In our cohorts there is a clear common terminal 6q deletion phenotype characterised by a small head size, dysplastic outer ears, hypertelorism, vision problems, abnormal eye movements, dental abnormalities, feeding problems, recurrent infections, respiratory problems, spinal cord abnormalities, abnormal vertebrae, scoliosis, joint hypermobility, brain abnormalities (ventriculomegaly/hydrocephaly, corpus callosum abnormality and cortical dysplasia), seizures, hypotonia, ataxia, torticollis, balance problems, developmental delay, sleeping problems and hyperactivity. Since these characteristics are common in the affected individuals, it is most likely that the phenotype is caused by haploinsufficiency of the most distally located genes. Nonetheless, the phenotype is very variable, and not all characteristics are seen in all individuals.
The three most distally located HI-genes that could contribute to this common 6q terminal deletion phenotype are
TBP (Tata Box-Binding Protein, MIM*600,075)
, PSMB1 (Proteasome Subunit Beta-Type 1, MIM*602,017) and
DLL1.
TBP is known to cause late-onset neurological disorders such as spinocerebellar ataxia (MIM#607136) and Parkinson’s disease (MIM#168600) through expansion of a CAG repeat [
39], but it is unclear whether loss of function of one allele has a phenotypic affect. Mice haploinsufficient for
Tbp did not show any abnormalities [
40], and no pathogenic loss-of-function variants have been reported in humans. For the gene
PSMB1, no pathogenic heterozygous variants have been reported. Recently, a presumed pathogenic homozygous missense variant was reported resulting in microcephaly, developmental delay and short stature in two sisters, aged 22 and 35 years [
41]. These clinical characteristics are present in our cohort but also appear in individuals with an interstitial 6q27 deletion that did not involve
PSMB1 [
7,
29]. Furthermore, there are multiple individuals with a deletion including both
TBP and
PSMB1 in the database of genomic variance (DGV) [
42]. The presence of these deletions in people with a normal or unrelated phenotype makes it unlikely that
TBP or
PSMB1 play a major role in the terminal 6q deletion phenotype.
DLL1
The third distally located HI-gene of interest for the common 6q terminal deletion phenotype is
DLL1. Recently, Fischer-Zirnsak et al. [
29] described 14 patients with pathogenic heterozygous variants of
DLL1 and one patient with a deletion including
DLL1. These patients presented with hypotonia, scoliosis and a neurodevelopmental phenotype including variable brain abnormalities (ventriculomegaly/hydrocephaly, corpus callosum abnormality and cortical dysplasia), seizures and autism spectrum disorder. This phenotype was registered in OMIM as neurodevelopmental disorder with nonspecific brain abnormalities and with or without seizures (NEDBAS, OMIM#618709). Additionally, the following clinical characteristics were reported in at least one patient with a heterozygous
DLL1 variant: PNH, large cisterna magna, strabismus, feeding problems, sleep apnoea, recurrent infections, hemivertebrae, sacral dimple, joint hypermobility, ataxia and hyperactivity [
29]. As all these clinical characteristics are also seen in our terminal 6q deletion cohorts, it is very likely that haploinsufficiency of
DLL1 makes an important contribution to the common terminal 6q deletion phenotype. Lesieur-Sebellin et al. characterised the features detected by prenatal ultrasound in 22 foetuses with terminal 6q deletions and pointed out the gene
DLL1 as the major contributor to the detected phenotype [
43].
In 2005, Eash et al. [
1] reported a patient (Eash_1) with the smallest terminal deletion seen thus far, 390 kb, which only included the HI-genes
TBP and
PSMB1. This patient’s phenotype was comparable to the common terminal 6q deletion phenotype we describe and included microcephaly, brain abnormalities (corpus callosum abnormality, hydrocephaly), seizures, vertebral abnormalities, hypotonia and developmental delay (Tables
2,
3 and Additional file
3: Table S3; subgroup T-PSMB1). Most other characteristics of the common terminal 6q deletion phenotype were not reported as being present or absent in the Eash et al. case. The breakpoints of this deletion where defined by BAC and PAC FISH clones at approximately 500 kb intervals [
1]. If the possible maximum size of the deletion is taken into consideration (Fig.
1), the
DLL1 gene could actually be part of the deletion. We tried contacting the authors about additional genetic studies performed for this individual, but without success. Considering the overlapping phenotype and the ambiguities in breakpoint definition, we expect that
DLL1 is also part of, or influenced by, the deletion in this case.
DLL1 codes for a ligand of the Notch receptor. The Notch signalling developmental pathway is involved in cell-to-cell communication and cell patterning and differentiation. Notch signalling plays a role in the development of multiple organs and tissues, including the somite-derived organs, nervous system, heart, vasculature, haematopoietic system, cochlea and pancreas [
44]. In our cohorts, we did not clearly see any abnormalities for the four latter organ systems, but the nervous system, somite-derived organs and heart were affected.
In mice, delayed expression of
Dll1 leads to premature cell differentiation, resulting in a smaller brain and fused vertebrae and ribs [
45]. This is in line with the variable brain abnormalities, microcephaly and abnormalities of the vertebrae seen in our cohorts and in the heterozygous
DLL1 variant patients reported by Fischer-Zirnsak et al. [
29].
CHDs were not reported in the patients by Fischer-Zirnsak et al. [
29] and also not seen in our T-DLL1 subgroup, so the effect on heart development may not be fully penetrant. We did find CHDs in larger deletions that include
DLL1. In these patients,
DLL1 seems the most likely candidate gene to cause CHDs given its role in the Notch pathway and reported pathogenic variants in
NOTCH1 in patients with a CHD [
46]. During heart development, Notch signalling plays a crucial role in the formation of the membranous walls of the atrial and ventricular chambers and of the outflow tract [
47]. Interruption in Notch signalling could explain the CHDs in our cohorts: ASDs, VSDs and a coarctation of the aorta in one individual. Bu et al. reported a patient with an ASD and persistent left superior vena cava with a heterozygous
DLL1 variant that was classified as likely pathogenic. This
DLL1 variant patient also had a cleft palate, but no further phenotype information was given [
48]. A cleft palate was also seen in two of our patients, although these two had the largest terminal deletions of our cohort.
Other genes
Besides
DLL1, other genes were also previously proposed to play a role in the terminal 6q deletion phenotype, especially in larger terminal deletions that display additional clinical characteristics. Below, we briefly summarise these in the context of our findings.Several genes have been linked to brain abnormalities.
ERMARD (Endoplasmic Reticulum Membrane-Associated RNA Degradation Protein, MIM*615,532 (also known as C6orf70)) might be involved in PNH, since Conti et al. described nine patients with a deletion including
ERMARD and one patient with a heterozygous missense variant and PNH [
7]. Unfortunately, we do not have information on the prevalence of PNH for our parent cohort. One patient in our literature cohort did present with PNH, but her deletion did not include
ERMARD [
28].
ERMARD is also not a predicted HI-gene (%HI: 84.86, pLI: 0.00). Based on this information and the fact that the deletion patients presenting with PNH of Conti et al. all had a deletion that also included
DLL1, it remains unclear whether haploinsufficiency of
DLL1 or
ERMARD, or both, can cause PNH.
Peddibhotla et al. suggested two other genes that may be involved in structural brain abnormalities:
THBS2 (Thrombospondin II, MIM*188,061) and
PHF10 (Phd Finger Protein 10, MIM*613,069). These genes were deleted in all seven of their terminal 6q deletion patients [
9]. However, both genes are not predicted HI-genes, and no pathogenic variants causing structural brain abnormalities in humans have been described thus far.
Lastly,
QKI has been linked to brain abnormalities because it plays a role in myelination by regulating several myelin-specific genes [
49]. Five individuals in our terminal 6q deletion cohort presented with delayed myelination, and all have a
QKI deletion (Additional file
2: Fig. S3). One individual with an interstitial 6q26 deletion (Id073) also presents with delayed myelination, but
QKI is not part of her deletion, and 32 out of 37 patients with a deletion of
QKI did not have delayed myelination. Likewise, Backx et al. reported a woman with a reciprocal balanced translocation t(5;6)(q23.1;q26) disrupting the
QKI gene, resulting in 50% reduced QKI expression, who did not present with myelination problems [
50]. Thus, although it is likely that
QKI plays a role in myelination, there seems to be incomplete penetrance of this clinical feature.
Vertebral abnormalities are part of the common terminal 6q deletion phenotype, and
TBXT (T-Box Transcription Factor T, MIM*601,397) is suggested to play a role in the aetiology of hemivertebrae [
18]. An identical missense variant in this gene was identified in three unrelated patients with congenital vertebral malformations. This variant was proposed to increase the risk of congenital vertebral malformations, but not sufficiently on its own [
18,
51]. Nevertheless,
TBXT was not deleted in all patients with hemivertebrae in our cohort. We therefore think this phenotype is more likely linked to
DLL1.
Dental problems, including abnormal morphology and reduced number of teeth, are also part of the common terminal 6q deletion phenotype. The gene
SMOC2 is related to dental problems in carriers of pathogenic homozygous variants, including oligodontia, microdontia and abnormally shaped teeth [
52‐
54]. However, no pathogenic heterozygous
SMOC2 variants have been identified thus far, and
SMOC2 was not deleted in all individuals with dental problems in our cohort (Additional File
2: Fig. S4).
CHDs were seen in 12 patients with terminal deletions including at least
AFDN (deletions larger than 2.7 Mb). Next to
DLL1, these deletions also included the gene
THBS2. In two large CHD cohort studies, two variants of unknown significance in
THBS2 were found. One patient presented with a tetralogy of Fallot [
55]. The other patient presented with subaortic stenosis, bicuspid aortic valve, mitral valve stenosis and regurgitation and a coarctation of the aorta [
56]. In contrast, our CHD patients mainly presented with septal defects. Since there is no further proof for the role of
THBS2 in CHD, and all terminal 6q deletions also include the more likely candidate gene
DLL1, we regard the contribution of
THBS2 to CHD in the 6q deletion phenotype as less likely.
Recent work showed that
QKI also plays a role in cardiovascular development and function in mice and might be involved in cardiomyopathies and cardiovascular disease in humans [
57]. One (1/42) of our patients (Id185, aged 1.5 years) with a deletion including
QKI was reported to have hypertrophic cardiomyopathy. Further research is needed to investigate whether there is a relation between cardiomyopathies and a deletion of
QKI and thus whether individuals with a
QKI deletion need to be screened for cardiomyopathy.
Almost all the individuals in our cohort with a deletion including
DLL1 had developmental delay. Besides
DLL1,
QKI seems to mark a tipping point in the extent of developmental delay. Normal development was seen in a couple of individuals without a deletion of
QKI, whereas severe developmental delay was only seen in individuals with a deletion including
QKI (Additional file
2: Fig. S2, Table
3). The woman with a reciprocal balanced translocation t(5;6)(q23.1;q26) disrupting the
QKI gene reported by Backx et al. also presented with borderline developmental delay [
50].
QKI probably has an additive effect on the level of developmental delay next to the deletion of
DLL1, which on its own can lead to moderate developmental delay in small (500 kb) deletions [
28].
A range of behavioural problems was seen throughout the whole group of terminal deletions, with information available for all 34 individuals from the parent cohort but only 8 of 51 literature cases, foetuses excluded (Additional file
3: Table S3). Self-harming behaviour was seen significantly (Fisher’s Exact Test
p = 0.03) more often in the larger terminal deletions. Fischer et al. reported autism spectrum disorder as part of the syndrome linked to
DLL1 haploinsufficiency. In their cohort, 5 out of their 13
DLL1 variant patients and their one
DLL1 deletion patient had autism spectrum disorder [
29]. In our cohorts, however, autistic behaviour was present in only 4/33 individuals with a terminal deletion including
DLL1. However, autistic behaviour was also seen in 3/5 individuals with an interstitial 6q26 deletion that did not include
DLL1, suggesting that
DLL1 haploinsufficiency is not the only cause for autistic behaviour.
The individuals with terminal 6q deletions included in our cohort were grouped based on the number of predicted HI-genes involved in their deletion. However, only two HI-genes, DLL1 and QKI, could be linked to some of the observed clinical characteristics. This is reflected in the phenotypic differences between the patients with deletions smaller and larger than 7.1 Mb.
Limitations of the study and suggestions for future research
Recruiting parents via social media and collecting phenotypes directly from the parents via the online Chromosome 6 Questionnaire resulted in an extensive dataset for the parent cohort. However, not all clinical characteristics were addressed in literature, resulting in missing data. For example, it was only known for one out of ten (10%) individuals in the T-PDE10A subgroup if they had sleeping problems, while this was known for 14 out of 20 (70%) in the T-PRKN subgroup. The main difference between these subgroups was the ratio of parent cohort versus literature cohort cases, which was 1:9 for T-PDE10a and 15:5 for T-PRKN. In another study, we investigated the availability of data on specific phenotype information in literature case reports compared to data collected directly from parents in the terminal 6q deletion cohort presented here. We show that we collected significantly more data from parents, for almost all phenotypic features, in comparison to the literature [
60].
A risk of not reporting data on absent phenotype features is that incorrect conclusions can be drawn. For example, balance problems were often reported in the whole terminal 6q deletion cohort (25/35, 71%), but vestibular and/or cerebellar dysfunction was only reported as a cause of these balance problems in individuals with a deletion including PRKN (larger than 7.9 Mb) (5/16, 31%). It remains unknown whether vestibular and/or cerebellar dysfunctions only cause balance problems in patients with larger deletions, or if the causes of balance problems in those with smaller deletions were simply not investigated or reported.
Our phenotype data was collected directly from parents, which might raise questions on the quality of the data. However, in our data consistency study we show that phenotype data collected directly from parents is highly consistent with data extracted from the medical files on the same individual [
60].
Another topic for which information is still very limited is the natural history of disease and adulthood. For two individuals with a terminal 6q deletion smaller than 2.7 Mb, it is known that they did not have developmental delay and could live independently. For another two individuals with a deletion larger than 7.9 Mb (including PRKN), it is known that they could not (fully) take care of themselves and could not live independently. However, for most individuals who have reached adulthood, information on their level of performance is very limited and it was unclear whether they could live independently. Follow-up on adults with terminal 6q deletions is needed to give insight into adult functioning and development of new clinical features at older ages.
Our recommendations for investigations now focus on two groups—deletions < 7.1 Mb and deletions > 7.1 Mb including the gene QKI—since this was a clear tipping point in the reported phenotypes. However, we cannot be absolutely sure that these clinical characteristics are only seen in the larger deletions and will never be reported in individuals with smaller deletions. Therefore we have tried to be cautious in our recommendations for investigations. Since the phenotype can be very variable, it is important to assess each patient on an individual level. Nonetheless, the general differences we have reported can be helpful in counselling (expecting) parents. For the future, we hope to be able to give more detailed recommendations based on deletion sizes, but this is only possible if detailed information for an even larger study population is collected.