To the Editor: Family-based whole-exome sequencing has proven to be an effective diagnostic strategy for the identification of causative variants in individuals with intellectual disability (ID) and congenital malformations (CM). De novo mutations play a major role in ID/CM and it is estimated that they are responsible for up to 40% of cases in non-consanguineous populations.1 Most whole-exome sequencing filtering strategies applied in laboratories worldwide are focused on de novo, X-linked and recessive inheritance. However, in their article “Imprinting: The Achilles Heel of Trio-Based Exome Sequencing,” Aten et al.2 recently highlighted the importance of taking into account mutations in imprinted genes as a cause of ID. They described the difficulties they confronted in the identification of the causative variant in a large family with several affected members. The family was studied in parallel in two independent centers using different diagnostic approaches. A trio-based approach was used for one part of the family and single-exome sequencing for another member; both failed to identify the pathogenic mutation. Only when the analysis of the family pedigree showed that all affected individuals were linked through their fathers was the causative variant, a paternally inherited frameshift mutation in MAGEL2, identified. Truncating mutations in the paternally expressed allele of MAGEL2, located in the imprinted 15q11q13 Prader–Willi syndrome region, are responsible for Schaaf–Yang syndrome.3 An additional pitfall when applying the common filtering strategies may be encountered if parental mosaicism is present. Recent unpublished data from the Deciphering Developmental Disorders study estimate that around 2% of pathogenic de novo mutations in children are mosaic in parental tissues.

We report a seven-year-old boy initially referred to our clinic at 3 years of age for clinical evaluation. He is the second child of a non-consanguineous healthy couple with no family history of note. His elder brother is healthy. Pregnancy was uneventful (normal fetal movements) and delivery was at term by C-section due to breech presentation. The birth weight was 2,830 g (15th centile). In the neonatal period, hypotonia, poor suck and scarce spontaneous movements were noted. On examination at 3 years of age, the patient showed marked growth delay (height: −4 s.d., weight: −3 s.d., occipitofrontal circumference: −2.5 s.d.). He had dry skin and an abnormal hair growth pattern. Dysmorphic features consisted of dolicocephaly, low-set ears, a broad nasal root, a deep philtrum and widely spaced teeth. He had mild contractures of both knees, tapering digits with camptodactyly of fingers 2 to 5, and poorly developed palmar creases. He had male genitalia with a hypoplastic penis and scrotum. Bilateral cryptorchidism had been surgically corrected. His psychomotor development was markedly delayed: he was unable to walk independently, speech was almost absent, and he was just able to comprehend simple orders. He had suffered two seizures (normal magnetic resonance imaging and electroencephalogram) and was undergoing gastroenterology and endocrine follow-up due to chronic constipation and recurrent hypoglycemias of unknown origin. No clinical diagnosis could be established. Initial genetic testing included a karyotype, a custom-designed 60 K oligonucleotide array (KaryoArray v3.0) and the CytoSNP-850 K Beadchip (Illumina), all with normal results. Subsequent trio exome analysis using a de novo filtering approach revealed an apparently de novo heterozygous frameshift mutation in MAGEL2 (NM_019066.4):c.1996dupC (p.Gln666fs) previously described in other patients with Schaaf–Yang syndrome. Sanger sequencing validation confirmed the variant in the proband and enabled us to identify the same frameshift variant in a mosaic state in his father. In fact, reanalysis of the parental data visualizing the BAM file detected this variant in two out of 50 reads. This finding was confirmed in the father using a custom next-generation sequencing clinical panel containing 1,253 genes involved in intellectual disability, autism spectrum disorders, and other common genetic disorders (Clinical panel V1.0). The variant was found again in 2 of 50 reads, and the mutant allele fraction was estimated to be around 4%. In this case, the low somatic mosaicism detected in the father allowed the identification of the variant in the proband when filtering by de novo variants because the genotyping tool used to obtain the variants (GATK HaplotypeCaller) establishes a cut-off of 10% of reads to call a de novo allele. At the time of diagnosis the father was expecting a baby with a different partner. The recurrence risk of 1 to 2% in the case of a de novo mutation due to the possibility of parental gonadal mosaicism substantially increased to 10 to 20% in view of MAGEL2 being an imprinted gene and the presence of mosaicism in paternal tissues.

In conclusion, we report on another case of molecularly confirmed Schaaf–Yang syndrome, which—to the best of our knowledge—is the first report of a MAGEL2 mutation inherited from a mosaic father. We also highlight the difficulties encountered in analyzing single patients when the causal variant is located in an imprinted gene or a parental mosaicism is present. Generally, trio-based analysis is a good approach in cases of unexplained ID/CM when a de novo variant is suspected; however, as illustrated by this case, a specific analysis pipeline for imprinted genes that does not include inheritance filtering should also be considered. To this extent de novo, inherited variants, and variants present in a mosaic state in the parents will be detected. In cases where the causative mutation is located in a non-imprinted gene, low parental mosaicism may be missed when applying a de novo filtering strategy and might be subsequently suspected in the direct visualization of the trio BAM file and/or during Sanger sequencing confirmation. Meanwhile, a pathogenic dominant mutation in a proband may be overlooked if parental mosaicism is present in a higher percentage and a common de novo filtering strategy is applied. Therefore, we also recommend a specific bioinformatic algorithm for imprinted genes, and raise awareness of parental mosaicism as a possible pitfall in routine de novo analysis and its implications in genetic counseling.