families) agreed to be screened for their respective family-specific mutations. The relative cohort comprised 41 parents (16 fathers and 25 mothers), 15 siblings, one child, and four second degree relatives.
Germline Mutations in Index Cases
The 33 index patients from the Registry harboured germline mutations in the CFH, CD46, and CFI genes (Table 1). There were 20 different mutations in the CFH gene, 15 missense and five truncating, nine different mutations in the CD46 gene, four missense and five truncating, and three different mutations of the CFI gene, all missense. None of these mutations occurred in 100 controls.
Table 1. Demographic and genotype data of 33 probands with atypical hemolytic uremic syndrome
Age at HUS Onset in Index Patients
Age at HUS Onset in Relatives
Patients with HUS in the Family
Mutation positive Relatives/Tested Relatives
FH aHUS Mutation Database Mutation Type
+ Pathogenicity SIFT: deleterious; SNAP: non-neutral; PolyPhen: probably damaging with high confidence supposed to affect protein function or structure; MutationTaster: disease causing–Likely not pathogenic: SIFT: tolerated; SNAP: neutral; PolyPhen: benign; MutationTaster: polymorphism ±PolyPhen: possibly damaging supposed to affect protein function or structure Type I indicates that the mutant protein is either absent from the plasma or present in lower amounts. This indicates the mutation has a structural effect on the mutant protein–that is, reducing the stability Type II indicates that the mutant protein is present in normal amounts in plasma. This indicates that the mutation has a functional effect on the protein – that is, affecting substrate binding *mutation with unknown effect on resultant protein.
Of the 33 index patients, 30 had one heterozygous mutation, 19 in CFH, eight in CD46, and three in CFI. Three index patients had double mutations. One index patient had a homozygous CFH mutation. Another index patient had a compound heterozygous CFH mutation. A third index patient had a compound heterozygous mutation in the CFH and the CD46 genes. All CFH mutations except five were missense mutations; additionally, two stop codon mutations, one splice site mutation, one deletion of three nucleotides, and one deletion of four nucleotides were identified. All CD46 mutations except three were missense mutations; the remaining three were one splice site mutation and two different single nucleotide frameshift mutations. All CFI mutations were missense mutations.
Results of Mutation Screening for Respective Family-Specific Mutations in Relatives
Of the 61 relatives screened for mutations, 31 (51%) had germline mutations. These were present in 21 of 41 parents, seven of 15 siblings, zero of one child, and three of four 2nd degree relatives.
Both parents of the index patient with the homozygous CFH mutation were found to carry the same mutation except as heterozygous. Of the index patient with a compound heterozygous CFH mutation, one parent carried one allele, and the other parent, the other. Of the index case with the compound heterozygous CFH and CD46 mutations, each parent showed one of these mutations.
Demographic and Clinical Findings
All index patients and relatives were caucasian. There were 21 index patients with CFH mutations (including the one with a CFH and a CD46 mutation), comprising 11 males and 10 females. Age at diagnosis of aHUS was 6–65 (mean 34) years. There were 10 index patients with CD46 mutations (including the one with a CD46 and a CFH mutation), six males and four females, with an age at diagnosis of aHUS of 19–56 (mean 37) years. Three index patients had CFI mutations, all were female, and age at diagnosis of aHUS was 20, 39, and 43 years.
Episodes of aHUS occurred in six mutation positive relatives, but not in mutation negative relatives (6/31 vs. 0/30, P= 0.01). These episodes occurred prior to this genetic study in five relatives and in one relative the episode occurred after recognition of the carrier status of the index patient. Further, of 23 mutation positive relatives of the parental and grandparental generation, only two had episodes of HUS.
Serum creatinine values were available for all relatives who underwent genetic screening. The estimated glomerular filtration rate (eGFR) was calculated. All these eGRF data—except those of the described relatives with episodes of aHUS—were within normal range, that is, above 90 ml/min/1.73 m2.
Penetrance of aHUS in Probands and Relatives Carrying Mutations
The criteria for being affected by aHUS were identical for the index patients and for the relatives: all had hemolytic nonimmunogenic anemia, low platelet count and acute renal failure. For purposes of penetrance calculations, we removed the three probands with compound heterozygous mutations and their relatives. Age-related penetrance for the aHUS phenotype was higher in the 30 index cases (probands) than in their 30 relatives who are mutation positive for any of CFH, CD46 and CFI genes (Fig. 1). The penetrance by age 40 was 67% and reached 100% by age 65 for the index patients (Fig. 1). In contrast, the penetrance at age 40 for the mutation positive relatives was 11% (P < 0.001). The maximum penetrance in the relatives was 15% by age 44.
For CFH mutation carriers, the penetrance by age 40 in 18 index cases was 67% and reached 100% at age 65 (Fig. 2). In the 19 mutation positive relatives, the penetrance at age 40 was only 6% (P < 0.001). The maximum penetrance in the relatives was 6% by age 37. For CD46 mutation carriers, the penetrance at age 40 in nine index cases was 67% and reached 100% by age 56 (Fig. 3). In nine mutation positive relatives, the penetrance at age 40 was 24%. The maximum penetrance in the relatives was 24% by age 31 (P= 0.009).
There were only three index patients and three relatives who were CFI mutation carriers so formal age-related penetrance could not be calculated. The index patients were 20, 39 and 43 years old at diagnosis of aHUS. One mutation positive relative had aHUS at age 44, and the other two mutation positive relatives did not manifest with aHUS until the ages of 79 and 80.
We also tested the possibility of a potential bias due to nonpathogenic DNA missense variations being favoured amongst the relatives. For this purpose, we utilized all missense variants regardless of in silico predictions of pathogenicity. The missense variants were plotted for age at occurrence of aHUS and compared to truncating mutations and age at first aHUS episode. There were no differences in age-related penetrance between those with mutations and those with truncating mutations (P= 0.93). No differences in penetrance between truncating mutations and missense variants were detected irrespective of gene involved.
Molecular genetics has opened important avenues for risk assessment, predictive testing and thus, preventive medicine. The basis for prevention begins with identification of disease-predisposing genes which allows for accurate molecular diagnosis of the patient, which then allows for future risk prediction, genetic counselling and hence, tailored management. Once a family-specific mutation is identified typically in an affected proband, predictive testing of the mutation can be offered to all at-risk family members which in turn allows subsetting of those at genetic risk and those who are not. This is an extremely powerful piece of knowledge that would allow for genotype-specific surveillance and even prevention. This type of gene-based risk assessment and management is best illustrated by heritable cancer syndromes (Eng, 2010). Such strategy for molecular-based personalization of preventive measures is particularly germane when the disease is severe and/or fulminant carrying high morbidity and/or mortality, such as appears to be the situation for aHUS, with end-stage renal disease being the inexorable endpoint of unrecognized aHUS. Furthermore, even transplanted kidneys have a high failure rate, succumbing to aHUS (Goodship, 2006). Molecular-targeted therapies such as the monoclonal antibody and complement inhibitor eculizumab are still in their infancy (Kose et al., 2010). It would be desirable as well if knowledge of the specific genotype would permit prediction for response to such therapies. For example, one study suggests that plasma therapy induced remissions in 55 to 80% of episodes in patients with CFH, C3, or THBD mutations or autoantibodies, whereas patients with CFI mutations were poor responders (Noris et al., 2010). A single case report shows a teenage male with aHUS and germline C3 mutation, who had failed both renal allografts, stabilize after eculizumab treatment (Al-Akash et al., 2011).
Historically, risk and penetrance were assumed to be identical between index cases presenting with disease and relatives who are mutation carriers. This assumption has come about because penetrance has usually been studied with proband data only or mixed proband and relative data. For aHUS, available published data provide an overall penetrance of 50% up to 60% (Caprioli et al., 2006; Kavanagh et al., 2008). Only one report provides details for the different susceptibility genes, for example, 48% penetrance for mutations in the CFH gene, 53% for mutations in the CD46 gene and 50% in the CFI gene (Caprioli et al., 2006). Potential differences in age-related penetrance in probands versus that in relatives were not addressed. Here, we have carefully examined age-related penetrance of aHUS in mutation positive presenting probands and in mutation positive family members. Predominantly, we have shown that mutation positive relatives systematically show dramatically decreased age-related penetrance compared to mutation positive presenting index cases, irrespective of gene involved. Our observations here should allow for evidence-based family counselling and risk-assessment after screening of relatives for family-specific germline mutations. While proband studies suggest that CFH mutations confer the most severe disease with earliest onset and highest mortality (Noris et al., 2010) and our proband penetrance estimates confirm this, our CFH mutation positive relatives show a low penetrance and onset after the age of 25. Although sample sizes are not very high, the difference in age related penetrance is statistically significantly different for probands versus relatives. These observations should encourage others to independently validate these data in larger cohorts.
More recently, the concept of the proband presenting with disease and the load of disease in the family representing the sum total of penetrance and expressivity of a gene mutation has resurfaced. For example, penetrance calculations of breast and ovarian cancers in families with germline BRCA1 and BRCA2 mutations has recently considered the concept of load of cancers in the family (Evans et al., 2008). Thus, the index presentation of aHUS from each family necessarily would carry the highest penetrance, as we note in our study, with much lower penetrance in carrier relatives. Indeed, the majority of our index cases’ mutation positive family members were parents and so, the fact that they had not presented with aHUS by the time of parenthood reflects the lowered penetrance.
The very low penetrance of aHUS in relatives found to carry the family-specific mutation does raise the question if the detected DNA variants are truly pathogenic. This is a difficult question to pose and even more difficult to resolve. Such a situation arose for rare germline variants believed, for a decade, to have been pathogenic for multiple endocrine neoplasia type 2 and von Hippel-Lindau disease (Erlic et al., 2010). Because of population-based registries and careful family studies, we were able to show that these variants were not primarily predisposing to these syndromes and represent either harmless polymorphisms or modifiers of phenotype but only in the presence of other variants/mutations. Providers and patients with newly detected mutations often view mutations on internet sites, such as http://www.fh-hus.org for FH, CD46, and FI mutations, and many are not well curated. Even the most curated sites such as dbSNP often list pathogenic mutations, proven by multiple criteria, as polymorphisms. In an attempt to be completely transparent so that others may judge for themselves, we have therefore included not only the undoubtedly pathogenic truncating mutations, but also the missense variants regardless if in silico analyses predict them as pathogenic or not. Notably, our results show that the penetrance between those with truncating mutations and those with missense mutations does not differ significantly. Additionally, we have removed the only three probands with compound heterozygous (double) mutations and their relatives (parents) who are heterozygotes. This is to remove bias from dosage effect in the penetrance calculations. Even with this conservative analysis, we show a statistically significant difference in penetrance between probands and mutation positive relatives.
In summary, we have shown that probands carrying germline mutations predisposing to aHUS have a much higher penetrance over their mutation positive relatives, regardless of gene and across all ages. The reasons behind the different penetrance values are currently unknown and necessitate further study. Some factors that may account for the different penetrance between mutation positive probands and mutation positive relatives might be genetic modifiers and unshared environmental exposures.
Hartmut PH Neumann is supported by a grant of the German Research Foundation (Deutsche Forschungsgemeinschaft Ne 571/4-6). Charis Eng is the Sondra J. and Stephen R. Hardis Endowed Chair of Cancer Genomic Medicine at the Cleveland Clinic, and an ACS Clinical Research Professor.