Chromosome microarray analysis has revolutionised the diagnosis of patients with neurocognitive impairment and has resulted in the association of many phenotypes with copy number changes in particular genes.

Recently in this journal, Brunetti-Pierri et al.1 published a report of seven patients with duplications at 14q12 that included the FOXG1 (MIM 164874). These patients presented with relatively severe neurodevelopmental phenotypes comprising intellectual disability, epilepsy and severe speech delay. Previously, we had reported a similarly affected patient with a deletion that also included FOXG1,2 and more recently three patients with West syndrome were reported to have duplications that included FOXG1.3, 4 FOXG1 encodes a brain-specific transcriptional repressor,5 and deletions and point mutations in FOXG1 are known to cause the congenital variant of Rett syndrome,6 making duplication of FOXG1 a strong candidate for the neurocognitive impairment in patient with 14q12 duplications.

In the course of a research project to identify genetic causes of hemifacial microsomia (MIM 164210), we identified a father–son pair who both have an interstitial duplication of ∼88 kb at 14q12 (chr14:28,236,716-28,325,210; UCSC genome browser, NCBI Build 36/hg18) (Figure 1). The duplication was identified using an Illumina CytoSNP-12 microarray (Illumina Inc., San Diego, CA, USA) and independently confirmed using a synthetic multiplex ligation-dependent probe amplification assay7 and/or Affymetrix 2.7M microarray (Affymetrix Inc., Santa Clara, CA, USA). The duplication contains only two genes, FOXG1 and C14orf23. Importantly, apart from hemifacial microsomia, the father and son are phenotypically normal, have normal intellect, do not have epilepsy, and have no family history of epilepsy or cognitive impairment.

Figure 1
figure 1

UCSC genome browser (NCBI36/hg18) view of reported proximal 14q duplications (blue bars) encompassing FOXG1 and neighbouring genes. Grey shadow highlights the duplicated region reported here. The colour reproduction of this figure is available at the European Journal of Human Genetics online.

Full size image

It is difficult to reconcile the normal neurocognitive phenotype in this father and son pair with the relatively severe impairment reported in other patients with duplications that include FOXG1; however, several possible explanations must be considered. First, it remains possible that FOXG1 duplication is benign, and that the neurocognitive impairment reported in patients with 14q12 duplication is the result of duplication of other genes in the vicinity. Second, FOXG1 duplication may be incompletely penetrant, manifesting clinical abnormality only in the presence of other genetic or environmental factors. Third, in our father–son pair it is possible that of the three detected copies of FOXG1, only two are functional. Finally, FOXG1 may be subject to long-range regulatory elements, with gene transcription being differentially affected according to the location of the duplication breakpoints.

A duplication at 14q12 that encompasses FOXG1is also recorded in the Children's Hospital of Philadelphia CNV database, which comprises CNV data from 2026 healthy children aged 0–18.8 This duplication is similar in size to the ∼3 Mb minimal duplicated region that includes FOXG1, c14orf32 and PRKD1 described in affected patients reported by Brunetti-Pierri et al.1 Patients carrying these small-sized duplications (cases 1 and 5 in Figure 1) were assessed as non-dysmorphic, so it is possible the healthy CHOP patient is yet to present with developmental problems, infantile spasms or other seizures. Alternatively, this case provides further evidence that FOXG1 duplication may be benign or incompletely penetrant.

On the basis of these data, the role of duplication of FOXG1 in the pathogenesis of cognitive impairment and epilepsy remains uncertain. This case is a salient reminder that our understanding of the relationships between CNVs and phenotype is far from complete, and of the importance of reporting CNVs that are found in the presence of normal phenotypes. This is particularly important in the context of the increasing use of molecular karyotyping for prenatal diagnosis, where decision-making may be based on evidence of questionable validity.