Hirschsprung disease (HSCR, aganglionic megacolon) represents the main genetic cause of functional intestinal obstruction with an incidence of 1/5000 live births. This developmental disorder is a neurocristopathy and is characterised by the absence of the enteric ganglia along a variable length of the intestine. In the last decades, the development of surgical approaches has importantly decreased mortality and morbidity which allowed the emergence of familial cases. Isolated HSCR appears to be a non-Mendelian malformation with low, sex-dependent penetrance, and variable expression according to the length of the aganglionic segment. While all Mendelian modes of inheritance have been described in syndromic HSCR, isolated HSCR stands as a model for genetic disorders with complex patterns of inheritance. The tyrosine kinase receptor RET is the major gene with both rare coding sequence mutations and/or a frequent variant located in an enhancer element predisposing to the disease. Hitherto, 10 genes and five loci have been found to be involved in HSCR development.
Statistics from Altmetric.com
Harald Hirschsprung, a Danish paediatrician, first described in 1888 two unrelated boys who died from chronic severe constipation with abdominal distension resulting in congenital megacolon.1 The absence of intramural ganglion cells of the myenteric and submucosal plexuses (Auerbach and Meissner plexuses, respectively) downstream of the dilated part of the colon was recognised as the cause of the disease in the 1940s.2 This allowed a simple and reliable diagnostic confirmation from rectal suction biopsies using histochemical staining for acetylcholinesterase (AchE).3 In 1948, Swenson and Bill developed a surgical procedure4 and the survival of patients uncovered familial transmission of Hirschsprung disease (HSCR).5 In 1974, Bolande proposed the term neurocristopathy for syndromes or tumours involving the neural crest (NC) cells.6 HSCR resulting from an anomaly of the enteric nervous system (ENS) of NC origin is therefore regarded as a neurocristopathy.6–8
Isolated HSCR appears to be of complex, non-Mendelian inheritance with low, sex-dependent penetrance, variable expression according to the length of the aganglionic segment and suggestive of the involvement of one or more gene(s) with low penetrance.5 9 These parameters must be taken into account for accurate evaluation of recurrence risk in relatives. With a relative risk as high as 200, HSCR appears an excellent model to study common multifactorial diseases. The major susceptibility gene is RET, which is also involved in multiple endocrine neoplasia type 2 (MEN 2) and familial medullary thyroid carcinoma (FMTC). Coding sequence mutations are identified in about 50% and 15% of familial and sporadic HSCR cases, respectively. The far most frequent HSCR predisposing event at the RET locus is a haplotype which comprises an SNP lying in an enhancer element of RET intron 1. The identification of modifier genes is currently underway by using various approaches and an international consortium has been settled in 2004 in order to achieve this goal.
HSCR occurs as an isolated trait in 70% of patients, is associated with a chromosomal abnormality in 12% of the cases, and with additional congenital anomalies in 18% of the cases.10–15 In the latter group of patients, some monogenic syndromes can be recognised. Indeed, thus far, genetic heterogeneity in HSCR has been demonstrated with 10 specific genes involved. The aim of this paper is to update a 6 year old review on clinical and molecular data about isolated and syndromic HSCR.
DEFINITION AND CLASSIFICATION
HSCR is a congenital malformation of the hindgut characterised by the absence of parasympathetic intrinsic ganglion cells in the submucosal and myenteric plexuses.2 It is regarded as the consequence of the premature arrest of the craniocaudal migration of vagal neural crest cells in the hindgut between the fifth and 12th week of gestation to form the enteric nervous system (ENS) and is therefore regarded as a neurocristopathy.6 16 While the internal anal sphincter is the constant inferior limit, patients could be classified as short-segment HSCR (S-HSCR: 80% of cases) when the aganglionic segment does not extend beyond the upper sigmoid, and long-segment HSCR (L-HSCR: 20% of cases) when aganglionosis extends proximal to the sigmoid. Four HSCR variants have been reported: (1) total colonic aganglionosis (TCA, 3–8% of cases)17; (2) total intestinal HSCR when the whole bowel is involved17; (3) ultra-short segment HSCR involving the distal rectum below the pelvic floor and the anus18; (4) suspended HSCR, a controversial condition, where a portion of the colon is aganglionic above a normal distal segment.
CLINICAL FEATURES AND DIAGNOSIS
In most cases, the diagnosis of HSCR is made in the newborn period15 due to intestinal obstruction with the following features: (1) delayed of passage of meconium (>24 h after birth); (2) abdominal distension that is relieved by rectal stimulation or enemas; (3) vomiting; and (4) neonatal enterocolitis. Some patients are diagnosed later in infancy or in adulthood with severe constipation, chronic abdominal distension, vomiting, and failure to thrive.19 Finally, although a rare presentation, unexplained perforation of the caecum or appendix should make the diagnosis considered.
On abdominal x ray a distended small bowel and proximal colon, with absence of rectal gas, are common findings. The classical image is a dilated proximal colon with the aganglionic cone narrowing towards the distal gut. On barium enema a small rectum with uncoordinated contractions is seen. The transitional zone represents the site where the narrow aganglionic bowel joins the dilated ganglionic bowel. On a delayed plain x ray taken 24 h after the enema, barium retention is observed. Anorectal manometry shows absence of relaxation of the internal sphincter (rectal inhibitory reflex) in response to rectal distension.20 The reliability of this test becomes excellent from day 12 after birth where the normal rectoenteric reflex is present.21 Suction rectal biopsy remains the gold standard for confirming the diagnosis in most cases demonstrating an increased acetyl cholinesterase activity.22 Nonetheless, full thickness rectal biopsy is the golden standard in reaching the diagnosis. Furthermore, seromuscular biopsies will be needed at operation to define the proximal limit of the aganglionic segment.
Other causes of intestinal obstruction should be discussed when abdominal distension and failure to pass meconium occur in a newborn infant: (1) meconium ileus resulting from cystic fibrosis; (2) intestinal malformations such as lower ileal and colonic atresia, isolated or occasionally associated with HSCR, intestinal malrotation or duplication; (3) ENS anomalies grouped as chronic intestinal pseudo-obstruction syndromes; and (4) functional intestinal obstruction resulting from maternal infection, maternal intoxication or congenital hypothyroidism.
TREATMENT AND PROGNOSIS
The treatment of HSCR is surgical. After careful preoperative management, the underlying principle is to place the normal bowel at the anus and to release the tonic contraction of the internal anal sphincter. Since the initial protocol of Swenson described in 1948,4 a series of operative approaches, such as the Soave and Duhamel procedures, have been developed.23 24 A one stage procedure is possible when diagnosis is made early, before colonic dilatation, in short segment disease. Otherwise, a primary colostomy is required. For long segment disease and total colonic aganglionosis, temporary enterostomy is often the first step in management before definitive surgery. Laparoscopic and transanal pull-through techniques have been proposed more recently in HSCR surgery.25 These techniques can provide patients with almost scarless surgery. Comparative long term results are pending.26 27 Neuronal precursor cells isolated from the developing human ENS may open the route to cell therapy.28 29 Fistula or stenosis of the anastomosis and enterocolitis are the main short term complications.30 Long term complications include chronic constipation (10–15%) and soiling.31 32 Mortality has been below 6% since the 1980s and may be related to short term complications or caused by the associated malformations.31 However, the treatment of children with TCA remains hazardous.33 34
The incidence of HSCR is estimated at 1/5000 live births.5 However, the incidence varies significantly among ethnic groups (1.0, 1.5, 2.1, and 2.8 per 10 000 live births in Hispanics, Caucasian-Americans, African-Americans, and Asians, respectively).15 S-HSCR is far more frequent than L-HSCR (80% and 20%, respectively).10 12 There is a sex bias with a preponderance of affected males and a sex ratio of 4/1.35 Interestingly, the male:female ratio is significantly higher for S-HSCR (4.2–4.4) than for L-HSCR (1.2–1.9) (table 1).15 35
MOLECULAR GENETICS IN ISOLATED HSCR
Several genes have been implicated in isolated HSCR, the two major ones being RET and EDNRB.
The RET signalling pathway
The first susceptibility locus was mapped to 10q11.2 in multigenerational families segregating HSCR as an incompletely penetrant autosomal dominant trait.36 37 This region had been targeted because of the observation of an interstitial deletion of chromosome 10q11 in patients with TCA and mental retardation.38 The proto-oncogene RET (REarranged during Transfection), identified as disease causing in MEN 239 40 and mapping in 10q11.2, was regarded as a candidate gene owing to: (1) co-occurrence of MEN 2A and HSCR in some families; and (2) expression in neural-crest derived cells. Consequently, RET gene mutations were identified in HSCR patients (fig 1).41 42 Over 100 mutations have been identified including large deletions encompassing the RET gene, microdeletions and insertions, nonsense, missense and splicing mutations.43–46 There is no mutational hot spot at variance with MEN 2A, where mutations occur in a cluster of six cysteines (exon 10: residues 609, 611, 618, 620; exon 11: residues 630,634),39 40 47 and MEN 2B where the mutation is almost unique (M918T, exon 16, tyrosine kinase domain).48–51 In vitro, MEN 2 mutations have been shown to be activating mutations leading to constitutive dimerisation of the receptor and to transformation,52 while haploinsufficiency is the most likely mechanism for HSCR mutations.53–57 Biochemical studies demonstrated variable consequences of some HSCR mutations (misfolding, failure to transport the protein to the cell surface, abolished biological activity).54 56 58 However, a simple activating versus inactivating model of gene action is not sufficient to explain the co-occurrence of HSCR and MEN 2A in patients with a MEN 2A RET gene mutation.51 59
Despite extensive mutation screening, a RET mutation is identified in only 50% of familial and 15–20% of sporadic HSCR cases.43 44 60 61 However, most families with few exceptions are compatible with linkage at the RET locus.62 Case–control and transmission disequilibrium test in several ethnic backgrounds had first pointed to a frequent SNP lying in exon 2 and leading to a silent change as over represented and transmitted in patients (fig 1).63–68 Later, the same observation was made for haplotypes comprising this SNP lying in exon 2 and an SNP at–5 from the transcription start site of RET.69–72 As there was no convincing evidence for a functional role of these two SNPs,71 73 74 the most likely hypothesis was that an ancient, low-penetrant founder locus was in linkage disequilibrium with the haplotype of the two SNPs previously identified and distant of about 25 kb.70 Comparative genomics focused on conserved non-coding sequences and an SNP lying in intron 1 was shown associated to HSCR susceptibility, making a 20-fold greater contribution to risk than coding sequence mutations.75 This T>C SNP lies in an enhancer-like sequence and the T allele reduces in vitro enhancer activity.75 Moreover, this sequence drives reporter expression in tissue consistent with the one of Ret during mouse and zebra fish development.75 76 Interestingly, the frequency of the predisposing T allele varies according to HSCR prevalence in various ethnic backgrounds from about 20–50% in European and Chinese, respectively.75 77 The T allele high frequency in control populations emphasises, as speculated by the oligogenic model, the pivotal role of the RET gene in HSCR susceptibility despite low penetrance. Finally, the penetrance of the T allele for the HSCR trait is both dose-dependent and greater in males than in females.75 Conversely, an SNP lying in the 3′ UTR of the RET gene and lowering stability to RET mRNA degradation has been shown to be under transmitted in HSCR cases.78 Again, this SNP lies on a haplotype that is of variable frequency according to ethnicity (about 8–4% in Caucasian and absent in Chinese).71 79 80 A recombination spot lies on intron 5 at the RET locus.59 66
RET is a 1114 amino acid transmembrane receptor with a cadherin-like extracellular domain, a cysteine-rich region and a intracellular tyrosine kinase domain.81 The role of Ret in mice development has been expanded to kidney,82–84 spermatogenesis85–88 and Peyer’s patch.89 90 Between the two RET major isoforms (RET9 and RET51) with different C-terminal tails as the result of alternative splicing, RET9 is critical for both kidney and ENS development.91
GDNF, known as a major survival factor for many types of neurons, was shown to be the RET ligand by both phenotypic similarities between Ret −/− and Gdnf −/− knock-out mice,92–94 and xenopus embryo bioassays.95 GDNF is a TGF-B related 211 residue protein, proteolytically cleaved to a 134 residue mature protein that homodimerise. To activate RET, GDNF needs the presence of a glycosylphosphatidylinositol (GPI)-linked co-receptor GFRA1.96 97 Four related GPI-linked co-receptors, GFRA1-4,98 and four related soluble growth factor ligands of RET have been identified, namely: GDNF, NTN,99 persephin (PSPN)100 and artemin (ARTN).101 Specific combinations of these proteins are necessary for the development and maintenance of both central and peripheral neurons, and all can signal through RET. GDNF mutations have been identified in only six HSCR patients to date, and could be regarded as a rare cause of HSCR (<5%).102–104 Moreover, GDNF mutations may not be sufficient to lead to HSCR since 4/6 patients have additional contributory factors, such as RET mutations or trisomy 21.102 103 Similarly, an NTN mutation has been identified in one family, in conjunction with a RET mutation.105 Finally, although Gfra1 homozygous knock-out mice are phenotypically very similar to Ret and Gdnf −/− mice, no GFRA1 mutations have been identified in HSCR patients except a deletion at the locus with incomplete penetrance in one family.69 106–109 Worth noting, RET exerts a pro-apoptotic effect that is inhibited by GDNF and some RET gene mutations may impair the control of this activity by GDNF.110
The endothelin signalling pathway
The endothelin pathway was first studied for its vasoconstrictive effect and putative role in hypertension. EDNRB and EDNRA are G-protein-coupled heptahelical receptors that transduce signals through the endothelins (EDN1, 2, 3).111 112
A susceptibility locus for HSCR in 13q22 was pointed out for three main reasons: (1) a significant lod-score at 13q22 in a large inbred Old Order Mennonite community with multiple cases of HSCR113–115; (2) de novo interstitial deletion of 13q22 in several patients with HSCR116; (3) synteny between the murine locus for piebald-lethal (sl), a model of aganglionosis, and 13q22 in human. The critical role of the endothelin pathway in HSCR was demonstrated with the finding that piebald-lethal was allelic to the Ednrb knock-out mouse and harboured an Ednrb mutation (table 2).117 Subsequently, an EDNRB missense mutation was identified in the Mennonite kindred (W276C).118 However, the W276C mutation was neither necessary (affected wild-type homozygotes) nor sufficient (non-affected mutant homozygotes) to cause HSCR, and penetrance was sex-dependent (greater in males than in females).118 piebald-lethal was considered a mouse model for WS4 in humans, and some of the Mennonite affected individuals had pigmentary anomalies and sensorineural deafness in addition to HSCR.113 114 This prompted a screen of the EDNRB gene in WS4, and homozygous mutations in a fraction of WS4 families were found.44 At the same time, an Edn3 mutation was identified in the lethal spotting (sl) natural mouse model for WS4119 and, subsequently, EDN3 homozygote mutations were identified in WS4 in humans (table 2).120 121
Both EDNRB and EDN3 were screened in large series of isolated HSCR patients. While EDN3 mutations were seldom found,130 EDNRB mutations were identified in approximately 5% of the patients.126–129 It is worth mentioning that the penetrance of EDN3 and EDNRB heterozygous mutations is incomplete in those HSCR patients, de novo mutations have not hitherto been observed, and that S-HSCR is largely predominant. Interstitial 13q22 deletions encompassing the EDNRB gene in HSCR patients makes haploinsufficiency the most likely mechanism for HSCR (table 3). Although EDNRB binds all three endothelins, the similarity of phenotype of the Ednrb knock-out mice to that of the Edn3 knock-out mice suggests that EDNRB’s major ligand is EDN3.
Preproendothelins are proteolytically cleaved by two related membrane-bound metalloproteases to give rise to the mature 21-residue endothelin. Ece1 processes only Edn1 and Edn3. Ece1 knock-out mice show craniofacial defects and cardiac abnormalities in addition to colonic aganglionosis.132 A heterozygous ECE1 mutation has been identified in a single patient combining HSCR, craniofacial and cardiac defects (R742C).131
The last known mouse model for WS4 in human is dominant megalon (Dom), homozygous Dom mutation being embryonic lethal.152 The Dom gene is Sox10, a member of the SRY (sex determining factor)-like, high mobility group (HMG) DNA binding proteins.125 Subsequently, truncating heterozygote SOX10 mutations have been identified in patients with WS4,122–124 Yemenite deaf-blind-hypopigmentation syndrome153 and WS2 (Bondurand et al in Am J Hum Genet website) but also in patients presenting in addition neurological impairment due to central and peripheral dysmyelination.67 123 The latter combination is known as PCWH for Peripheral demyelination-Central dysmyelinating leucodystrophy-Waardenburg syndrome and Hirschsprung disease. Genotype–phenotype correlation relies on nonsense-mediated decay being effective (WS4) or not (PCWH).154 The penetrance of the HSCR trait appears to be high, although sibs sharing a mutation and discordant for HSCR have been described in one family.124 Therefore, SOX10 is unlikely to be a major gene in isolated HSCR.
Interaction between pathways
Ret and Ednrb signalling pathways were considered biochemically independent. However, G-protein-coupled receptors and tyrosine kinase receptors could be engaged in crosstalk. Moreover an HSCR patient heterozygote for weak hypomorphic mutations in both RET and EDNRB has recently been reported.155 Each mutation was inherited from a healthy parent. Genetic interactions between EDNRB and RET have been demonstrated in the Mennonite population where HSCR predisposition is high (incidence of 1/500).118 156 Finally, no complementation of aganglionosis could be observed in mouse inter crosses between hypomorphic piebald alleles of Ednrb (Ednrbs/s) and a null allele of Ret.156
Sox10 is involved in cell lineage determination and is capable of transactivating MITF synergistically with PAX3.157 Similarly, Ednrb transcripts are either absent or drastically reduced in Dom−/− and Dom+/− mice, respectively.158 Sox10;Ednrb and Sox10;Edn3 double mutants have a severe ENS defect with no enteric progenitor cells extending beyond the stomach at all embryonic stages studied.159 Interestingly, genetic interactions for the HSCR trait have been shown between RET and PHOX2B and BBS genes responsible when mutated for CCHS and Bardet-Biedl syndromes, respectively.148 Such correlation was not found between RET and SOX10. Along these lines, a genome-wide screen aimed at localising modifier genes for the aganglionosis of Dom mice did not point to the Ret locus but, among others, a locus encompassing the PHOX2b gene.160
Taking it all together, several general comments can be made: (1) RET is the major gene in HSCR with either CDS mutations or, more frequently, a low penetrant SNP lying in an enhancer element within intron 1; (2) RET mutation penetrance is incomplete and sex-dependant; (3) genotype–phenotype correlation is poor; 4) HSCR is genetically heterogeneous and due to mutations in distinct pathways; (5) some patients with mutations in more than one HSCR susceptibility gene are known (RET+GDNF, RET+NTN, RET+EDNRB); (6) the RET gene plays a role in HSCR penetrance of some but not all syndromic HSCR (see below).
MULTIGENIC INHERITANCE OF ISOLATED HIRSCHSPRUNG DISEASE
As mentioned above, RET plays a key role in HSCR genesis and multiple genes may be required to modulate clinical expression. On the other hand, genetic heterogeneity where mutation in one of several genes is sufficient for phenotypic expression of HSCR has been demonstrated (RET, EDNRB, EDN3, ECE1). Segregation studies in HSCR showed that the recurrence risk in siblings varies from 1.5–33% depending on the gender and the length of the aganglionic segment in the proband, and the gender of the sibling (table 1).5 35 Consequently, HSCR has been assumed to be a sex modified multifactorial disorder, the effect of genes playing a major role as compared to environmental factors (relative risk of 200).
According to the segregation analysis where an autosomal dominant model in L-HSCR and a multifactorial model in S-HSCR were more likely, different approaches have been chosen to test these hypotheses in L-HSCR and S-HSCR independently.
Linkage analysis in 12 HSCR families with three or more affected individuals in two or more generations where L-HSCR is largely predominant.62 All but one family showed linkage to the RET locus. Mutational analysis identified a nonsense or missense mutation at highly conserved residue in six families, a splice mutation in two families and no coding sequence variation in three families. Linkage to a novel locus in 9q31 was identified only in families with no or hypomorphic RET gene mutation. Therefore, a severe RET mutation may lead to phenotypic expression by haploinsufficiency while hypomorphic RET mutations would require the action of other mutations.
A sib-pair analysis in 49 families with S-HSCR probands.161 This studies shows that only three loci on chromosomes 3p21, 10q11 and 19q12 are both necessary and sufficient to explain the incidence and sibling recurrence risk in HSCR. A multiplicative risk across loci with most affected individuals being heterozygotes at all three loci seems the best genetic model. Finally, linkage to 9q31 was confirmed in the sib-pairs with no or hypomorphic RET mutation.
A genome-wide association study was conducted in 43 Mennonite family trios and identified a susceptibility locus on 16q23 in addition to the loci of the two predisposing genes in this population (RET and EDNRB at 10q11.2 and 13q22, respectively).156
Linkage analysis in a multigenerational HSCR family where the RET gene had been previously excluded, showed linkage to 4q31-q32.162
The route to the identification of modifier genes is now based on various approaches. A differential screen for ENS expressed genes was conducted by a 22 000 probe DNA micro array of embryonic Ret+/+ and Ret−/− mice and identified over 300 genes over expressed in Ret+/+ mice.