A novel mosaic variant on SMC1A reported in buccal mucosa cells, albeit not in blood, of a patient with Cornelia de Lange–like presentation

Mosaicism in Cornelia de Lange syndrome (CdLS) has been reported in clinically diagnosed CdLS patients with negative molecular testing using blood as the specimen, particularly in the NIPBL gene. Here we report a novel mosaic variant in SMC1A identified in the buccal swab DNA of a patient with a mild CdLS phenotype. Our patient presented with global developmental delay, dysmorphic features, microcephaly, and short stature, with no limb defect. Face2Gene, a digital tool that analyzes facial morphology, demonstrated a 97% match between our patient and the CdLS gestalt. An initial next-generation sequencing (NGS)-based CdLS panel test, including NIPBL, HDAC8, RAD21, SMC1A, and SMC3, completed using DNA isolated from leukocytes, was negative, and subsequent trio exome sequencing was nondiagnostic. The exome identified biallelic variants of uncertain significance in a candidate gene, NSMCE2. In the pursuit of a molecular diagnosis, a second NGS-based CdLS panel test was ordered on a buccal swab specimen and a novel variant, c.793_795delGAG (p.Glu265del) in SMC1A, was detected at 60% mosaicism. Retrospective analysis of the former panel and exome data revealed the SMC1A variant at 4% and 2%, respectively, both far below standard reporting thresholds. Given that mosaicism has been frequently reported in CdLS, we suggest selecting a different tissue for testing in clinically suspected CdLS cases, even after negative molecular results via blood specimen.

emotional skills (71), and cognition and adaptive skills (70 and 70). He had few words and was not toilet trained by the time of his initial genetics evaluation at 3 yr and 8 mo. Patient had normal head tomography and single-nucleotide polymorphism (SNP) microarray. At the initial genetic evaluation at 3 yr and 8 mo old, his height was 89.10 cm (0.39%ile and −2.66 SDS), weight was 14.10kg (20%ile, −0.84 SDS), and head circumference was 46.1 cm, below the third percentile. His features were reminiscent of the Cornelia de Lange phenotype (see Table 1). Using Face2Gene, a facial digital analysis tool, his gestalt score was a strong match for the CdLS (Fig. 1). Thus, a CdLS next-generation sequencing (NGS)-based gene panel (sequencing and copy-number variant [CNV] analysis), including NIPBL, HDAC8, RAD21, SMC1A, and SMC3, was ordered on a blood specimen, and the result was negative, with no variants of interest reported. In the interim, the patient was diagnosed with growth hormone deficiency. His initial growth velocity was 5.8 cm/year (10th percentile), but after initiating growth hormone therapy, velocity improved to 8.9 cm/year (90th percentile). The response to growth hormone has not been studied systematically on CdLS patients; however, it is consistent with the experience reported by de Graaf et al. (2017). When the patient returned to the clinic at the age of 5 yr and 1 mo, his height was 100.2 cm, improved to the third percentile (−2.01 SDS). However, microcephaly persisted and his head circumference was 47.5 cm, remaining below the third percentile. In addition, his penis remained relatively small and the length measured at the 10th percentile. Notably, his motor abilities had A novel mosaic variant on SMC1A C O L D S P R I N G H A R B O R significantly improved, whereas his speech delay persisted, and he only had about 10 words at 5 yr old and had received an autism spectrum disorder diagnosis. Recent assessment in the context of an individualized education plan, at age 6, reported aggression toward others, requiring constant supervision and maximum teacher cues to complete work at school. He is not toilet trained. He participates in a speech-impaired program and uses assisted technology devices, continuing to qualify for in-school speech and occupational therapy. Given his persistent delays, dysmorphic features, and the negative CdLS sequencing panel, a trio exome sequencing was ordered to look for an alternative diagnosis that mimics CdLS. The exome identified two variants of uncertain significance, in trans, in a candidate gene, NSMCE2 (see Table 3). The NSMCE2 gene encodes a small ubiquitin-like modifier (SUMO) ligase, which plays a role in promoting homologous recombination and DNA damage repair (Potts and Yu 2005). Two patients have been reported by Payne et al. with NSMCE2 compound heterozygous frameshift mutations; both presented with primordial dwarfism,

Molecular Case Studies
gonadal failure, and insulin-resistant diabetes, with fatty liver and hypertriglyceridemia. Our patient had some overlapping features with those presented by Payne et al. such as microcephaly and linear growth deficiency. Although liver function and insulin levels on our patient were normal, his fasting lipid profile indicated hypercholesterolemia (total cholesterol 275 mg/dL, reference range <170 mg/dL and LDL cholesterol 201 mg/dL; reference range <110 mg/dL), as well as hypertriglyceridemia (160 mg/dL, reference range <75 mg/dL). But fasting lipid profiles from both parents were normal. We verified with the exome reporting laboratory that there were no reportable variants in known genes causing hyperlipidemia identifiable through exome sequencing. It is uncertain if variants on NSMCE2 contribute to his hyperlipidemia in any way. Despite some overlapping features, there were significant clinical differences between our patient and those described by Payne et al. as our patient has no signs of insulin resistance or lipodystrophy. Moreover, our patient's facial features were very divergent from those described by Payne et al. who displayed prominent midface, a small jaw, acanthosis nigricans, and skin tags.
At a follow-up clinic visit at 6 yr old, considering the relatively high mosaicism rate of up to 30% reported (Krawczynska et al. 2019) in the clinically diagnosed mutation-negative CdLS patients, a CdLS gene panel (sequencing and CNV analysis), including ANKRD11, HDAC8, KMT2A, NIPBL, RAD21, SMC1A, and SMC3, on buccal swab specimens was ordered and identified a novel hemizygous variant, c.793_795delGAG, p.Glu265del, in SMC1A in 60% of sequencing reads (33 of 56 reads), finally provided a definitive molecular diagnosis (Table 2).

TECHNICAL ANALYSIS
Cornelia de Lange NGS-based sequencing and a CNV analysis panel (HDAC8, NIPBL, RAD21, SMC1A, and SMC3) through EGL Genetics Laboratory (EGL) were performed in 2017 with blood as the specimen (please refer to Table 1 in Supplemental Materials for sequencing coverage information). In solution, hybridization of all coding exons within the genes tested was performed on genomic DNA: direct sequencing was done using next-generation short base pair read sequencing. Intronic variants >3 nt from the exon/intron boundary are not reported unless known to be pathogenic.
Trio exome sequencing through GeneDx laboratory was performed in 2018, with blood as the specimen from the proband and saliva as a specimen from both parents (please refer to Tables 2 and 3 in Supplemental Materials for sequencing coverage information). Genomic DNA from the specimen was enriched for coding regions and splice site junctions for most genes on the human genome, sequenced on an Illumina platform, and then filtered and analyzed using their custom analysis tool.
CdLS NGS-based sequencing and a CNV analysis panel (ANKRD11, HDAC8, KMT2A, NIPBL, RAD21, SMC1A, and SMC3) through GeneDx laboratory was performed in 2019 with a buccal swab specimen (please refer to Table 4 in Supplemental Materials for sequencing coverage information). The same methodology as trio exome sequencing is used, except sequencing was limited to the genes listed above. The variant on SMC1A was identified by the CdLS sequencing and CNV analysis. The panel was completed with the buccal swab sample in 2019. Upon discussion with GeneDx, which completed exome sequencing in 2018, the variant was detected at 4% mosaicism in blood, which was below the laboratory's 20% threshold for reporting. EGL, which performed Cornelia de Lange panel sequencing in 2017, disclosed that the variant was present in 2% (4 out of 199) of reads in blood specimen. Again, it is below threshold for reporting.
Variants of the NSMCE2 gene have been described by Payne et al. (2014) in two female patients with primordial dwarfism, severe insulin resistance, microcephaly, small jaw, fatty liver, hypertriglyceridemia, and ovarian failure.

VARIANT INTERPRETATION
For the SMC1A variant, the criteria considered for classification were Cornelia de Lange-like phenotype, de novo variants not observed in large population cohorts (Lek et al. 2016), inframe deletion of one amino acid in a nonrepeat region, in silico analysis (including protein predictors and evolutionary conservation) supports a deleterious effect, and has not been previously published as pathogenic or benign.

SUMMARY
CdLS is a well-known genetic syndrome, recognizable by a distinctive appearance in patients with the classical phenotype, caused by heterozygous pathogenic variants on NIPBL in the majority of cases. In cases of milder phenotypes, facial characteristics can be less striking, limb anomalies are often absent, and cognitive impairment is less pronounced. Mild CdLS is more commonly caused by heterozygous pathogenic variants on SMC3, RAD21, and hemizygous pathogenic variants in HDAC8 or SMC1A. Somatic mosaicism has been frequently reported in the NIPBL gene. These NIPBL mosaic cases have generally correlated with a classic appearance and severe manifestations, suggesting that detected mosaicism does not necessarily ameliorate clinical features in this condition, especially for cases caused by variants in NIPBL (Castronovo et al. 2010;Gervasini et al. 2013). Somatic mosaicism has been reported less often in other genes that produce a milder CdLS phenotype.
To the best of our knowledge, our case is only the second reported CdLS-like patient with somatic mosaicism on SMC1A; this contrasts with high mosaicism on the NIPBL gene, which is well known. The scarcity of mosaic cases reported in the SMC1A gene could be attributed to its milder and less recognizable clinical features, or variable mosaicism levels in different

Molecular Case Studies
tissues might be missed on certain types of specimens, as we saw in our patient. In addition, mosaicism levels can vary over time, even on the same tissue. Ansari et al. (2014) reported SMC1A mosaicism of 50% and 10%, respectively, in two separate saliva samples of the same individual, producing an atypical phenotype with moderate growth retardation. Furthermore, germline mosaicism has been reported in several cases of the NIPBL gene but has also been suggested in a few SMC1A cases. They included a family with two affected daughters with a mild CdLS phenotype and parents with negative testing for the familial variant in blood, as well as a family with two affected daughters and a father with qualitative testing (chromatograph), indicative of low-level somatic mosaicism on SMC1A and no clinical features of the syndrome (Deardorff et al. 2007;Slavin et al. 2012). All of the above present a challenge for molecular testing strategies and genetic counseling in cases of CdLS-like patients but with an initially negative molecular result.
Here we report a case of somatic mosaicism on SMC1A that was detected at 60% in a buccal swab specimen, but detected at only 2% and 4% in two different blood specimens, by retrospectively reanalyzing raw data on previous testing because it is below the reporting threshold. This remarkable difference can be explained by the different origins of DNA in the specimens used. Buccal swabs contain a high proportion of epithelial cells, especially in children, in whom the mean proportion of buccal cells in buccal swabs was 90.3% in a recent study (Theda et al. 2018), whereas leukocytes are the cellular population used in blood samples.
In the current era of clinical genetics, the most cost-effective strategy for diagnosis of syndromes with poorly defined clinical features has become exome sequencing. This approach has the potential to identify new candidate genes in conditions with locus heterogeneity. Our case exemplifies a diagnostic odyssey and illustrates the challenge of somatic mosaicism in clinical practice, as low-level mosaicism can be disregarded by reporting laboratories if below a certain threshold. Our case illustrated that undetected mosaicism is a plausible explanation for unexpected negative molecular results, especially if a blood sample is the only specimen provided for testing. Selection against mutant cells, particularly in lymphocytes, has been suggested as one plausible mechanism and needs to be taken into consideration (Ansari et al. 2014). Fortunately, in CdLS, the computer-assisted facial analysis tool has a high sensitivity for CdLS. A diagnosis of CdLS was within the top five predicted syndromes for 97.9% of cases in a study by LaTorre-Pellicer, and sensitivity was high even for atypical cases and genes other than NIPBL (87.5% for SMC1A) (LaTorre-Pellicer et al. 2020). Thus, a facial diagnostic tool can assist clinicians to a certain degree in deciding on the best test and specimen to send to increase diagnostic yield.
In conclusion, we suggest selecting different tissues for testing if CdLS is clinically suspected, even after negative results are obtained using a blood specimen, as mosaicism has been frequently reported in this condition.

Ethics Statement
ACMG patient consent for use of medical photography to be used in medical publications, including medical journals, textbooks, and electronic publications, was granted in writing by the parent of the patient.
A novel mosaic variant on SMC1A