Functional genotype classification groups distinguish disease severity in recessive dystrophic epidermolysis bullosa

Abstract

Background

Recessive dystrophic epidermolysis bullosa (RDEB) is a genetic disorder caused by pathogenic variants in COL7A1.

Objectives

To determine the association between different COL7A1 variants and clinical disease severity in 236 North American patients with RDEB.

Methods

Published reports or in silico predictions were used to assess the impact of pathogenic variants in COL7A1 on type VII collagen (C7) protein function. Three impact categories were postulated: genotypes that would be likely to cause a low impact on C7 function (splice B/missense, missense/missense); a medium impact [premature termination codon (PTC)/splice B, splice A/splice B, PTC/missense, splice A/missense, splice B/splice B]; and a high impact (PTC/PTC, PTC/splice A, splice A/splice A). Splice A variants are predicted to cause downstream PTCs, while splice B variants cause in-frame exon skipping and are therefore less deleterious.

Results

The severity of functional impact was significantly associated with a history of gastrostomy tube placement, oesophageal dilation, hand surgery, anaemia, renal disease, chronic wounds, diffuse skin involvement and a history of squamous cell carcinoma. The odds of death were 3.5 time higher in the high-impact vs. medium-impact group (95% confidence interval 1.24–8.50; P = 0.02). Patients in the high-impact group had worse clinical outcomes.

Conclusions

Functional genotype categories are a feasible approach to risk-stratify patients based on predicted C7 function.

Recessive dystrophic epidermolysis bullosa (RDEB) is a genetic disorder caused by pathogenic variants in COL7A1. Data from 236 North American patients with RDEB support the categorization of genotypes into low-, medium- and high-impact groups based on the type of DNA mutation. Patients in the high-impact group had worse clinical outcomes.

Linked Article: Lwin Br J Dermatol 2025; 192:794–795.

What is already known about this topic?

• Recessive dystrophic epidermolysis bullosa (RDEB), caused by pathogenic variants in COL7A1, is the most severe subtype of dystrophic epidermolysis bullosa.

• There is a lack of large genotype–phenotype association studies in RDEB.

What does this study add?

• Our data support the categorization of genotypes into low-, medium- and high-impact groups based on the type of DNA mutation.

• Patients with RDEB in the high-impact genotype group have more severe clinical disease and increased odds of dying.

Recessive dystrophic epidermolysis bullosa (RDEB) is a rare genetic blistering disease characterized by variants in COL7A1, which encodes type VII collagen (C7). Lack of functional C7 leads to the loss of anchoring fibrils and epidermal–dermal adhesion,¹ resulting in recurrent or chronic wounds with significant pain and pruritus, and numerous multiorgan complications,^2–4 including aggressive squamous cell carcinoma (SCC), often leading to early mortality.⁵ Patients with RDEB require specialized, frequent and complex medical care, and report a lower quality of life.^6,7

Variants leading to premature termination codons (PTC) in COL7A1 are correlated with severe RDEB phenotypes,^8–10 but biallelic missense variants often exhibit milder skin disease and fewer extracutaneous complications,¹¹ possibly attributed to the relatively milder reduction in expression/function of C7 vs. the complete loss of C7 generally seen in individuals with biallelic PTC variants.¹² The functional impact of splice site and missense variants or heterozygote combinations (PTC/splice or splice/missense) on clinical disease severity is poorly understood. Missense mutations and splice mutations have been shown to lead to reduced C7 expression and function.^9,13

A functional genotype classification based on the type of DNA mutation, categorizing variants into a low-impact genotype group (splice/missense, missense/missense), a medium-impact group (PTC/missense, splice/splice) or a high-impact group (PTC/PTC, PTC/splice) was previously proposed.¹⁴ In this prospective cohort study of 236 patients with RDEB, we sought to establish general genotype–phenotype associations in a larger RDEB cohort and revise the previously published functional genotype classification system incorporating novel subgroups from the splice site category.

Materials and methods

Study design

Participants of any age seen between 1 May 2008 and 1 March 2023 were recruited from 13 EB centres in North America and were enrolled in the Epidermolysis Bullosa Clinical Research Consortium’s (EBCRC) Clinical Characterization and Outcomes Database. All participants were clinically diagnosed with RDEB and had variants in COL7A1, identified by Sanger or next-generation ­sequencing.

Patients were seen at routine clinic visits. Notes from time of entry into the study up to the last clinic note available were reviewed to obtain key clinical outcomes, including age at the most recent clinic visit, type of RDEB-specific skin changes involved, anatomical sites and the presence of chronic wounds (defined as > 6 weeks). Diffuse skin involvement was defined as ≥ 8 body regions affected by skin changes (out of a maximum of 12 body regions), intermediate as 4–7 body regions affected and limited as 1–3 body regions affected. Examples of RDEB skin changes examined in this study included blisters, erosions/ulcers, naevi, milia, scarring and keratoderma. Additional data extracted from clinic notes included a history of gastrostomy tube (G-tube) placement; history of oesophageal dilation; history of hand surgery; presence of hand and foot pseudosyndactyly; microstomia; nail involvement; alopecia; ocular involvement; cardiac disease; renal disease; history of SCC; clinical subtype according to the system proposed by Has et al.;¹⁵ and mortality data, including frequency and cause of death. Mean haemoglobin levels (g dL^–1) were obtained by averaging the three most recent haemoglobin levels, irrespective of whether the patient was on oral or intravenous iron supplements or had received blood transfusions.

Variant classification

Genetic reports were de-identified and reviewed for variant location and notation, and protein change and pathogenic significance were extracted. If applicable, variants were reclassified based on in silico predictions and up-to-date literature, including a review of the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/).

We proposed a novel classification whereby splice site variants were further subclassified according to downstream impact (splice A and splice B subgroups). Splice A variants were predicted to functionally result in a frameshift of the reading frame and a downstream PTC. Point mutation variants affecting splice sites and leading to either exon skipping of 1 of the 11 ‘unskippable’ exons [containing base pair (bp) numbers not divisible by 3]¹⁶ or point mutations leading to the emergence of cryptic splice sites with subsequent out-of-frame integration/deletion of a part of the adjacent intron/exon (bp numbers of integrated/deleted segments not divisible by 3) were reclassified as splice A variants. However, variants that lead to in-frame exon skipping of 1 of the 107 ‘skippable’ exons (containing bp numbers divisible by 3) and variants with published functional studies demonstrating milder impact were classified as splice B variants (Figure S1; see Supporting Information). In-frame variants such as insertion and/or deletion variants that are a multiple of threes or variants that have an equal number of deleted and inserted nucleotides were grouped with missense variants as low-impact variants.

Statistical analysis

Descriptive statistics are provided as mean (SD) or median (range) for continuous variables, and as frequency and percentage for categorical variables. To assess differences between the three genotype groups (high-, medium- and low-impact), a Kruskal–Wallis test was performed for continuous data with a non-normal distribution, and a χ² or Fisher’s exact test for categorical data, as appropriate. Logistical regression was used to estimate the mortality risk of impact variants adjusting for age at the most recent clinic visit as a covariate. All tests were two-sided and P-values < 0.05 were considered to be statistically significant. All analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC, USA).

Results

Cohort characteristics

In total, 236 patients with a clinical and genetic diagnosis of RDEB were included [mean (SD) age at most recent visit 16.3 (11.2) years; range 0–57.8]. The majority of patients were White (n = 157/236; 66.5%) and most were children (n = 137/236; 58.1%); 31.8% (n = 75/236) were < 9 years old. The most common mode of inheritance was a heterozygotic pattern (n = 181/236; 76.7%) (Table 1).

Table 1.

Demographic and genetic characteristics of 236 patients with recessive dystrophic epidermolysis bullosa (RDEB) included in this study

Characteristic	n (%)
Age at most recent visit (years), mean (SD)	16.3 (11.2)
Range	0–57.8
Sex
Female	114 (48.3)
Male	122 (51.7)
Age group (years)
0–9	75 (31.8)
10–17	62 (26.3)
18–24	50 (21.2)
≥ 25	41 (17.4)
Unknown (not answered)	8 (3.4)
Race
Asian	17 (7.2)
Black or African American	9 (3.8)
Middle Eastern or North African	13 (5.5)
Multiracial	4 (1.7)
Native Hawaiian or Pacific Islander	1 (0.4)
White	157 (66.5)
Unknown (not answered)	35 (14.8)
Clinical site location
Cincinnati Children’s Hospital, OH, USA	78 (33.1)
Stanford University Hospital, CA, USA	51 (21.6)
Children’s Hospital Colorado, CO, USA	23 (9.7)
Hospital for Sick Children, Toronto, ON, Canada	19 (8.1)
Northwestern University, IL, USA	15 (6.4)
University of Massachusetts, MA, USA	9 (3.8)
The Children’s Hospital of Philadelphia, PA, USA	9 (3.8)
University of California San Diego, CA, USA	8 (3.4)
Phoenix Children’s Hospital, AZ, USA	6 (2.5)
Columbia University, NY, USA	5 (2.1)
Dell Children’s Medical Center, TX, USA	4 (1.7)
Instituto Nacional de Pediatría, Mexico City, Mexico	4 (1.7)
Sainte-Justine Hospital, QC, Canada	5 (2.1)
RDEB subtypea
Severe	130 (55.1)
Intermediate	73 (30.9)
Localized	11 (4.7)
Inversa	8 (3.4)
Self-improving	3 (1.3)
Unknown (not answered)	11 (4.7)
Reclassified functional variant impactb
High-impact group	123 (52.1)
PTC/PTC	88 (37.3)
PTC/splice A	27 (11.4)
Splice A/splice A	8 (3.4)
Medium-impact group	86 (36.4)
PTC/splice B	16 (6.8)
Splice A/splice B	3 (1.3)
PTC/missensec	50 (21.2)
Splice A/missensec	14 (5.9)
Splice B/splice B	3 (1.3)
Low-impact group	27 (11.4)
Splice B/missensec	7 (3.0)
Missense/missense	20 (8.5)
Variant classification (n = 472)
Pathogenic/likely pathogenic	444 (94.1)
Variant of unknown significance	15 (3.2)
Not previously reportedd	13 (2.8)
Variant location in COL7A1
NC1/NC1	41 (17.4)
NC1/TH	63 (26.7)
NC1/NC2	5 (2.1)
TH/TH	115 (48.7)
TH/NC2	9 (3.8)
NC2/NC2	2 (0.8)
Unknown	1 (0.4)
Inheritance pattern
Heterozygous	181 (76.7)
Homozygous	55 (23.3)

Data are presented as n (%) unless otherwise stated. NC1, noncollagenous domain 1; NC2, noncollagenous domain 2; PTC, premature termination codon; TH, triple helix domain. ^aClassification was based on that proposed by Has et al.^15bImpact was determined by reclassifying variant consequences reported in genetic test reports by predicted or known effect on type VII collagen expression and function. ^cIn-frame variants (four c.5720_5721delGAinsAT variants and one c.155_157del variant) were grouped with missense. ^dAs noted on genetic testing reports and ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), 7 of these 20 variants have been reported in the literature (see Table 3).

Genotype characteristics

The majority of de-identified genetic reports were obtained from GeneDx (https://www.genedx.com/). There were 472 variants, with 247 unique ones identified. The most frequent variant location in COL7A1 was within the triple helix region, which is the largest domain.⁹ The most common variants were c.6527dupC within exon 80 (5.9%), followed by c.7485+5G>A within intron 98 (3.4%) and c.425A>G within exon 3 (2.9%). Table 2 shows the most common recurrent COL7A1 variants identified in this cohort. The most common variant locations were in exon 80, intron 98 and exon 3. Figure 1 shows variant frequency by location in COL7A1. The majority of variants were classified as pathogenic/likely pathogenic (92.6%). We identified 20 variants not previously reported in ClinVar, 12 of which have been reported in the literature, including in a recent publication that characterized COL7A1 variations in a large cohort of patients with dystrophic EB in South Asia (Table 3).¹⁷ After reclassification of splice site variants into splice A and splice B, genotypes were grouped into three categories, depending on the type of DNA mutation: low-impact (splice B/missense, missense/missense; n = 27); medium-impact (PTC/splice B, splice A/splice B, PTC/missense, splice A/missense, splice B/splice B; n = 86); and high-impact (PTC/PTC, PTC/splice A, splice A/splice A; n = 123).

Table 2.

Most common^a recurrent COL7A1 variants identified in 247 unique alleles from 236 patients with recessive dystrophic epidermolysis bullosa included in this study

Variant DNA change	Amino acid	Exon/intron location	Domain	Functional consequence	n (%)b
c.6527dupC	p.G2177WfsX113	Exon 80	TH	PTC	28 (5.9)
c.7485 + 5G>A	IVS98+5G>A	Intron 98	TH	Splice B	16 (3.4)
c.425A>G	p.K142R	Exon 3	NC1	Splice A	14 (2.9)c
c.8440C>T	p.R2814X	Exon 114	NC2	PTC	8 (1.7)
c.2005C>T	p.R669X	Exon 15	NC1	PTC	8 (1.7)
c.7012C>T	p.2338X	Exon 90	TH	PTC	8 (1.7)
c.1732C>T	p.578X	Exon 13	NC1	PTC	7 (1.5)
c.497dupA	p.V168GfsX12	Exon 4	NC1	PTC	7 (1.5)c
c.356_357delCA	p.T119RfsX9	Exon 3	NC1	PTC	6 (1.3)c
c.3840delC	p.G1281VfsX44	Exon 31	TH	PTC	6 (1.3)c
c.553C>T	p.R185X	Exon 54	NC1	PTC	6 (1.3)
c.1637-1G>A	IVS12-1G>A	Intron 12	NC1	Splice A	6 (1.3)c
c.5048_5051dupGAAA	p.N1684KfsX13	Exon 54	TH	PTC	5 (1.1)
c.5047C>T	p.R1683X	Exon 54	TH	PTC	5 (1.1)c
c.6187C>T	p.R2063W	Exon 74	TH	Missense	5 (1.1)
c.6501G>A	p.P2167P	Exon 79	TH	Splice A	5 (1.1)
c.682+1G>A	IV5+1G>A	Intron 5	NC1	Splice A	5 (1.1)c

NC1, noncollagenous domain 1; NC2, noncollagenous domain 2; PTC, premature termination codon; TH, triple helix domain. ^aVariants comprising > 1% of the study cohort (n ≥ 5) are listed here. ^bPercentage of variants obtained by dividing the number of times the variants occur in the dataset by the total number of variants (472 total variants – 236 × 2 variants for each participant). ^cRecurrent COL7A1 variants that were exclusively found in White patients.

Figure 1.

Stacked bar chart representing variant frequency (from a total of 472 variants) by exons/introns of COL7A1 from 236 patients with recessive dystrophic epidermolysis bullosa (RDEB). Blue = premature termination codon (PTC) variant; orange = splice A; green = splice B; yellow = missense. Exons/introns shaded in yellow (1–28), light blue (29–111) and red (112–118) represent the noncollagenous domain 1, triple helix and noncollagenous domain 2 domains of COL7A1. UTR, untranslated region.

Table 3.

Variants in our cohort of patients with recessive dystrophic epidermolysis bullosa (RDEB) not identified in ClinVar^a

Variant	Amino acid	Exon/intron	Location	Consequence	Reference
Unreported	IVS85-1_2delAG	Intron 85	TH	Splice site	b
c.1358-1G>A	IVS10-1G>A	Intron 10	NC1	Splice site	Lucky31
c.7929+1G>A	IVS106+1G>A	Intron 106	TH	Splice site	Kern32
c.8226+1G>A	IVS110+1G>A	Intron 110	TH	Splice site	b
c.2315-2delA	IVS17-2delA	Intron 17	NC1	Splice site	b
c.5533-2_-1delAG	IVS64-2_-1delAG	Intron 64	TH	Splice site	Woodley33
c.4182_4188dup7	p.A1397WfsX7	Exon 36	TH	PTC	Woodley33
c.2017G>T	p.E673X	Exon 15	NC1	PTC	Marinkovich34
Unreported	p.G1580D	Exon 49	TH	Missense	b
c.5009G>A	p.G1670D	Exon 54	TH	Missense	b
c.5345G>T	p.G1782V	Exon 61	TH	Missense	Kern13
c.5417G>A	p.G1806E	Exon 62	TH	Missense	b
c.7424G>A	p.G2475E	Exon 97	TH	Missense	b
c.5942delA	p.K1981RfsX23	Exon 72	TH	PTC	b
c.7856delT	p.M2619RfsX12	Exon 105	TH	PTC	Yu8
c.4754delCc	p.P1585LfsX125	Exon 49	TH	PTC	b
c.4353C>T	p.R1451X	Exon 40	TH	PTC	b
c.1010delA	Unreported	Exon 8	NC1	PTC	b
c.4818+3_4818+6dup	Unreported	Exon 50	TH	Splice site	b

NC1, noncollagenous domain 1; NC2, noncollagenous domain 2; PTC, premature termination codon; TH, triple helix domain. ^aAs noted in genetic testing reports and ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/). ^bNovel variant not previously reported in the literature. ^cVariant found in both alleles of a patient with homozygous RDEB.

Positive association of genotype group and extracutaneous disease

Overall, patients in the high-impact group were more likely to have more severe clinical disease than those in the medium- or low-impact groups (Table 4). Patients in the high-impact group had a significantly higher prevalence of G-tube placement (high-impact vs. medium-impact vs. low-impact: 52.5% vs. 38.8% vs. 28.0%; P = 0.04), oesophageal dilations (high-impact vs. medium-impact vs. low-­impact: 63.3% vs. 45.4% vs. 55.6%; P = 0.03), hand surgery (high-impact vs. medium-impact vs. low-impact: 36.8% vs. 14.5% vs. 26.9%; P = 0.002) and renal disease (high-­impact vs. medium-impact vs. low-impact: 8.9% vs. 1.2% vs. 3.7%; P = 0.03). Patients in the high-impact group had a significantly higher prevalence of treatment for anaemia (high-impact vs. medium-impact vs. low-impact: 21.1% vs. 11.6% vs. 3.7%; P = 0.04), as well as lower haemoglobin levels (high-impact vs. medium-impact vs. low-impact: 9.8 g dL^–1 vs. 11.6 g dL^–1 vs. 11.9 g dL^–1; P < 0.001). The high-­impact group was also associated with a significantly higher prevalence of the severe subtype of RDEB and a lower prevalence of the localized subtype (P < 0.001).

Table 4.

Associations between the impact of reclassified functional variants and clinical manifestations^a in 236 patients with recessive dystrophic epidermolysis bullosa (RDEB) included in this study

Characteristic	All patients (n = 236)	Low impact (n = 27)	Medium impact (n = 86)	High impact (n = 123)	P-valueb
Age at last clinic visit (years), mean (SD)	16.3 (11.2)	20.0 (16.0)	15.2 (11.5)	16.2 (9.5)	0.27
Distribution (n = 201)c					0.03
Limited	46 (22.9)	2 (8.3)	25 (33.3)	19 (18.6)
Intermediate	59 (29.4)	11 (45.8)	16 (21.3)	32 (31.4)
Diffuse	96 (47.8)	11 (45.8)	34 (45.3)	51 (50.0)
Presence of chronic woundsd	135 (68.5)	12 (54.6)	42 (59.2)	81 (77.9)	0.01
Presence of hand pseudosyndactyly	180 (76.3)	22 (81.5)	62 (72.1)	96 (78.1)	0.48
History of hand surgery	62 (27.4)	7 (26.9)	12 (14.5)	43 (36.8)	0.002
Age at first hand surgery (years), median (range)	8.0 (2.0–21.1)	9.3 (4.5–21.1)	10.2 (5.5–16.3)	7.4 (2.0–16.8)	0.04
Presence of foot pseudosyndactyly	157 (66.5)	21 (77.8)	53 (61.6)	83 (67.5)	0.28
Absence of nails	145 (61.4)	16 (59.3)	47 (54.7)	82 (66.7)	0.21
Hair involvement	185 (78.4)	21 (77.8)	69 (80.2)	95 (77.2)	0.87
History of G-tube placement	103 (44.8)	7 (28.0)	33 (38.8)	63 (52.5)	0.03
History of oesophageal dilation	130 (55.8)	15 (55.6)	39 (45.4)	76 (63.3)	0.03
Age at first oesophageal dilation (years), median (range)	6.5 (1.7–33.0)	10.0 (4.0–22.0)	5.4 (1.8–33.0)	6.8 (1.7–32.1)	0.16
History of anaemia treatment	37 (15.7)	1 (3.7)	10 (11.6)	26 (21.1)	0.04
Haemoglobin level (g dL–1), median (range)e	10.8 (5.2–20.0)	11.9 (7.9–16.5)	11.6 (5.2–15.2)	9.8 (6.1–20.0)	< 0.001
History of IV iron transfusion	50 (21.2)	4 (14.8)	15 (17.4)	31 (25.2)	0.31
No. of iron transfusionsf	2 (1–6)	6 (1–6)	1 (1–6)	2 (1–6)	0.12
Presence of ocular manifestations	152 (64.4)	17 (63.0)	51 (59.3)	84 (68.3)	0.40
Mouth opening (mm), median (range)g	18 (5–40)	20 (8–40)	20.3 (6–35)	14 (5–40)	0.39
Presence of cardiac manifestations	26 (11.0)	1 (3.7)	10 (11.6)	15 (12.2)	0.53
Presence of renal manifestations	13 (5.5)	1 (3.7)	1 (1.2)	11 (8.9)	0.03
History of SCC	21 (9.2)	2 (7.4)	3 (3.6)	16 (13.5)	0.04
Age at first SCC (years), median (range)	24.2 (15.4–40.1)	28.8 (23.5–34.1)	28.0 (28.0–28.0)	23.6 (15.4–40.1)	0.51
RDEB-related deaths	32 (13.7)	0 (0)	7 (8.1)	25 (20.8)	0.002
Age at death (years), median (range)	18.3 (0–39.1)	NA	12.6 (2.0– 21.6)	18.7 (0–39.1)	0.24
RDEB subtype (n = 225)h					< 0.001
Severe	130 (55.1)	10 (37.0)	31 (37.8)	89 (76.7)
Intermediate	73 (30.9)	15 (55.6)	32 (39.0)	26 (22.4)
Localizedi	22 (9.3)	2 (7.4)	19 (23.2)	1 (0.9)

Data are presented as n (%) unless otherwise stated. G-tube, gastrostomy tube; IV, intravenous; NA, not applicable; SCC, squamous cell carcinoma. ^aData exclude unreported data missing for each variable. ^bP-values < 0.05 are considered statistically significant (bold). ^cA participant was assigned to one of the three distribution categories depending on the number of reported body regions affected by blisters and erosions/ulcers. Limited (1–3 areas involved), intermediate (4–7 areas involved) or diffuse (≥ 8 areas involved) categories were ascertained from disease involvement of ≥ 1 of the following areas: face, scalp, neck, chest, abdomen, back, axillae, arms, hands, genitalia/buttocks, legs, feet. ^dDefined as RDEB wounds that have been continuously open for > 6 weeks. ^eHaemoglobin levels averaged over the previous 1–3 visits. ^fNumber of IV iron transfusions within the past 6 months. ^gA patient with a mouth opening < 30 mm was considered to have microstomia. ^hClassification based on that proposed by Has et al.¹⁵ⁱThe localized subset includes the classification categories outlined in Has et al.:¹⁵ ‘localized’, ‘inversa’ and ‘self-improving’ (all preceding categories display milder phenotypes than severe or intermediate).

Death occurred more frequently in the high-impact group (high-impact vs. medium-impact vs. low-impact: 20.8% vs. 8.1% vs. 0.0%; P = 0.002). For every 10-year increase in age, the odds of dying were 3.25 times higher [95% confidence interval (CI) 1.24–8.50; P = 0.02] for patients in the high-­impact group vs. the medium-impact group. After adjusting for participant age, there was a 1.49-fold (95% CI 1.03–2.17; P = 0.04) higher odds of death in the high-impact group vs. the medium-impact group. It was not feasible to compare these data with those for the low-impact group as no deaths were reported in this group at the time of data collection.

Further modelling was performed to assess whether the trends were solely driven by PTC/PTC genotypes in the high impact group; therefore, the above analysis was re-run with PTC/PTC genotypes excluded (Table S1; see Supporting Information). The overall trend for history of SCC, history of hand surgery (P = 0.002) and history of oesophageal dilation was persistent. The overall trend for a history of G-tube placement, anaemia, the presence of chronic wounds and early death persisted in a severity-dependent manner (from the high- to medium- to low-impact groups).

Positive association of genotype group and skin disease

Patients in the high-impact group had a higher incidence of chronic wound presence (high-impact vs. medium-impact vs. low-impact: 77.9% vs. 59.2% vs. 54.6%; P = 0.01), a more diffuse distribution of skin changes (affecting ≥ 8 body regions; P = 0.03) and a greater history of SCC (high-impact vs. medium-impact vs. low-impact: 13.5% vs. 3.6% vs. 7.4%; P = 0.04) (Table 4).

Discussion

Patients in the high-impact group generally had more severe cutaneous and extracutaneous clinical outcomes. Patients in the high-impact group reported a higher prevalence of chronic wounds. The clinical significance of chronic wounds was demonstrated in a prospective observational study of patients with RDEB, in which chronic open wounds were reported as more painful than recurrent wounds.¹⁸ Moreover, the functional classification proposed is in general agreement with the subtype classification put forward by Has et al.,¹⁵ with increasing functional severity generally demonstrating a more severe RDEB subtype. Furthermore, patients in the high-impact group had a higher prevalence of a history of G-tube placement and oesophageal dilation. Patients with RDEB have increased protein loss and higher caloric requirements due to greater cutaneous involvement and feeding difficulties.¹⁹ Anaemia is a common and known complication of RDEB.²⁰ Overall, patients in the high-impact group had lower haemoglobin levels. This finding is of clinical relevance as patients in the high-impact group may require more frequent surveillance and more frequent transfusions. The general trends found in our study are consistent with the existing literature and with a recent study by Gupta et al.,¹⁷ which found a higher frequency of extracutaneous manifestations (especially oesophageal involvement), deformities and more diffuse skin involvement in South Asian patients with PTCs vs. other variations.

Patients in the high-impact group having more severe disease is consistent with previous observations that PTC mutations lead to a severe phenotype.¹⁰ PTC variants include point mutations that result in a stop codon (UAA, UAG or UGA) in the associated mRNA transcript or frameshift mutations resulting from insertions or deletions of nucleotides that are not a multiple of threes. As it is well known that the NC2 domain is necessary for anchoring fibril assembly, a distal PTC disrupting NC2 function or a proximal PTC mutation in the NC1 coding region would each be expected to ablate anchoring fibril assembly/function, and would each be expected to have the same severe impact on dermal–epidermal cohesion.²¹

A functional approach to characterizing COL7A1 variants is critical to understanding the extensive phenotypic variability observed in RDEB.²² One could argue that the trends seen in the high-impact group (PTC/PTC, PTC/splice A, splice A/splice A) are driven disproportionately by two PTC genotypes (PTC/PTC); however, positive associations with some key clinical manifestations persisted when the analysis was re-run with the exclusion of PTC/PTC.

A study of Chinese patients with EB concluded that most patients with RDEB had variants that were more randomly distributed within COL7A1 compared with patients with dominant dystrophic epidermolysis bullosa (DDEB).²³ In contrast, although direct comparisons cannot be made due to the lack of data from patients with DDEB in this study, our data seem to suggest the presence of hotspot variants or founder variants in certain populations, particularly in the exon 80 region of COL7A1. This aligns with a study on gene editing in RDEB, which focused on a highly prevalent variant in the Spanish population within exon 80.²⁴ Gardella et al. noted the predominance of variants affecting exons 73–74 in a cohort of Italian patients with RDEB, corroborating the findings in our study of a significant peak within those exons (Figure 1).⁹ They identified six recurrent variants in their cohort, including c.425A>G within exon 3, which was also over-represented in our and other cohorts.²⁵ This variant has been seen mainly in patients of European ancestry. Gupta et al. did not find hotspot variants in a study of 68 patients with dystrophic epidermolysis bullosa in South Asia.¹⁷ These findings have implications for future CRISPR/Cas9-centred gene-correction trials to focus on hotspots for gene editing.²⁶

Genetic analysis can be complicated by variants that are termed ‘variants of uncertain significance’ (VUS; i.e. variants that have incomplete or conflicting data on pathogenicity).²⁷ Given that patients harbouring one VUS had a second, pathogenic variant, and that – phenotypically – all patients probably had RDEB (according to the patients’ treating EBCRC dermatologist), it is suggested that all VUS variants in this cohort are most likely to be pathogenic, as two pathogenic variants are required for the diagnosis of RDEB. Additionally, upon review of each of the patient’s clinical VUS data, we verified that they all had symptoms and signs consistent with RDEB. Thus, we corroborated the hypothesis that these VUS are likely to be disease-causing. We have identified 20 ‘not previously reported’ variants that are similarly deemed pathogenic. We defined a variant as ‘not previously reported’ if the variant in question had not been previously reported in GeneDx, in ClinVar or in the literature. No participants in our cohort had more than one VUS or two variants ‘not previously reported’.

The limitations of our study include limited data on the history of SCC and renal disease. Also, age at disease onset was not available, which precluded performing age adjustments for these variables during analysis. Given the mostly paediatric nature of the study cohort [mean (SD) age 16.3 (11.2) years] and the inability to adjust age-sensitive clinical manifestations such as the presence of SCC for age, the classification system proposed does not capture the true prevalence of clinical manifestations in the RDEB population and thus cannot be used to guide clinical care practices without further validation by a larger (and older) study cohort. Another age-related limitation pertains to the inability to determine whether RDEB in infant participants will eventually subside; this subtype of RDEB is known as the self-improving subtype or ‘bullous dermolysis of the newborn’, and is known to resolve typically within 24 months of life.⁴ While only 9 of 236 (3.8%) patients were last seen when they were < 24 months of age, we were incapable of determining whether their RDEB phenotype would persist or resolve. This study was also limited by the comparatively small number of participants included in the low-impact group and thus introducing possible selection bias; patients with severe RDEB may be more likely to present to dedicated specialty centres, given their more severe clinical disease. The relatively smaller cohort size of the low- and medium-impact groups may also indicate that the clinical severity between those included medium- and low-impact groups does not appear to correlate well with genotype in several aspects. For example, the low-impact group had higher rates of oesophageal dilation, hand surgery, renal disease and a history of SCC vs. the medium-impact group. Full-body skin examinations were not completed for all patients. This may explain the lack of trend in terms of the presence of RDEB-specific skin changes that did not yield a statistically significant correlation with the functional impact groups in our study. Further studies should assess clinical disease severity using validated epidermolysis bullosa (EB) scoring instruments and correlate it with the functional genotype impact group, as well as account for concurrent medications that may have confounded the associations. There is considerable clinical overlap between patients with dominant and recessive dystrophic EB. Despite expert opinion, there was no ascertainment of the diagnosis by a second EB specialist, and one cannot always reliably distinguish a patient with recessive EB from one with a dominant EB by phenotype alone, especially in patients with one or more VUS mutations. Another limitation includes the potential confounding of iron supplementation on the haemoglobin levels of the participants, especially those included in the higher-impact group; the inability to determine whether the collated haemoglobin levels represent the patients’ levels before or after iron supplementation precluded an accurate assessment of anaemia across the functional impact groups. Although one can establish general genotype–­phenotype correlation rules, exceptions to these rules exist and – in some cases – severe RDEB genotypes can present a relatively mild phenotype and vice versa.^28,29 This may be due to nutritional status, wound care regimen and management of wound infections, as well as genetic factors such as accompanying gene mutations outside of COL7A1.³⁰ While splice A variants are expected to have a more deleterious impact on phenotype than splice B variants, alternative splicing may occur for splice A variants and partly salvage C7 expression, making it less impactful on protein availability than PTC variants.²⁸ Ultimately, functional studies are the gold standard to determine the true impact of a variant on protein expression.

RDEB represents a severe form of EB and characterization of clinical milestones and disease progression is paramount to help establish genotype–phenotype associations. This multicentre study of 236 North American patients with RDEB has refined a previously published functional genotype classification that now incorporates a more specific delineation of splice site variants based on downstream impact. Furthermore, the data support the notion that patients in the high-impact group (PTC/PTC, PTC/splice A, splice A/splice A) were associated with more unfavourable clinical outcomes. Future correlation studies using this refined functional genotype framework should employ more diverse cohorts with RDEB and other EB subtypes.

Appendix 1.

Complete list of author affiliations

Pirunthan Pathmarajah,¹ Edward Eid,¹ Jaron Nazaroff,¹ Jodi So,¹ Vaishali Mittal,¹ Nicki Harris,¹ Shufeng Li,¹ Anne W Lucky,² Emily S Gorell,² Kathleen G Peoples,³ Elena Pope,⁴ Irene Lara-Corrales,⁴ Amy S Paller,^5,6 Karen Wiss,^7,8 Marissa J Perman,⁹ Lawrence F Eichenfield,^10,11 Moise L Levy,^12,13 Kimberly D Morel,¹⁴ Maria T García-Romero,¹⁵ Catherine C McCuaig,¹⁶ Melissa Saber,¹⁷ M Peter Marinkovich,^1,18 Anthony Oro,¹ Anna L Bruckner^19,20 and Jean Y Tang¹

¹Department of Dermatology, Stanford University School of Medicine, Stanford, CA, USA

²Division of Dermatology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA

³Children’s Hospital Colorado, Aurora, CO, USA

⁴Division of Dermatology, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada

⁵Department of Dermatology, Northwestern University Feinberg School of Medicine, Chicago, IL, USA

⁶Department of Paediatrics, Northwestern University Feinberg School of Medicine, Chicago, IL, USA

⁷Department of Dermatology, UMass Chan Medical School, Worcester, MA, USA

⁸Department of Pediatrics, UMass Chan Medical School, Worcester, MA, USA

⁹Section of Dermatology, Children’s Hospital of Philadelphia and Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA

¹⁰Department of Dermatology, University of California San Diego and Rady Children’s Hospital San Diego, San Diego, CA, USA

¹¹Department of Pediatrics, University of California San Diego and Rady Children’s Hospital San Diego, San Diego, CA, USA

¹²Department of Pediatrics, Dell Medical School, University of Texas at Austin, Dell Children’s Medical Center, Austin, TX, USA

¹³Department of Internal Medicine (Dermatology), Dell Medical School, University of Texas at Austin, Dell Children’s Medical Center, Austin, TX, USA

¹⁴Department of Dermatology, Columbia University Irving Medical Center, New York, NY, USA

¹⁵Department of Dermatology, Instituto Nacional de Pediatría, Mexico City, Mexico

¹⁶Division of Dermatology, Sainte-Justine University Hospital Center, University of Montreal, Montreal, QC, Canada

¹⁷Department of Dermatology, University of Montreal, Montreal, QC, Canada

¹⁸Dermatology, Veterans Affairs Medical Center, Palo Alto, CA, USA

¹⁹Department of Dermatology, University of Colorado School of Medicine, Aurora, CO, USA

²⁰Department of Pediatrics, University of Colorado School of Medicine, Aurora, CO, USA

Funding sources

This study was supported by a grant from the Epidermolysis Bullosa Research Partnership. This work is also supported by National Institutes of Health/National Center for Advancing Translational Sciences Colorado (NCATSC) Clinical and Translational Science Awards (CTSA) grant no. UM1 TR004399.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Ethics statement

This prospective, observational study was approved by the Stanford University Institutional Review Board (IRB #20314). The University of Colorado Denver was the central data coordination centre. All 13 collaborating centres have previously received local Institutional Review Board approval for data contribution to the Epidermolysis Bullosa Clinical Research Consortium’s Clinical Characterization and Outcomes Database.

Patient consent

Age-appropriate informed consent was obtained from all participants.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's website.