Our study is the first comprehensive molecular characterisation of de novo 13q deletions by means of array‐CGH and FISH, and allows us to address karyotype–phenotype correlations depending on the extent of the deletions. In agreement with previous reports,6
the overall phenotype severity in our series varies widely, with some patients being only slightly affected and carrying only dysmorphic features and minor anomalies, and others being seriously compromised.
Three patient groups have been distinguished on the basis of previous attempts to correlate defined 13q‐deletion intervals with specific phenotypic signs3
: patients carrying proximal deletions not extending into the q32 band and characterised by variable dysmorphic features, mild or moderate mental retardation and growth retardation (group 1); patients with deletions including at least part of q32 and the most serious phenotype, with severe mental and growth retardation and one or more major malformations most often involving the brain, eyes, distal limbs, genitourinary and gastrointestinal traits (group 2); and patients with more distal deletions involving bands q33–34, who usually show growth and severe mental retardation, but without gross malformations (group 3). In our cohort, four patients (2, 4, 5 and 6) fit into group 1, as they carry proximal deletions not extending to q32; eight patients (1, 3, 7–12) show the deletion of the entire 13q32 band (with the exception of patient 3 who lacks only the q32.1 sub‐band) and are assigned to group 2; and two patients (13 and 14) seem to belong to group 3 as they carry more distal deletions exclusively involving the q33–q34 bands. Table 2 lists the clinical features indicated as being typical of patients with 13q32 deletions,3,6
and their presence or absence in our three patient groups.
Table 2Main clinical features of patients with 13q deletions grouped according to the inclusion or exclusion of the 13q32 band
Patients of groups 1 and 3 do not perfectly fit the suggested phenotypic features,3
especially with regard to mental retardation: in group 1, this was scored mild and moderate in patients 2 and 5, but severe in patient 6; and, in group 3, patients 13 and 14 show only moderate mental retardation, but they lack brain anomalies and other major malformations, according to Brown et al
However, our findings confirm that group 2 patients (lacking the 13q32 band) are the most seriously affected: all of the patients for whom data are available show severe and profound mental retardation, growth delay, microcephaly and brain anomalies or NTDs (except for patient 3, who lacks only the q32.1 sub‐band; table 2). The microphthalmia and coloboma previously reported in patients with the 13q32 deletion6
are confined to our group 2 patients (table 1 patients 1, 8, 10–12). Patients 8 and 10–12 share coloboma (table 1) and a common deleted region ranging from 13q32.2 to 13q33.2 (fig 1). In addition to coloboma, patients 10–12 show microphthalmia and blindness (table 1), and a common deleted interval extending from 13q33.2 (the distal bkp of the “coloboma” region) to 13q33.3 (the deletion breakpoint of patient 13 who does not display any eye abnormality; fig 1). The EFNB2
(ephrin‐B2) gene maps within the 13q33.2–13q33.3 interval. This gene belongs to a family of ligands with specificity for the eph receptors, which, in animal models, participate in several aspects of visual system development.16
It might be hypothesised that haploinsufficiency in this gene is responsible for the more severe eye anomalies found in patients 10–12 as compared with patient 8. Hand and foot anomalies in our sample are not restricted to group 2 (table 1), but patients 10 and 11 show the absent or hypoplastic thumbs considered to be the most representative of these features.3,15,17
Given that patients 10–12 present with the most severe eye abnormalities, but diverge from the absent or hypoplastic thumbs phenotype, assignment to 13q33.2–13q33.3 of gene or genes involved in absent or hypoplastic thumbs cannot be definitely inferred.
Patients 1 and 2 carry deletions including the RB1 locus (13q14.2). No signs of retinoblastoma were evident at prenatal ultrasound examination in patient 1, whereas patient 2 did not develop the tumour until age 11 years. In contrast with most patients with germline RB1 mutations, a relevant proportion of those carrying a cytogenetically visible deletion involving the 13q14 band show unilateral disease and some develop no tumours at all.18,19
This is the case in our postnatal patient 2.
Congenital heart defects are heterogeneous and do not seem to be associated with 13q32 or any other specific deleted segments; and renal and gastrointestinal malformations are infrequent, even in group 2.
Bartchs et al20
suggested that the 13q32.2–q34 region plays an important role in genital development. Two of our group 2 patients show ambiguous genitalia (patient 10) and a minor genital abnormality (patient 11; table 1), thus confirming the possible involvement of the q32 band in genital development. However, patient 6 (carrying a 13q22.1 or q22.2–q31.3 deletion) shows ambiguous genitalia, thus suggesting that other proximal genes may be involved in genital anomalies due to haploinsufficiency.
The critical 13q32 band responsible for the severe phenotypes of group 2 patients has subsequently been restricted to a 1.2‐Mb region between markers D13S136 and D13S147,6
containing the ZIC2
gene at q32.3.7,8
The observation that heterozygotic ZIC2 mutations are associated with HPE in patients without chromosomal abnormalities led to the hypothesis that ZIC2 haploinsufficiency may partially underlie the brain malformations seen in patients with the 13q deletion and, in line with this hypothesis, the only one of our group 2 patients without CNS anomalies (patient 3) retains the q32.3 sub‐band and, consequently, the ZIC2
gene. Luo et al9
have proposed a critical region at 13q33.2‐qter, distal to and not overlapping Brown's critical band specifically involved in NTDs, and suggested that one or more genes mapping to this band and distal to ZIC2 may cause NTDs as a result of haploinsufficiency. Two of our patients show NTDs (tables 1 and 2): patient 7 with a 13q22‐qter deletion, and patient 9 with an interstitial 13q31.1–q33.1 deletion and the maintenance of the entire proposed critical region. The distal breakpoint at q33.1 indeed maps 1.5 Mb proximally to the microsatellite markers D13S274‐D13S1311, fixing the q33.2 breakpoint of the terminal deletion indicated by Luo et al
On the basis of these observations, we presume that other dose‐sensitive genes proximal to q33.2 may be involved in NTDs if they are haploinsufficient, and this favours the hypothesis of the presence of more than one locus for NTDs in the 13q32‐qter region.6
The role of ZIC2 mutations as a common cause of human NTDs was excluded by Brown et al
but it should be noted that ZIC5 (another member of the ZIC gene family) maps 16 kb proximally to ZIC2. Deficiency of the Zic5 mouse orthologue is associated with NTDs,22
which makes the hypothesis that ZIC5 deletion may be responsible for human NTDs attractive even if no human ZIC5 mutations have yet been described.23
Another locus within 13q22–33 (different from ZIC2) has been proposed as a cause of DWM, a specific CNS abnormality associated with distal 13q deletion.10,11
Our patient 8 has DWM, and patients 11 and 12 show DWMs respectively associated with cerebellar hypoplasia and agenesis of the corpus callosum. As the breakpoints of their deletions (fig 1) are localised distally to the centromeric boundary of the previously reported DWM critical region,10
we narrowed the minimal DWM‐associated deletion interval to 13q32.2–33.2, and a recent report of a 13q31.2/32.1‐qter deletion in a fetus with DWM and other malformations supports our finding.24
The first locus involved in human DWM has been localised to an interval encompassing the ZIC1
genes in the 3q2 region and, on the basis of the observation that heterozygotic deletions of both mouse orthologues Zic1 and Zic4 cause a DWM‐like phenotype, it has been hypothesised that a heterozygotic loss of both ZIC1 and ZIC4 genes may cause DWM in patients with 3q deletion.25
ZIC1 and ZIC4 genes map to chromosome 3 in a configuration that is paired with that of ZIC2 and ZIC5 on chromosome 13. Like the four previously reported patients with DWM,10,11,24
our patients with DWM lack the 13q32.3 band encompassing the ZIC2
genes. It is therefore possible to assume that the loss of both ZIC2 and ZIC5 may cause DWM in patients with the 13q deletion, like the loss of ZIC1 and ZIC4 in patients with the 3q deletion. Grinberg et al23
have suggested that the loss of ZIC2 may lead to HPE, and the loss of both ZIC2 and ZIC5 may lead to DWM with HPE. However, all our patients with CNS anomalies lack both genes and have different CNS features, thus making it difficult to support this assumption. Clinical studies on patients with 13q deletions including only one of the two genes may be useful for establishing their possible role in determining CNS‐specific phenotypes, although further studies on human mutations and their phenotype effects are necessary to clarify the possible role of ZIC5 in CNS anomalies.
- Chromosome 13q deletion is associated with varying phenotypes and there is still no acknowledged consensus correlation between the monosomy of distinct 13q regions and specific clinical features.
- We studied 14 Italian patients carrying de novo 13q deletions by means of array‐comparative genomic hybridisation or fluorescent in situ hybridisation.
- All our patients showed severe to mild mental retardation: eight had central nervous system (CNS) anomalies, six had eye abnormalities, nine had facial dysmorphisms and 10 had hand and/or feet anomalies. The size of the deleted regions varied from 4.2 to 75.7 Mb.
- Our study confirms that patients lacking the 13q32 band are the most seriously affected. Dose‐sensitive genes proximal to q33.2 may be involved in neural tube defects (NTDs). The minimal deletion interval associated with the Dandy–Walker malformation (DWM) was narrowed to the 13q32.2–33.2 region and we proposed ZIC2 and ZIC5 genes as underlying various CNS malformations.
An additional consideration is that the loss of ZIC2 and/or ZIC5 may underlie the various CNS malformations seen in patients with the 13q deletion. Indeed, loss of the same genes leading to different CNS phenotypes can be ascribed to the simultaneous loss of contiguous but different chromosomal regions that may influence their expression. Alternatively, variable expression of such genes may be caused by modifier loci modulating the severity of the traits. Furthermore, it is well known that patients carrying deletions of precisely the same chromosomal regions may show different phenotypic features depending on the genetic background in which they are placed.