INTRODUCTION

The corpus callosum is the major interhemispheric fiber bundle in the brain [Aboitiz and Montiel, 2003], and consists of about 200 million axons in humans, that is, approximately 2–3% of all cortical fibers, thus making it the largest fiber tract within the central nervous system.

Formation of the corpus callosum begins as early as 6 weeks of gestation when axons destined to cross the midline can be seen growing medially within the hemispheres. At 11–12 weeks of gestation, the first fibers cross the midline through the massa commissuralis, which is located between the anterior and hippocampal commissures, to form the corpus callosum. In the developing brain, axon tracts generally form according to a conserved ontogenic sequence as shown in animal models [Hatten, 1999; Mihrshahi, 2006]. Most axon tracts develop along non-neural substrate cells such as glia, which guide the first pioneering axons to their targets [Holley, 1982; Nordlander and Singer, 1982; Norris and Kalil, 1991]. Thus, a number of glial populations have been found to play a role in the development of the corpus callosum. One of these populations, the so-called midline zipper glia, has been shown to guide the process of midline fusion, a necessary event in the run-up of the formation of the corpus callosum [Silver, 1993]. Two other glia populations, a “glial wedge” being formed in the dorsomedial lateral ventricles, and another glial population being formed in the region of the indusium griseum [Shu and Richards, 2001; Lent et al., 2005], have been identified and shown to play important roles in corpus callosum development. Another structure called “midline sling,” mainly consisting of migratory neurons, forms a midline bridge along which callosal axons can grow to reach the contralateral hemisphere. It has been shown that absence of, or damage to, this sling results in agenesis of the corpus callosum [Silver et al., 1982]. By 18–20 weeks of gestation, the corpus callosum has assumed its final shape except that it will continue to thicken and grow caudally.

Agenesis of the corpus callosum (ACC) is one of the most frequent malformations in brain with a reported incidence raging between 0.5 and 70 in 10,000 [Myrianthopolous, 1977; Jeret et al., 1986]. The prevalence in children with developmental problems has been estimated to be as high as 230 in 10,000 [Jeret et al., 1986, 1987]. Two types of ACC can be distinguished morphologically: (1) ACC type 1, in which axons form but are unable to cross the midline; they consecutively form large aberrant fiber bundles known as Probst bundles along the medial hemispheric walls. (2) ACC type 2, in which commissural axons fail to form; therefore, no Probst bundles are found.

Disorders of the corpus callosum can also be observed in association with major malformations of the embryonic forebrain prior to formation of the anlage of the corpus callosum (e.g., holoprosencephaly, HPE).

ACC is a heterogeneous condition, which can be observed either as isolated condition or as one manifestation in the context of a congenital syndrome. Among the most frequent clinical findings in patients with ACC are mental retardation (60%), visual problems (33%), speech delay (29%), seizures (25%), and feeding problems (20%) [Schilmoeller and Schilmoeller, 2000]. Furthermore, even in cases with no developmental delay and normal intelligence mild behavioral or social problems as well as the attention-deficit-hyperactivity disorder (ADHD) have been described [Brown and Paul, 2000; Doherty et al., 2006].

ACC can be caused by exogenous factors, for example, maternal alcohol use during pregnancy [Sowell et al., 2001] or maternal phenylketonuria [Levy et al., 1996] as well as by genetic factors. Several syndromes that include ACC having autosomal-dominant, autosomal-recessive and X-linked inheritance have been recognized [Dobyns, 1996; Online Mendelian Inheritance in Man (OMIM), 2008], and several causative gene mutations have been identified so far (Table I). In addition, ACC has been observed in constitutional trisomies as well as in some chromosomal rearrangements like del(4)(p16), del(6)(q23), dup(8)(p21p23), dup(11)(q23qter), suggesting that causative genes may be located in these chromosomal regions (Table II).

On the other hand, cases with isolated ACC and developmental delay without detectable chromosomal changes have also been published with apparent autosomal dominant, autosomal recessive or X-linked modes of inheritance. To our knowledge, causative genes have not been found to date [Serur and Jeret, 1988; Dobyns, 1996].

Much is known about ACC being a part of certain conditions (Tables I and II), or in terms of its association with social, behavioral and medical problems [Brown and Paul, 2000; Schilmoeller and Schilmoeller, 2000; Doherty et al., 2006]. In contrast, there is confusion about the terminology concerning partial absence of the corpus callosum (DCC), where various desgnations are used including hypogenesis, hypoplasia or partial agenesis. In fact, radiological terminology to describe corpus callosum abnormalities used in literature is rather confusing and heterogeneous. For the purpose of this study, only complete absence of the corpus callosum was addressed as agenesis (ACC), and partial absence of the corpus callosum as dysgenesis (DCC). This reflects on the findings by Rubinstein et al. [1994] that the partial appearance of the corpus callosum may be due to a process to overcome initial abnormalities of midline structures resulting in a variety of shape, size and location of an observed callosal structure not necessarily corresponding to a “normal” corpus callosum.

We were interested in finding out, whether we could relate the ACC and DCC seen in patient series classified using a standardized protocol for describing the available neuroimaging to certain underlying genetic causes. Our purpose was to uncover unknown chromosomal regions associated with, or other changes contributing to ACC and DCC.

METHODS

In this retrospective study, data of children seen in our institution for suspected disorders of the corpus callosum between 1984 and 2006 were reviewed. A total of 172 patients were identified in whom corpus callosum abnormalities were documented in the files (group 1). According to our records, the corpus callosum was completely absent in 63 patients and partially absent in 28 patients. In another eight patients the corpus callosun was missing due to holoprosencephaly (HPE), and 73 patients had a corpus callosal hypoplasia (CH).

Another 23 patients (group 2) were added to the study, when seen in our institution, with chromosomal rearrangements known to be associated with corpus callosum changes. In total, we identified 195 patients in groups 1 and 2. In a total number of 126 cases, nine computed tomographies (CT) and 117 magnetic resonance images (MRI) of the brain were performed. In 82 cases, in which a corpus callosum abnormality had been described or suspected, either no neuroimaging other than ultrasound had been performed (n = 65) or detailed imaging studies were not available (n = 17).

In all 126 cases with available neuroimaging, the images were re-evaluated using a standardized protocol focusing on midline and cortical defects by a neuroradiologist specialized in pediatric neuroradiology. In this structured evaluation only complete absence of the corpus callosum was addressed as agenesis (ACC), and partial absence of the corpus callosum as dysgenesis (DCC) [Rubinstein et al., 1994].

In 43 cases, the previous corpus callosum findings were revised: Three times from ACC to DCC, five times from ACC to corpus callosum hypoplasia (CH), four times from DCC to CH, four times from DCC to ACC, once from CH to ACC, five times from CH to DCC, once from CH to HPE, and once from HPE to DCC. Nineteen patients, who had previously been described to have ACC (n = 2), DCC (n = 3), HPE (n = 1), or CH (n = 13) had no corpus callosum abnormalities at all.

Of the 35 patients with ACC, 28 were seen by a clinical geneticist. In the seven remaining cases, an appointment was offered, but the parents declined. Of the 18 patients with DCC, 13 were seen by a clinical geneticist. In the five remaining cases an appointment was offered, but the parents either declined (n = 4) or were lost to follow-up.

Clinical genetic investigation of a patient included a thorough phenotype assessment including minor anomalies of head, neck, skin, and extremities. Genetic laboratory investigation included chromosome and subtelomeric analyses. For the purpose of this study, a diagnosis was considered chromosomal only, if an unbalanced karyotype was either in conventional, subtelomeric or array-CGH analysis. According to the obtained genetic data and/or a clinically suspected diagnosis, further investigation (e.g., particular gene analysis, microarray-based comparative genomic hybridization in case of severe mental retardation and/or dysmorphic phenotype) was considered and offered.

Array-CGH analysis using a whole genome tiling path BAC array was performed as described previously [Klopocki et al., 2006]. In brief, patient and reference DNA were labeled using a Bioprime CGH labeling kit (Invitrogen, Karlsruhe, Germany) and hybridized on the array (SlideBooster, Implen, Munich, Germany). Analysis and visualization were performed with CGHPRO software. Copy number changes were determined by a conservative log₂ ratio threshold (gain ≥ 0.3; loss ≤ −0.3). Profile deviations consisting of three or more neighboring BACs are considered as genomic aberrations.

RESULTS

After the neuroradiological re-assessment, 110 patients were identified with distinctive features of the corpus callosum:

• Thirty-five patients with ACC (34 from group 1 and one from group 2).

• Eighteen patients with DCC (all from group 1).

• Fifty patients with CH (47 from group 1 and 3 from group 2).

• Seven patients with HPE (all from group 1).

Of the 35 patients with ACC, 28 were seen by a clinical geneticist, two of them being brother (A-13) and sister (A-11), and the following diagnoses were established:

• Mowat–Wilson syndrome (n = 2; OMIM #235730).

• Walker–Warburg syndrome (n = 1; OMIM #236670).

• Oro-facial-digital syndrome type 1(n = 1; OMIM #311200).

• Six chromosomal changes (only one of them (A-14) being from group 2).

An additional patient had an apparently balanced reciprocal translocation, which led to the disruption and a predicted loss of function of a gene (FOXG1B gene), which had not been described previously [Shoichet et al., 2005]. In a total of 19 patients subtelomeric chromosome analyses were performed with normal results.

Of the 18 patients with DCC, 13 were seen by a clinical geneticist, and two chromosomal rearrangements were found (none of the patients were from group 2). In 12 cases subtelomeric analyses were performed, which revealed an unbalanced karyotype in one case. Beside the two siblings (A-13, A-11) with ACC, no other patients were related.

Seventeen patients with ACC and two patients with DCC showed Probst bundles, while in nine patients with ACC and nine patients with DCC Probst bundles were absent. In two cases with ACC and two cases with DCC it remained unclear due to insufficient image quality whether Probst bundles are present or not. A repeat MRI was declined by the parents.

Associated cortical malformations were frequent in our study population:

• polymicrogyria was noted in eight patients with ACC (29%) and in one patient with DCC,

• pachygyria/incomplete lissencephaly was seen in nine patients with ACC (25%) and in one patient with DCC, and

• heterotopia was noted in four patients with ACC (14%) and in one patient with DCC.

Dandy–Walker malformation, Chiari malformation or delayed myelination constituted less frequent findings (Table III).

In 11 of 28 patients with ACC seen by a geneticist, a genetic diagnosis could be established—in the remaining 18 cases the genetic basis of the ACC remained unknown (Table IV). Five cases had non-chromosomal diagnoses:

• Two boys (A-12, A-26) had Mowat–Wilson syndrome (OMIM #235730): One of them (A-12), who was $1\frac{3}{12}$ years old at the time of investigation, had a de novo deletion of two base pairs in exon 5 (nt553-554) of the ZFHX1B gene. The other one (A-26), who was $1\frac{7}{12}$ years old at the time of investigation, had a de novo deletion of a single nucleotide (nt2176) of the same gene leading to a truncated polypeptide.

• One girl (A-28), who was $8\frac{3}{12}$ years old at the time of investigation, clinically had an oro-facial digital syndrome type 1 (OMIM #311200), which was confirmed by molecular genetic analysis of the CXORF5 gene revealing a frameshift mutation in exon 5 of the gene not found in the girl's mother.

• Another female patient (A-23) first seen at the age of 14 months by a clinical geneticist, had an apparently balanced translocation t(2;14)(p22;q13) leading to a disruption of the FOXG1B gene at the breakpoint of chromosome 14 published by Shoichet et al. [2005].

• A male patient (A-19) with Walker–Warburg syndrome (OMIM #236670), who had a congenital hydrocephalus, dysplastic cortex with agyria, optical atrophy, retinal hemorrhage, and congenital muscular dystrophy, passed away at the age of 9 months. Unfortunately, no samples were available to perform further genetic testing besides chromosome analysis which was normal.

Six cases with ACC, mental retardation, and dysmorphic features had various chromosomal changes:

• A $10\frac{7}{12}$-year-old boy (A-07) had a mosaicism in fibroblast culture with a karyotype of 46,XY[19]/46,XY,del(18)(pter → q21:)[31], but a normal karyotype of 46,XY in lymphocytes.

• A $1\frac{8}{12}$-year-old boy (A-15) had partial monosomy 3p with karyotype 46,XY,del(3)(pter→ p25).

• A $6\frac{3}{12}$-year-old boy (A-24) had trisomy 8 mosaic, karyotype 46,XY[7]/47,XY, +8[93].

• A 13-month-old boy (A-20) with severe mental retardation, macrocephaly, hearing impairment and dysmorphic appearance had trisomy 8pter → 8q11.1 and trisomy 12q11.1 → 12pter mosaic due to a de novo translocation 8p;12p resulting in a dicentric marker chromosome, karyotype 47,XY, +dic(8;12)( 8pter → 8q11.1::12q11.1 → 12pter)[28]/46,XY[72].

• A 9-year-old female patient (A-14) had a partial trisomy 8p in combination with a partial monosomy of the very distal region of 8p due to an inverted duplication 8p23.1 → p11.2 with a deletion of 8p23.1 → pter.

• In a $3\frac{5}{12}$-year-old male patient (A-16) with ACC, severe mental retardation, seizures, and dysmorphic features, who had delayed myelination as well as a complex malformation of cortical development with pachygyria and polymicrogyria, chromosome and subtelomeric analyses were performed with normal results. In this case, we additionally applied a whole genome tiling path BAC array in order to investigate the genomic DNA for submicroscopic aberrations. We detected two genomic aberrations: a duplication of 6q25.3-q26 as well as a duplication of 11q25. While the duplication 6q has been classified as a genomic variant (database of genomic variants, version December 13, 2005; http://projects.tcag.ca/variation/), the 230 kb duplication of 11q has not been described as genomic variant yet (Fig. 1). The parents of the patient were investigated as well, both showing no duplication of 11q25.

In 2 of 13 patients with DCC a chromosomal diagnosis could be established (Table IV):

• A $2\frac{9}{12}$-year-old girl (D-01) had a partial trisomy 11q and a partial monosomy 6q due a translocation (6;11) revealed by subtelomeric analysis, karyotype: 46,XX,ish der(6)t(6;11)(6qtel-,11qtel+). Analyses of the parents revealed that the child's mother carried a balanced translocation involving chromosomes 6, 11 and 14. Breakpoints were refined by chromosome microdissection showing karyotype: 46,XX,rev ish t(6;11;14) (6pter → q26::11q23.3 → qter; 11pter → q23.3::14q22 → qter; 14pter → q22::6q26→qter).

• Another 12-month-old female patient (D-11) had a partial monosomy 7q, karyotype 46,XX,del (7q32 → qter). Remarkably, Sonic Hedgehog, one of the major genes accounting for HPE and midline defects in brain, is located in region 7q36.

The major clinical findings of all the patients with ACC and DCC are summarized in Tables V and VI. Findings present in a patient were marked with “+,” traits absent with “−” If information concerning a particular trait was not informative, a question mark (?) was used.

The two siblings A-13 and A-11 with ACC, were listed and counted individually because they were discordant for major clinical findings, though they are likely to have a common genetic etiology for the corpus callosum abnormality. The female patient A-11 had a developmental delay and muscular hypotonia. Her brother A-13 developed normally and was healthy. The ACC in his case was revealed by chance due to a postnatal ultrasound. The differences between these siblings may be due to a different genetic background modifying a common genetic cause for ACC, which could be also gender specific, or the corpus callosum changes and the associated clinical findings have in fact different genetic or even non genetic causes in both of these two children.

Concerning the patients with ACC, 25 out of 28 (89%) had a developmental delay. Twenty-one of 24 patients (88%) had a delay in speech development and 21 of 25 patients (84%) had feeding problems. Fifteen of 25 patients (60%) had visual problems, and three of 17 (18%) were hearing impaired. Thirteen of 21 patients (62%) developed seizures, 21 out of 23 (91%) had abnormal muscular tone.

Concerning the patients with DCC, all had developmental delay, and a delay in speech development and eight out of 13 patients (62%) had feeding problems. Eight of 12 patients (67%) had visual problems, and one of 11 (10%) were hearing impaired. Eight of nine patients (89%) developed seizures, 10 out of 12 (83%) had abnormal muscular tone.

Only slightly different frequencies of the major clinical findings were found, when patients without specific diagnosis are taken separately. These findings including the frequencies are given in Tables VII and VIII. Concerning the patients with ACC without a diagnosis, 14 out of 17 (82%) had a developmental delay. Twelve of 15 patients (80%) had a delay in speech development and 11 of 15 patients (73%) had feeding problems. Eight of 15 patients (57%) had visual problems, nine of 13 patients (69%) developed seizures, and 11 out of 13 (85%) had abnormal muscular tone.

Concerning the patients with DCC, 6 out of 10 patients (60%) had feeding and/or visual problems, 7 of 8 patients (89%) developed seizures, and 8 out of 10 (88%) had abnormal muscular tone.

Interestingly none of these patients with ACC, and only one out of 10 patients with DCC (10%) was hearing impaired, suggesting that hearing problems were part of the specific diagnoses in the other patients.

DISCUSSION

Since ACC is one of the most frequent brain malformations in children with developmental delay it is crucial to establish consistent criteria in the radiological assessment of the corpus callosum description in order to achieve reproducible and comparable results. In this study, we decided to include only patients with complete absence of the corpus callosum (ACC), or partial absence (dysgenesis) of the corpus callosum (DCC), based on the terminology that has been introduced by Rubinstein et al. [1994]. Known genetic causes for the absence of the corpus callosum are chromosomal rearrangements and several genetic disorders with autosomal dominant, autosomal recessive and X-linked mode of inheritance (see also Tables I and II).

The results of our study confirmed data known from the literature, and moreover added to the current knowledge (Table IX). In the case of patient A-16, where array-CGH analysis revealed a 230 kb spanning de novo microduplication in 11q25. This has not been described as genomic variant before (see also Fig. 1) and was not present in the patient's parents. In the light of partial trisomy of the region 11q23 → qter is known to be involved in ACC [Rott et al., 1972] this finding is suggestive that the region in this patient is the smallest region of overlap described so far involved in corpus callosum formation (see also Fig. 1). In the case of A-23, who had a cytogenetically balanced translocation t(2;14)(p22;q13), we were able to show that this rearrangement led to a disruption of the FOXG1B gene at the breakpoint of chromosome 14 [Shoichet et al., 2005]. As the chromosomal region 14q13 is known to be involved in HPE formation [Kamnasaran et al., 2005], the FOXG1B gene could be a good candidate not only for ACC but for HPE as well.

Interestingly, in none of the patients with DCC a non-chromosomal diagnosis could be established. In six out of 28 patient with ACC a chromosomal change was found, but only in two out of 13 patients with DCC, one of these identified with subtelomeric analysis, suggesting that DCC is not seen in complex chromosomal and non-chromosomal disorders.

As mentioned before, the most common clinical findings in patients with ACC are described in the literature to be mental retardation (60%), visual problems (33%), speech delay (29%), seizures (25%), abnormal muscular tone (25%), and feeding problems [Schilmoeller and Schilmoeller, 2000]. Though the study of Schilmoeller and Schilmoeller [2000] involved more patients than in the present study (596 families from the US and 12 other countries provided dermographic information, and a profile of their child by completing a questionnaire), the approach was completely different from our study, and, therefore, comparing results and frequencies is difficult: In the study reported by Schilmoeller and Schilmoeller [2000] all information on patients were collected merely on the basis of a questionnaire completed by the families themselves and not by professionals. Consistent clinical investigation was not offered and performed in the patients and Magnetic Resonance Imaging was not performed in all of them (only in 33.3%); neuroimaging was not evaluated according to standardized criteria. Therefore, the higher frequencies for these features found in the present study may be due to different usage both of the radiological terminology, and of the diagnostic criteria underlying the assessment and evaluation of the particular clinical features. Unfortunately, Schilmoeller and Schilmoeller [2000] did not report on genetic or biological causes of the patients as well.

There are several limitations in our study that need to be taken into account when interpreting the data. Recent neuroimaging data were not available in all patients. As our study included exclusively patients, in whom corpus callosum abnormalities had previously been reported in the files, it remains unknown how many patients seen in our institution actually had a pathology of the corpus callosum. It is known that even patients with normal intelligence, mild behavioral or social problems as well as the attention-deficit-hyperactivity disorder (ADHD) may have ACC [Brown and Paul, 2000; Doherty et al., 2006].

Though the genetic basis of the complete and partial absence of the corpus callosum was identified in 13 of 41 patients (32%), the cause remained unknown in 68%. Further studies need to be performed to elucidate the genetic and biological bases of corpus callosum formation, and its complete or partial absence as well as associated brain malformation, and clinical findings.