Molecular analysis of holoprosencephaly in South America.

Holoprosencephaly (HPE) is a spectrum of brain and facial malformations primarily reflecting genetic factors, such as chromosomal abnormalities and gene mutations. Here, we present a clinical and molecular analysis of 195 probands with HPE or microforms; approximately 72% of the patients were derived from the Latin American Collaborative Study of Congenital Malformations (ECLAMC), and 82% of the patients were newborns. Alobar HPE was the predominant brain defect in almost all facial defect categories, except for patients without oral cleft and median or lateral oral clefts. Ethmocephaly, cebocephaly, and premaxillary agenesis were primarily observed among female patients. Premaxillary agenesis occurred in six of the nine diabetic mothers. Recurrence of HPE or microform was approximately 19%. The frequency of microdeletions, detected using Multiplex Ligation-dependant Probe Amplification (MLPA) was 17% in patients with a normal karyotype. Cytogenetics or QF-PCR analyses revealed chromosomal anomalies in 27% of the probands. Mutational analyses in genes SHH, ZIC2, SIX3 and TGIF were performed in 119 patients, revealing eight mutations in SHH, two mutations in SIX3 and two mutations in ZIC2. Thus, a detailed clinical description of new HPE cases with identified genetic anomalies might establish genotypic and phenotypic correlations and contribute to the development of additional strategies for the analysis of new cases.

The phenotypic variability and incomplete penetrance make it difficult to conduct genetic counseling in HPE. Families typically present autosomal dominant, but also autosomal recessive and possibly X-linked inheritance (Ming and Muenke, 1998;Muenke and Beachy, 2001). Autosomal dominant inheritance has an estimated penetrance of 80% (Odent et al., 1998), and the same mutation segregating within a family can present asymptomatic, mild and severe forms of the disease El-Jaick et al., 2005). We recently analyzed HPE cases from the ECLAMC (Estudio Latinoamericano de Malformaciones Congénitas-Latin-American Collaborative Study of Congenital Malformations) since 2000 (Orioli et al., 2001;El-Jaick et al., 2005, El-Jaick et al., 2007a, and in the present study, we evaluate the contribution of chromosomal anomalies and mutations in the four main HPE genes (SHH, SIX3, TGIF, and ZIC2).

ECLAMC
Most of the families studied (140/195; 72%) were derived from the ECLAMC (Estudio Latinoamericano de Malformaciones Congénitas-Latin American Collaborative Study of Congenital Malformations). ECLAMC is a hospital-based registry, which examines births in South America (Castilla and Orioli, 2004). In 1998, the ECLAMC initiated a molecular epidemiological program, called MOLECLAMC, and biological samples from newborns presenting congenital malformations were collected for molecular analysis.
Samples from MOLECLAMC primarily comprise blood spots on filter paper (Schleicher and Schuell no. 903 or IsoCode), and a pediatrician interviewed the mothers of the malformed infants using a form with 50 questions concerning risk factors, including environmental exposure, history of other familial congenital defects, and parental consanguinity (Castilla and Orioli, 2004). The DNA repository of MOLECLAMC also receives samples from pediatricians and geneticists associated with the ECLAMC. These cases occasionally involve older patients, representing 28% (55/195) of the families presented herein.

Patients
The patients were referred based on typical HPE facial dysmorphisms, with or without confirmation through brain imaging, on SMMCI or a family history of HPE. Aneuploidy analyses were performed on some HPE patients with no informed karyotype. Chromosomal microrearrangements were evaluated in some HPE and SMMCI patients without mutations in the four main HPE genes and no known chromosomal anomalies.
The patients referred to the Birth Defects Laboratory at the Federal University of Rio de Janeiro were analyzed in the last 14 years. Initially, we only analyzed mutations in the four main HPE genes. In 2008, the laboratory implemented the MLPA (Multiplex Ligation-Dependent Probe Amplification) technique, and in 2010, we initiated studies to screen for trisomies using QF-PCR. Not all patients could be studied using all techniques, reflecting the lack of adequate DNA quantity and quality, particularly for older samples.
The Ethics Committee of the ECLAMC (CEMIC: Centro de Educación Médica e Investigación Clínica, Buenos Aires, Argentina, IORG-0001315) approved the project, and written informed consent was obtained from adult patients and the parents of infant patients.

DNA extraction
The DNA from blood spotted onto Schleicher and Schuell no. 903 filter paper (Schleicher & Schuell Inc., Keene, NH, USA) was extracted using QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA) or DNeasy 96 Tissue Kit (Qiagen, Valencia, CA, USA), and DNA from blood spotted onto IsoCode filter papers (Schleicher & Schuell Inc., Keene, NH, USA) was extracted through boiling in water. The DNA from peripheral blood samples on EDTA was extracted through salting out (Miller et al., 1988).

PCR and direct sequencing
The coding region and exon-intron boundaries of SHH, SIX3, TGIF, and ZIC2 were amplified as previously described (El-Jaick 2007a,b;Savastano et al., 2014). Automated sequencing was performed using MegaBACE 1000 (Amersham Pharmacia Biotech In, London, UK) or ABI Prism 377 (Applied Biosystems, Foster City, CA, USA) DNA sequencing systems. The electropherograms were analyzed using Sequencher software (Gene Codes Corporation, Ann Arbor, MI, USA). SIFT and PolyPhen2 online software programs were used to predict the effects of nonsynonimous mutations. The gene mutation nomenclature used in this study was consistent with the recommendations of den Dunnen and Antonarakis (2000), and updates are available on the website for the Human Genome Variation Society (HGVS).

QF-PCR analysis of trisomies
HPE patients without confirmed karyotypes were further tested for aneuploidies on chromosomes 13, 18, 21 and X through QF-PCR, using the Aneufast Multiplex QF-PCR Kit (Genomed AG, Wollerau, Switzerland). Patients with alterations in more than one marker for the same chromosome were subjected to further tests using chromosome specific back-up markers. The analyses were performed using GeneMarker (Softgenetics, State College, PA, USA).
The facial descriptions of the 195 patients, including the two obligated carrier mothers, were classified as cyclopia (19 cases), ethmocephaly (6 cases), cebocephaly (17 cases), and premaxillary agenesis (41 cases). The facial types in 172 patients were distributed among seven additional categories: median cleft lip (8 cases), bilateral or unilateral cleft lip with or without cleft palate (8 cases), cleft palate (5 cases), no oral cleft (60 cases), no oral cleft with SMMCI (16 cases), atypical facial cleft (4 cases), and no specified facial type (11 cases). The numbers of patients in these eleven facial type categories are presented in Table 1.
Alobar HPE was frequently observed in almost all facial categories, except for the median, lateral and no oral cleft categories where semilobar HPE was apparently more common (Table 1). Interestingly, the semilobar HPE type was observed in four of five cases of trisomy 13 with HPE type specified. A significant excess of females was observed in the ethmocephaly, cebocephaly, and premaxillary agenesis categories, and the male proportion was normal in the other categories. The proportion of other defects unrelated to the HPE spectrum did not vary among the facial categories. The recurrence of HPE or microform in the 195 families was approximately 19% (37/195; 95% CI: 13.9-25.3) without variation among the facial groups. Parental consanguinity is a rare event, with only three cases among the 158 cases specified for the data (1.9%; 95% CI: 0.5-5.9). The mean maternal age among 149 mothers was 26.9 252 Holoprosencephaly in South America years (95% CI: 25.8-27.8), and did not vary among the facial groups. There were nine cases of maternal diabetes among the 139 probands (6.5%, 95% CI: 3.2-12.3), and six of these cases exhibited premaxillary agenesis. The proportion of chromosomal anomalies or gene mutations did not differ among the facial categories (Table 1).

Micro-rearrangement analysis and chromosomal anomalies
MLPA screening was performed in 58 patients: ten patients with common trisomies (chromosomes 13 and 18), five patients with structural chromosomal anomalies, 23 patients with normal karyotypes, and 20 patients without karyotype information, showing failure in 15 cases and inconclusive results in eight cases. Among the 35 cases with MLPA results, 28 patients had HPE and seven patients had microforms. We also observed normal MLPA results in 27 cases, in which 14 patients had normal karyotypes, one patient exhibited 46,XX, inv(5) (p14.3; q23.1), and 12 cases showed no karyotype information. Altered MLPA results were observed in three of the 17 cases with normal karyotypes (17.6%; 95% CI: 4.7-44.2) or in three of the 29 cases with or without informed karyotypes (10.3%; 95% CI: 2.7-28.5). We detected two ZIC2 deletions and one deletion encompassing the SIX3 and SIX2 genes. The other altered MLPA results were observed in two cases of trisomy 13, and one case of 18p-, t(6;7) and t(7;14). An affected sibling of the patient with t(6;7) was also examined to verify the occurrence of the SHH deletion. Table 2 shows the phenotype and MLPA alterations observed in individuals with gene deletions.
Informed karyotypes were observed in 65 of the probands, and 20 patients had chromosome anomalies. We successfully performed QF-PCR analysis on 26 probands without informed karyotypes and detected chromosomal aneuploidies in five patients: one case of trisomy 21, one case of trisomy 18, and three cases of trisomy 13. Considering informed karyotype and the QF-PCR analysis, we obtained karyotype information for 91 patients (46.7%). Chromosomal anomalies, including aneuploidies or structural anomalies, were observed in 25/91 patients (27.5%; 95% CI: 18.9-38.0). The increased frequency of chromosomal abnormalities was observed in patients with associated HPE (22/51; 43.1%). Only three patients with isolated HPE or microform (3/40; 7.5%) showed chromosomal aberrations: 18p-, inv(5) and an unspecified chromosomopathy. The inv(5) observed in a SMMCI patient, was also observed in her father and brother, both exhibiting a normal phenotype.

Mutation screening
Mutation analyses were successfully performed in 120 patients for SHH and TGIF, 125 patients for SIX3, and 151 patients for ZIC2. All four genes were successfully sequenced in 119 patients: 45 patients had HPE, or micro- Savastano et al. 253  In total, 47 variants were detected, 35 variants were classified as benign or of unknown significance based on several features, including the presence of the variant in population SNP databases, synonymous benign variants predicted using PolyPhen and SIFT software, or UTR and intronic variants with unknown significance. We detected 12 damaging mutations in 119 probands (10.1%; 95% CI: 5.6-17.3) ( Table 3). The pathogenetic mutations included eight SHH gene mutations (66%), two SIX3 gene mutations (17%), and two ZIC2 gene mutations (17%). All 12 patients presented isolated HPE or microform.
The mutation frequency is similar (10.9%) among the probands with normal karyotype or QF-PCR results (six mutations out of 55 cases). The pedigrees of the six families with mutation recurrences are presented in Figure 1. The parents from six probands were available for study to determine the mutation inheritance. The clinical characterization and mutations identified in the probands and parents are summarized in Table 3. Figure 2 shows clinical photos and eletropherograms from the three most recently studied mutated patients. The other patients have been previously described (Orioli et al., 2001;El-Jaick et al., 2005, 2007a.

SHH mutations
We detected eight mutations in the SHH gene. Three patients have been previously described. El-Jaick et al.
described a newborn girl presenting the mutation c. 332T > A, p.(I111N), with alobar HPE and premaxillary agenesis, inherited from the mother affected with SMMCI and hypotelorism. Orioli et al. (2001) described the other two patients: 1) a 8-year-old male presenting the mutation c. 419A > C, p.(H140P) without specified HPE type, ocular 254 Holoprosencephaly in South America  hypotelorism, SMMCI, and strabism, inherited from the mother affected with mild ocular hypotelorism; and 2) a newborn male presenting the mutation c. 548G > T, p.(C183F) with alobar HPE, microcephaly, ocular hypotelorism, bilateral microphthalmia without lid opening, a single nostril with choanal atresia, and an ogival palate. The mutation c. 214C > T, p.(R72*) likely represents a de novo mutation in a female infant with lobar HPE, microcephaly, and ocular hypotelorism (El-Jaick, 2005). The mutation c. 381delC, p.(W128Gfs*58) was observed in a male newborn prenatally diagnosed with HPE, presenting a typical face, with microcephaly, ocular hypotelorism, cleft lip, micropenis, and rudimentary scrotum with nonpalpable testes (Figure 2 e, f, and g). An analysis of the family history revealed a paternal half-brother with HPE and a father with ocular hypotelorism. Unfortunately, DNA from the father and half-brother was not available. The mutation c. 482T > A, p.(L161Q) was observed in a female proband with HPE, microcephaly, ocular hypotelorism, premaxillary agenesis, flat nose, and pectus excavatum, presenting the karyotype 46,XX, 16qh+. The family history revealed a sister with HPE. The mutation was inherited from the father, who was not affected (El-Jaick, 2005). The mutation c. 548G > A, p.(C183Y) was detected in a male proband with alobar HPE, showing the absence of the corpus callosum and fusion of the thalami, a cystic area in the region of the cerebellar vermis, extending cranially through the midline, lissencephaly, microcephaly, ocular hypotelorism, small palpebral fissures, single nostril, and premaxillary agenesis (Figure 2, a and b). The patient presented CC = 22 cm at birth (< 3 rd centile) and 29.8 cm at 13 months of age (< 3 rd centile), with a normal karyotype (46,XY). The patient evolved with significant psychomotor retardation, feeding difficulties and difficult to control seizures, dying at age five. The parents were not available for this study. The mutation c. 718A > C, p.(T240P) was present in a fe-  male infant with microcephaly, ocular hypotelorism, single nostril, and pronounced ogival palate (Figure 2, c and d).
The family history revealed a nephew of the maternal grandmother with hydrocephalus, and a paternal uncle with a clubfoot. No relatives were available for further analysis in this study.

SIX3 mutations
SIX3 mutations were identified in two probands. The mutation c. 522CA, p.(Y174*) was detected in a woman with ocular hypotelorism and normal intelligence (El-Jaick, 2005). The patient was pregnant with two fetuses affected with HPE. The mutation c. 686C > T, p.(P229L) was observed in an eight-year-old male patient (El-Jaick et al., 2007a) with SMMCI, ocular hypotelorism, mild bilateral distichiasis, mammary hypertelorism, hyperfolded helices, an indistinct and short philtrum, mild mandibular prognatism, normal stature, and a mild learning delay but normal cerebral CT.
The mutation c. 1411_1415delGTGTC, p.(V471Rfs*57) was detected in a male infant with a prena-tal diagnosis of semilobar HPE, presenting microcephaly, ocular hypertelorism, and a rare cleft nose. The family history revealed a paternal half-sister with HPE. The father presented the same mutation, likely in mosaicism, with mild ocular hypotelorism (Savastano et al., 2014).

Discussion
More than 80% (160/195) of the patients in this study were newborn, and two-thirds of these patients were obtained from the ECLAMC, except for the older patients and those with SMMCI. The controlled epidemiological variables, such as sex, maternal age, maternal diabetes, parental consanguinity, chromosomal or gene defects, and the recurrence of HPE or microform in the family, showed that this sample did not differ from that previously studied at ECLAMC (Orioli and Castilla, 2007).
The phenotypic variables, such as type of HPE, facial types, presence of other cerebral defects or the presence of other defects not associated with the HPE spectrum, differed from those previously observed (Orioli and Castilla, 2007) when newborn patients may not present features, such as SMMCI and/or mental retardation. The expansion of the classical facial classification in HPE (DeMyers et al., 1964) resulted in 11 facial types. We examined some epidemiological and phenotypic variables according to these 11 facial types, to detect useful differences among them, as a facial description is more easily obtained than cerebral imaging or karyotype. The semilobar HPE type was less fre- Savastano et al. 257 quently observed among the four classical facial types, but the sample is too small to access the significance of this finding. Because the semilobar HPE type was more frequently observed in trisomy 13 cases (four among five with HPE type specified), we attempted to characterize the other chromosome anomalies with specified HPE types and observed only alobar HPE in four cases (18 trisomy, 21 trisomy, t(6;7), and del 18p). The slight female predominance of 1.4:1 (female:male) observed in the present study was consistent with that previously observed in other studies (Croen et al., 1996;Forrester and Merz, 2000;Mercier et al., 2011). This female predominance was restricted to three facial types: ethmocephaly, cebocephaly, and premaxillary agenesis. Patients with cyclopia showed a normal male proportion. This sexual difference might reflect causal differences of HPE associated with facial types. The increased frequency to be confirmed of maternal diabetes in patients with premaxillary agenesis, and ZIC2 mutations typically associated with normal or mildly affected faces (Brown et al., 2001), might be examples of association between cause and facial type.
Chromosomal anomalies are described in 32% to 41% of HPE cases, and trisomy 13 is the most frequently observed anomaly (reviewed by Solomon et al., 2010a). In the present study, karyotyping or QF-PCR analyses revealed chromosomal anomalies in 27.5% (25/91) of the informative patients. This frequency must be underestimated as 104 patients (53.3%) were not studied or had inconclusive results through QF-PCR. Furthermore, structural anomalies were not detected using this method. As expected, an increased frequency of chromosomal anomalies was observed in patients with HPE and associated malformations.
In addition to the usual cytogenetic study, tracing microdeletions involving genes associated with HPE through techniques, such as MLPA, contributed significantly to the diagnosis. The rate of chromosomal microanomaly detection through MLPA in the present study, ranging from 10.3% (3/29) to 17.6% (3/17 with normal karyotype), is not significantly different from the 5% frequency of microdeletions involving the main four HPE genes in newborns (Bendavid et al., 2006a) or the 9.8% frequency observed in patients without point mutations (Bendavid et al., 2006b). The frequency of microrearrangements and mutations in the four main HPE genes varied between cohorts of fetuses and live-born children. While mutations are more frequent in livebirths, submicroscopic deletions occur more frequently in fetuses (Bendavid et al., 2006a(Bendavid et al., , 2006b(Bendavid et al., , 2009). In the present study, which primarily included livebirths, the frequency of microdeletions, 10.3% (2.7-28.5), and mutations, 12/119 (10.1%; 5.6-17.3), are similar in both groups.
Array-CGH studies have reported frequencies of 17% to 22% losses or gains in the genome of HPE patients, considering known and new HPE loci (Bendavid et al., 2009;Mercier et al., 2011). However, the meaning of these candidate loci in the etiology of HPE was uncertain. In a genotype-first approach study, Rosenfeld et al. (2010) identified 136 individuals with deletions in any of the 35 selected HPE loci identified in a cohort of 26,922 patients studied through array CGH. The authors observed HPE in 13 individuals, 11 patients with deletions involving the four primary HPE genes, one patient with a deletion in the HPE8 locus (14q13), and one patient with a deletion in FGF8. The other six loci involving HPE genes or candidates were only associated with HPE microforms, but most of the deletions identified in HPE candidate genes showed no HPE, nor microforms of HPE.
The phenotype of patients with microdeletions in one of the four main HPE genes is highly heterogeneous. Similar to intragenic mutations, these patients present phenotypes ranging from alobar HPE to microforms Rosenfeld et al., 2010). No microdeletions were observed in the seven patients with HPE-microform or the patient with 46,XX, inv(5) (p14.3; q23.1), whose brother and father were carriers with the same inversion, showing normal results using the P-187 HPE Kit for MLPA, where probes against the candidate HPE gene, FBXW11, were located in 5q35.1. The patient presented with SMMCI, agenesis of corpus callosum, iris colobome, ocular hypotelorism, and a mild mental deficiency. Additional studies in this family will be important to determine whether the gene in this locus is associated with HPE.
The mutation frequency observed in the present study was 12/119 (10.1%; 5.6-17.3). Dubourg et al. (2004) reported a similar rate of 17% in fetuses and living children with normal karyotypes, while Nanni et al. (2000) observed mutation in only one case studied, representing 4.3% of the sample of newborn cases in a population from California. Lazaro et al. (2004) observed a similar frequency of 16% when considering syndromic and nonsyndromic HPE and patients with midline facial and/or cerebral anomalies without neuroradiological HPE. Although we did not conduct a functional analysis of the identified mutations, these anomalies affected conserved protein regions and therefore can be considered as contributors to the HPE phenotype.
The mutation ratio is higher in liveborn children (20% to 23%) than in fetuses (12.5% to 14%), who usually present more severe phenotypes due to chromosomal anomalies Bendavid et al., 2009). However, even if most of the sample comprised newborns, the mutational frequency is low compared with the cited literature, likely reflecting the high frequency of cases without karyotype observed in the present study (53.3%).
The frequency of mutations in SHH, 8/119 (6.7%; 3.2-13.2), is consistent with the results obtained in previous studies. Nanni et al. (1999) observed the same frequency (6.7%) in 344 HPE patients. Bertolacini et al. (2009) described a frequency of 8.1%, and Lazaro et al. (2004) detected mutations in the SHH gene in 8.6% of cases. 258 Holoprosencephaly in South America Paulussen et al. (2010) observed a lower frequency of 3.5% in a Dutch cohort of non-syndromic HPE, including fetuses, neonatal deceased children, children and adults. Roessler et al. (2012) included the data from Dubourg et al. (2004) and Paulussen et al. (2010) in their analyses and observed an aggregated frequency of 5.9% in 475 studied individuals. With the exception of the cohort from Paulussen et al. (2010), SHH is the most commonly mutated gene among HPE individuals.
Consistent with the study from Solomon et al. (2012), we observed the prevalence of missense mutations (6/8, 75%) and a similar prevalence of nonsense (1) and frameshift (1) mutations in the SHH gene. A nonsense mutation in codon 72 was detected in the patient 818 (Table 2). A quarter of the mutations described in SHH were nonsense or frameshift mutations (Roessler et al., 2009b;Solomon et al., 2012). Except for the mutation p.M457Rfx*18, these anomalies eliminated the autocatalytic processing site from the protein and are considered as null alleles (Roessler et al., 2009b). The mutation observed in the patient 5327 (Table 2) was a deletion of the nucleotide at position c.381 in SHH, resulting in a change in the reading frame and generating a stop codon at 58 residues after the alteration. This frameshift mutation resulted in a truncated protein terminating 14 residues before the autocatalytic cleavage site, likely generating a non-functional allele. The variant c. 482T > A, p.(L161Q), detected in the proband 576 (Table  2), was inherited from her father. This mutation is considered damaging (score 1.00). It is located in domain N-SHH, a conserved region that occurs among species, causes the substitution of a hydrophobic amino acid (leucine) to a hydrophilic amino acid (glutamine) and was not observed in the Exome Variant Server. Solomon et al. (2012) described the same mutation in a patient with unknown HPE and no other information about the phenotype. Mutation c. 548G > A, p.(C183Y) was identified in patient number 2086 (Table 2) and was also identified in a female HPE patient described by Roessler et al. (2009b), with no detailed clinical information, who inherited the mutation from her father (Solomon et al., 2012). This mutation occurred in the same position of the mutation observed in patient number 81 (Table 2) previously described by Orioli et al. (2001). Indeed, codon 183 is a hot spot for mutations, as this codon was altered in at least four unrelated families: mutation p.(C183R), detected in a family described by Roessler et al. (2009b) and Solomon et al. (2012); mutation p.(C183F), detected in a patient previously described by Orioli et al. (2001); and the two unrelated probands with mutation p.(C183Y) described in the present study, Roessler et al. (2009b), and Solomon et al. (2012). The altered region was localized in domain N-SHH and is highly conserved among species. The PolyPhen2 analysis revealed the mutation p.(C183R) as likely damaging (score of 1.00). The variant c. 548G > A was not described in the Exome Variant Server among more than 6,000 sequenced individuals. The alter-ation c. 718A > C, p.(T240P), detected in patient 2815 (Table 2), affects the C-terminal domain of the SHH protein. Roessler et al. (2009b) reported two mutations affecting the adjacent amino acid. This region is conserved among different phyla, and this change leads to the substitution of a threonine, an amino acid with hydrophilic characteristics, with a hydrophobic proline. This mutation was not described in the Exome Variant Server, and the PolyPhen2 analysis considered this mutation as likely damaging (score of 0.773).
In families where the SHH gene mutation is segregating, widely variable expressivity among affected members is observed (Solomon et al., 2012). Indeed, in the present study, at least half of the probands with mutations in the SHH gene inherited the mutation from one parent who was mildly affected. There was no disproportion of sex in the probands with the SHH mutation (proportion was 1:1), consistent with the results of previous studies (Mercier et al., 2011;Solomon et al., 2012). The small sample size prevented comparisons of the distribution of HPE types or facial features.
The SIX3 gene presented a mutation frequency of 2/119 (1.7%; 0.3-6.5), which is not different from the 4% frequency described by Dubourg et al. (2004) in 200 individuals with normal karyotypes, including fetuses and living children and is also not different from the 5.1% frequency reported by Mercier et al. (2011). Paulussen et al. (2010) observed a higher frequency of 10.5%. Mutations in the SIX3 gene have been predominantly observed in exon 1 (Lacbawan et al., 2009).
The ZIC2 gene has been previously described as the second most frequently mutated gene in HPE cases . In the present study, the frequency was only 2/119 (1.7%; 0.3-6.5), a similar frequency of mutations as observed in the SIX3 gene. Paulussen et al. (2010) also observed a 10.5% mutation frequency in ZIC2 and in SIX3 genes, although these frequencies were higher than the frequencies observed in the present study. Other studies have reported mutation frequency between 3% and 8.4% (Brown et al., 2001;Dubourg et al., 2004;Solomon et al., 2010b;Mercier et al., 2011;Roessler et al., 2012). Patients presenting mutations in the ZIC2 gene typically have mild facial defects or no facial anomalies (Brown et al., 2001). It is likely that there was a bias in the present study towards more typical faces, which would lower mutation frequency in gene ZIC2 and increase chromosomal defect rate. The two mutations in ZIC2 detected in the present study were observed in patients with semilobar HPE and microcephaly, one patient with a relatively non-dysmorphic face, and another patient presenting a nose cleft (Table 2) (Orioli et al., 2001;Savastano et al., 2014). Interestingly, four patients presented with atypical facial clefts. However, the other patient with a nose cleft showed normal results in the mutational analysis. Recently, there have been attempts to establish genotype-phenotype relations in HPE patients through studies including hundreds of mutated individuals for SHH (Solomon et al., 2012), ZIC2 (Solomon et al., 2010b) and SIX3 genes (Lacbawan et al., 2009). Although the number of cases in the present study identified with a chromosomal or gene defect is small (40/195; 20.5%), certain features that were previously identified in the genotype-first approach study of Rosenfeld et al. (2010) as the possible common cause between Dandy-Walker anomaly and HPE were observed in the present study. A total of 47 patients presented other cerebral defects besides or instead of HPE, and eight patients showed Dandy-Walker anomalies: two patients with trisomy 13, one patient with trisomy 18, one patient with a normal karyotype, and the remaining four patients did not have informed karyotypes.
The absence of mutations in the TGIF gene in our sample likely reflects sample size constraints, as the frequencies of mutations in this gene have been reported as approximately 1% .
The frequency of chromosome anomalies and mutations observed in the present study suggests that patients diagnosed with HPE must first be karyotyped, as previously recommended by other authors (Mercier et al., 2010;Pineda-Alvarez et al., 2010). Molecular methods, such as QF-PCR, are efficient for the diagnosis of associated HPE cases when premature death of newborn precludes cytogenetics studies. Isolated HPE cases might present structural alterations not detected using QF-PCR. In these cases, the MLPA technique is an alternative for the detection of gains or losses in key genes associated with HPE. Current recommendations for the diagnosis of HPE patients also include array-CGH analyses, as a sensitive and efficient technique for the identification of gains and losses in known or new HPE loci, which could clarify the etiology of many additional cases (Mercier et al., 2010). A mutation in one of the four main HPE genes explains only 10% to 20% of cases, and therefore mutation screening is recommended after the exclusion of chromosomal anomalies.