CLIA Laboratory Testing for Facioscapulohumeral Dystrophy

Discussion

This retrospective analysis describes data from the University of Iowa, an academic institution offering CLIA-certified laboratory testing, for the purpose of reporting the relative frequencies of FSHD1 and FSHD2 and other genetic characteristics among individuals with a clinical rationale for undergoing FSHD testing. This is an unbiased sample of FSHD diagnostic testing. The results appear to be representative for the US population, as demonstrated by the geographic distribution shown in table 2. In our sample set of nearly 1,600 patients, a genetic diagnosis of FSHD was reached in 44% of cases. Of these, 94.5% (n = 664) had a genetic diagnosis of FSHD1 and 5.5% (n = 39) a diagnosis of FSHD2. The D4Z4 repeat arrays were significantly smaller in tests consistent with FSHD1. We compared our data to FSHD registry–derived data that reported the distribution of D4Z4 repeat sizes among 74 patients with a genetic diagnosis of FSHD1.⁴ The Iowa Molecular Pathology Laboratory data summarized in table 3 by allele size is similar to the published data: 1–3 D4Z4 repeats (Iowa 16.4% vs registry 12.2%), 4–6 D4Z4 repeats (Iowa 48.2% vs registry 50%), 7–10 repeats (Iowa 35.4% vs registry 37.8%).

Genetic testing for FSHD has been refined over many years. FSHD was first linked to the 4q35 region in 1990.¹⁵ Two years later, in 1992, clinical genetic testing for FSHD became a possibility with the identification of a deleted 4q35 as the genetic basis for the disease.^16,–,18 The deletion was determined to be within a region of 3.3 kb D4Z4 repeats that could be evaluated by Southern blots of EcoRI restriction enzyme digests. However, the interpretation of deleted 4q35 alleles was complicated by the high homology with D4Z4 repeats on chromosome 10 (10q26 locus). The next step in diagnostic refinement was the introduction of a EcoRI/BlnI double digestion step to distinguish the D4Z4 repeats derived from 4q35 and 10q26, as the chromosome 10 D4Z4 repeats have BlnI sites not present in 4q35 D4Z4 repeats.^19,20 XapI digestion further refined testing by digesting 4q35-derived D4Z4 by not those from chromosome 10.²¹ Sequencing DNA distal to the D4Z4 repeats led to the recognition of “A” and “B” haplotypes; abnormally short D4Z4 repeats on 4q35 without an A haplotype (4q35B alleles) were not associated with FSHD and A/B haplotyping was added to the diagnostic workflow.^22,23 In 2012, pathogenic variants in the SMCHD1 gene were found in families with an FSHD phenotype but without a 4q35 deletion.¹ The importance of methylation status was also recognized at this time. Sequencing of SMCHD1 was added to diagnostic testing, and analysis of methylation could be used to screen for those at risk for an SMCHD1 variant or to verify the pathogenicity of variants identified by sequencing the SMCHD1 gene.²⁴ This evolution resulted in the diagnostic workflow utilized by the Iowa Molecular Pathology Laboratory illustrated in figure 1.

Although the genetic understanding of FSHD has developed and grown over the years, the pathophysiology of FSHD is still not fully understood. The muscle wasting found in FSHD is associated with derepression of a gene distal to the D4Z4 repeat array known as the double homeobox 4 (DUX4) retrogene. Derepression of DUX4 is caused by chromatin relaxation due to either D4Z4 contraction (FSHD1) or hypomethylation due to a pathogenic SMCHD1 variant (FSHD2).²⁵ DUX4 codes for a transcription factor that is normally expressed in small amounts in early embryological development and is found in the testis and pluripotent cells but is silenced in adult somatic tissue.^7,26 Even small amounts of DUX4 in postnatal humans is toxic to skeletal muscle and results in apoptosis through a cascade of events including disruption of RNA metabolism and induction of oxidative stress.^27,28 The pathophysiology of FSHD and DUX4 has been difficult to study due to the retrogene's transient and sporadic misexpression in cells.²⁷

FSHD genetic testing is complex and one of the factors that complicates it is the high degree of homology between 4q35 and 10q26. D4Z4 repeat sizes at 10q26 that are similar to those of pathologic 4q35 alleles do not result in disease manifestations.²⁰ Thus, distinguishing 4q35 from 10q26 alleles is critical to the accurate interpretation of test results. The high homology found between 4q35 and 10q26 also increases the likelihood of translocations between the 2 loci. Translocation events between 4q and 10q are fairly common within the normal population and have been reported in about 20% of individuals.^19,29 Remarkably consistent with earlier published data, we found evidence of translocation in 20.9% of individuals tested at Iowa.

A second factor contributing to complexity of FSHD genetic testing is the presence of somatic mosaicism, first described in FSHD in 1993.^{6,17,30,–,32} We identified 23 cases with somatic mosaicism; they had 5–7 alleles from 4q35 and 10q26. This mosaicism can partially explain the clinical heterogeneity seen among individuals with FSHD.³³ Individuals who are somatic mosaics often have a milder phenotype or are asymptomatic.^10,33 Of the 10%–30% of FSHD cases that are de novo, about 50% display mosaicism either in a parent or (less commonly) in the proband.^10,34

The homology with chromosome 10q, somatic mosaicism, and additional genetic variations, such as p13E-11 probe binding site deletions, have led some to suggest that FSHD testing is inconclusive in about 20% of cases.^6,31 In our experience, a genetic diagnosis (FSHD1, FSHD2, FSHD1+FSHD2, or non-FSHD1,2) was reached in 96% of cases that underwent testing (n = 1,640). The testing was incomplete in 2.8% of these cases (n = 46), while testing results were borderline FSHD1 in 1.2% (n = 20).

While a D4Z4 repeat array of ≤10 units is consistent with a diagnosis of FSHD1, FSHD2 cases also have a relatively short D4Z4 repeat array. FSHD2 cases typically have a D4Z4 repeat array ranging from 8 to 20 units with the average of 12 units, but longer repeat arrays are reported.³⁵ In our study, 27 FSHD2 cases had 12 to 19 repeats and 12 cases had 21 or more repeats, with the highest being 31. One explanation that has been proposed for an FSHD2 diagnosis with a longer D4Z4 sequence is D4Z4 duplications allowing for DUX4 derepression.³⁵ However, review of the Southern blots at Iowa revealed that none of the 12 cases with more than 21 repeats had evidence of D4Z4 duplications.

One of the FSHD2 cases in our Iowa testing cohort has a 2.5 mb 18p deletion encompassing the SMCHD1 gene (supplemental table 1; data available from Dryad, doi.org/10.5061/dryad.f1vhhmgv3). Hemizygosity for SMCHD1 has previously been shown to result in D4Z4 chromatin relaxation characteristic of FSHD2 in 2 FSHD2 families with a 1.2 MB 18p deletion encompassing SMCHD.³⁶ SMCHD1 hemizygosity also resulted in D4Z4 hypomethylation in 72 of 82 cases with 18p deletions encompassing the SMCHD1 gene, and the authors proposed that a permissive 4q35A allele with 11–16 D4Z4 repeats might increase the risk of clinical FSHD in these individuals.³⁶ The single case with an 18p deletion in our study has 18 D4Z4 repeats on a permissive 4q35A allele and a D4Z4 methylation value of 14%.

We identified 4 cases with genetic characteristics of both FSHD1 and FSHD2 (table 1); when included in the total number of FSHD cases, they represent 0.6%. SMCHD1 was first recognized as a disease modifier in FSHD1 cases that were more severely affected than predicted by their D4Z4 allele size (8–10 repeats).^25,37 In a more recent report, 7 of 19 patients with FSHD with 9–10 D4Z4 repeats also had a pathogenic SMCHD1 variant, while none of the patients with shorter repeat arrays did. The authors suggested that the 8–10 D4Z4 unit alleles have reduced penetrance and FSHD types 1 and 2 overlap in this repeat range, creating a disease continuum.^8,38,39 Our cohort includes 2 cases with this genetic signature and 2 additional cases with 11 D4Z4 repeats and an SMCHD1 variant (see supplemental table 1; data available from Dryad, doi.org/10.5061/dryad.f1vhhmgv3). Due to the previously discussed inconsistencies of allele sizing by Southern blotting, these latter 2 cases were included in our FSHD1+FSHD2 category (n = 4).

Our results and those of others suggest that SMCHD1 is not the only gene associated with FSHD2. Although we identified 98 tests with 4q35 hypomethylation and a permissive 4qA allele, only 39 had pathogenic SMCHD1 variants resulting in a FSHD2 diagnosis (not including the FSHD1+FSHD2 cases). Of the remaining 59 cases, 9 had short EcoRI fragments resulting in a FSHD1 diagnosis. The other 50 cases were given non-FSHD1,2 genetic diagnoses in the current analysis (table 1); among these 50 cases were 20 individuals with between 11 and 20 D4Z4 repeats who are perhaps at higher risk for FSHD2. It is possible that some of these cases have a pathogenic variant in a non–protein coding region of the SMCHD1 gene that is not interrogated in the sequencing.²⁵ Recent research at Leiden University Medical Center shows that intronic variants resulting in new splice sites can be found in about 2% of the FSHD2 population.²⁵ Using this as guideline, one of our hypomethylated cases without a FSHD1 or FSHD2 diagnosis is predicted to carry an intronic variant. Alternatively, pathogenic variants in another gene or genes may affect 4q35 methylation status.⁸ One such gene, DNMT3B (DNA methyltransferase 3B), is found on chromosome 20q11.^2,40 Heterozygous autosomal dominant pathogenic variants of DNMT3B have been found in patients with clinical characteristics of FSHD and a relatively short D4Z4 repeat array (9–13 repeats) on a 4q35A allele with hypomethylation but no pathogenic SMCHD1 variant.² Homozygous pathogenic variants in this gene have been implicated in ICF syndrome (immunodeficiency, centromeric instability, and facial dysmorphism).⁴¹ Another possible FSHD2 gene is LRIF1 (ligand-dependent nuclear receptor-interacting factor 1). A homozygous duplication variant of LRIF1 was found in 1 patient with a clinical phenotype of FSHD. The patient had a 4q35A allele with 13 D4Z4 repeats and a D4Z4 methylation of 15%.⁴² Current research involving DNMT3B, LRIF1, and other possible modifiers of FSHD will contribute to understanding the molecular pathogenesis and may contribute to refined diagnostic testing.

The traditionally accepted and preferred method of FSHD testing relies on Southern blot analysis of restriction enzyme digests separated by pulsed field gel electrophoresis, as was used for the tests in this analysis. Southern blotting is labor-intensive, only estimates restriction fragment size, and requires large amounts of high molecular weight DNA.^31,32 Other testing methods have been explored to overcome these obstacles. One alternative to Southern blotting is molecular combing (MC), where combed DNA is fluorescently hybridized to regions of the D4Z4 repeat region on both chromosome 4 and chromosome 10.³¹ MC reduces the number of steps required for Southern blotting and is reported to provide resolution of cases termed undecided or borderline or where a genetic cause has not been determined by Southern blot.^6,31,43 Another alternative to Southern blotting is single-molecule optical mapping. DNA molecules are fluorescently tagged either by nick label DNA repair or direct insertion at sequence specific recognition sites. The DNA is electrophoresed and imaged through BioNano technology allowing for D4Z4 sizing and haplotyping.^32,44 Similar to MC, single-molecule optical mapping reduces the steps needed to diagnosis FSHD. Finally, bisulfite sequencing to measure the methylation level of the D4Z4 repeat region has been used to distinguish among FSHD1, FSHD2, and normal individuals.⁴⁵ Using a series of sequencing steps, 4qA alleles are identified and the level of methylation is used to assign a diagnostic category.⁴⁵ The University of Iowa Molecular Pathology Laboratory is in the process of transitioning to optical mapping technology to diagnose FSHD1; optical mapping will be used in combination with methylation assays and SMCHD1 sequencing to diagnose FSHD2.

This report of a large clinical sample of patients from across the United States who underwent testing for FSHD provides unique insights into the genetic epidemiology of this complex disease. Our results highlight some of the complexity involved in genetic testing that will have to be addressed as clinical laboratories move away from Southern blotting and toward more rapid and less expensive approaches to molecular diagnosis. The rapid growth in understanding the pathophysiology of FSHD in recent years is expected to lead to novel approaches to therapy and the population-based data presented here, showing subtype frequencies, will facilitate the design of future clinical trials.