In this study, we systematically investigated the clinical subtypes and molecular pathologies of 77 Taiwanese EB patients through WES, Sanger sequencing, TEM, and IF studies. As the initial tool for mutational analysis, WES correctly identified disease-associated variants in 37 of the 38 families (97.4%).
In recent years, next-generation sequencing (NGS) has become a first-line tool for mutational analysis for many genodermatoses, including EB. Such an approach has been proven feasible by several studies utilizing NGS with EB-specific multigene panels [11,12,13,14,15], yielding a diagnostic rate of 83.5–97.7%, depending on the study population and the gene panel used. Our results are consistent with such previous studies.
Of the 38 families in which WES was used as a first-line tool for mutational analysis, one family (the family of PT41) required the additional use of RNA sequencing to determine the pathogenic mutations. In this case, we used RNA sequencing to determine whether each disease-associated variant (all with CADD scores < 10) found by WES affected splicing. The results helped establish the diagnosis of AR-EB-pruriginosa, despite atypical IF results with conventional LH7.2 antibody for C7. Similar approaches utilizing RNA sequencing to study the transcriptomic changes of specific variants found by WES had also been used in mutational analysis for EB [16, 17]. Vahidnezhad et al. also used RNA sequencing to prove coexistence of both recessive simplex and junctional EB phenotypes in one patient with homozygous mutations in both EXPH5 and COL17A1 [16].
Interestingly, the results of transcriptomic analysis of c.5820 + 4A > G in PT41 with RNA sequencing and RT-PCR plus Sanger sequencing were discordant, with RNA sequencing showing intron 70-retained transcripts and PCR plus Sanger sequencing showing exon 70-skipped transcripts caused by c.5820 + 4A > G. Similar results were also seen in the mutational analysis of PT70, who shared this mutation. We believe that the higher amplification efficiency of exon 70-skipped transcripts due to the primers used in RT-PCR might have resulted in this disparity, and hence, the "biased" results of RT-PCR plus Sanger sequencing. The advantage of sequencing splice transcripts without preference, along with the ability to uncover deep intronic, silent, and synonymous exonic variants often overlooked by WES [18], makes RNA sequencing a useful additional technique to WES in the mutational analysis of genodermatoses.
Our study included 19 EBS patients (24.7%), six JEB patients (7.8%), and 52 DEB patients (67.5%). In contrast, most studies on the prevalence of EB using national EB registries showed EBS to be the most common subtype, accounting for over or close to 50% of all EB cases [19,20,21]. Published studies also revealed that EBS as a share of EB is highest in Northern Ireland (88%), followed by Scotland (58%), Australia (56%), the United States (54%), Japan (51%), and Norway (43%), with the lowest occurring in Croatia (16%) [20, 22]. Considering this, EBS seems under-represented and DEB over-represented in the current study. Such deviation from the world's EB epidemiology data could be explained by the fact that our study, which utilized data of patients who came to a tertiary hospital, probably selected for a more severe EB population. EBS, in which a large percentage being mild or self-improving, have a less severe phenotype in general, making patients with EBS less likely to seek medical attention and genetic counseling than DEB patients. Several other studies conducted in similar hospital settings also showed underrepresentation of EBS and overrepresentation of DEB [12, 13, 23].
In our study, only PT25 (JEB) and PT77 (AR-DEB-severe) had disease caused by homozygous mutations. This is expected because the rate of consanguineous marriages is low in Taiwan. In countries where consanguineous marriages are much more common, such as Iran and Kuwait, EB is usually inherited in a recessive mode and mutations are more frequently found at homozygous status. In addition, EB is caused by mutations in genes that usually are more rarely mutated in the disease [14, 24]. Indeed, in addition to the low prevalence of homozygous mutations in Taiwan, our study did not identify recurrent mutations suggestive of common ancestral alleles in our population study, either.
The phenotypes of the 12 patients with EBS caused by KRT5 or KRT14 mutations in our study correlated well with their genotypes. Patients with mutations lying in the highly conserved boundary motif of keratin 5 and keratin 14 demonstrated severe phenotypes; patients with mutations elsewhere had much less severe blistering. All three patients with AD-EBS-severe improved over time, presenting less acute lesions (blisters and erosions) and more hyperpigmentation.
The 52 DEB patients in our study included 34 patients with dominant DEB, 17 patients with recessive DEB, and one patient with severe DEB but unknown genotype. These patients had a total of 33 mutations, including 12 novel mutations. Nine mutations were found in unrelated Taiwanese families; the most frequent ones were p.Gly2043Arg and p.Pro1805Leu, both occurring in three families, the former of which being the most common glycine substitution mutation underlying dominant DEB worldwide [25, 26], leading to both reduced secretion of pro-C7 into the extracellular matrix and increased enzymatic susceptibility of C7 [26].
c.5414C > T (p.Pro1805Leu) in COL7A1 was a novel mutation seemingly specific to Taiwanese EB populations. The substitution of leucine in this mutation occurred on the Y residue of a Gly-X–Y repeat in exon 62. Since the proline at this residue is often hydroxylated to 4-hydroxyproline and involved in stabilization of collagen triple helices, this mutation might disrupt the thermal stability of the triple helices [27, 28]. Interestingly, within the three EB families with this mutation, all heterozygous carriers had a normal phenotype, suggesting a recessive nature of this mutation.
In general, the genotypes in our DEB cohort correlated relatively well with established genotype–phenotype correlations. In most of the AR-DEB-severe patients, the disease was caused by biallelic nonsense, frameshift, and certain splice-site mutations, all resulting in PTCs [29]. However, in two AR-DEB-severe patients (PT70, PT71), the disease was caused by one splice-site mutation causing PTC and the other causing inframe exon skipping. Still, none of the AR-DEB-intermediate patients in our study had biallelic nonsense mutations.
In the patient with severe DEB but unknown genotype (PT62), only one mutation, c.6182G > A (p.Gly2061Glu), was found by Sanger sequencing. Further analysis was not possible because the patient had died of the disease. This patient had an extremely severe phenotype, characterized by extensive blistering, scarring, growth retardation, flexure contractures, and pseudosyndactyly. In addition to p.Gly2061Glu, the patient might have had another recessive glycine substitution mutation, p.Gly2422Glu, based on its presence in the proband’s mother and siblings. Although uncommon, AR-DEB-severe could be caused by missense mutations only; the homozygosity of both c.7705G > C (p.Gly2569Arg) and c.8245G > A (p.Gly2749Arg) caused a severe phenotype in two families [30]. Nevertheless, due to a lack of direct mutational data, it was unknown whether the severe phenotype of PT62 resulted from the compound heterozygosity of the two missense mutations, an unidentified mutation in COL7A1, or other disease modifiers.
Fifteen of the 52 DEB patients (15/52, 28.8%) had EB pruriginosa in this study, including 14 AD-DEB-pruriginosa patients and one AR-DEB-pruriginosa patient. Typically, EB pruriginosa presents with intensely pruritic excoriated nodules, papules, and plaques on the extensor aspects of the extremities, while more generalized lesions are seen in some patients. The disease can be dominant or recessive, but the dominant form is more common [31]. EB pruriginosa is traditionally considered a rare subtype of EB, and the largest series of EB pruriginosa reported to date consisted of eight patients without mutational data [32], while some cases were reported under other names, including pretibial EB [5], a term used in older classifications. The large number of EB-pruriginosa patients reported by Lee et al. from Taiwan [5] and by our group indicates that EB pruriginosa is a relatively common subtype of DEB in Taiwan.
In our study, all 14 patients with AD-DEB-pruriginosa had glycine substitution mutations. This is consistent with a systemic review by Kim et al., which found glycine substitution mutations (52.7%) and in-frame skipping (33.8%) to be the most common mutations underlying EB pruriginosa [33]. It is noteworthy that mutations associated with EB pruriginosa in our studies showed marked inter-familial and intra-familial variations in phenotype. Of the six families in which other subtypes of DEB occurred, the same pathogenic variants resulted in AD-DEB-localized in five (5/6) and AD-DEB-intermediate in two families (2/6). Modifiers that led to a phenotypic difference of the same mutation could be genetic, epigenetic, or environmental, and remain mostly unknown [2].