Background
Recessive dystrophic epidermolysis bullosa (RDEB) is a rare, debilitating autosomal recessive disease caused by biallelic mutations in
COL7A1, the gene encoding type VII collagen (C7) [
1]. C7 is produced by basal keratinocytes and dermal fibroblasts, and is the primary component of anchoring fibrils (AF), specialized extracellular structures originating in the lamina densa that link to dermal collagen fibers to stabilize dermal-epidermal adhesion [
2,
3]. In RDEB, C7 expression is typically abnormal or absent, leading to widespread subepidermal blistering and a constellation of characteristic clinical findings including extensive wounding, scarring, strictures, musculoskeletal contractures, increased risk of aggressive squamous cell carcinoma (SCC) and premature mortality [
1,
4,
5]. There are currently no disease modifying treatments approved for RDEB. In recent years, several pre-clinical studies and clinical trials using gene therapy approaches aimed at correcting the underlying molecular phenotype of RDEB have been initiated [
6,
7]. However, much remains unknown about the long-term outcomes and safety implications of gene therapy in RDEB including the potential risk for insertional mutagenesis and malignancy associated with retroviral vectors [
8], or autoimmunity to genetically-modified cells or full-length C7 [
9].
As previously reported, autologous gene-corrected keratinocytes expressing full-length C7 using a retroviral vector (EB-101, previously named LZRSE-COL7A1 Engineered Autologous Epidermal Sheets [LEAES]) were developed to treat chronic open wounds in patients with severe RDEB in the first gene therapy trial for RDEB (ClinicalTrials.gov, NCT01263379) [
10,
11]. In the preliminary results of this Phase 1/2a open-label study, EB-101 was well-tolerated without any serious related adverse events during short-term follow-up. Full-length C7 expression was observed two years after grafting, demonstrating durable molecular correction of RDEB after treatment. Here, we present comprehensive, long-term (mean 5.9 years, range 4–8 years) efficacy and safety data on seven adults treated with EB-101 for severe RDEB wounds in the largest long-term follow-up study for a dermatologic gene therapy treatment to date.
Discussion
We previously presented safety and wound outcomes for a mean of 2.7 years of follow-up in seven adult subjects with severe RDEB treated with autologous gene-corrected keratinocyte grafts [
10,
11]. We now report clinical safety and efficacy data from a mean of 5.9 years (range 4–8 years) of follow-up in this Phase 1/2a trial. Our findings demonstrate that EB-101 is a safe, effective, and long-term treatment for chronic RDEB wounds.
During up to eight years of follow-up, no serious adverse events related to treatment were observed. Concerns for oncogenesis have previously been raised regarding the use of retroviral vectors for gene therapy following early clinical trials for X-linked severe combined immunodeficiency [
8]. Reassuringly, no participants developed cutaneous or extracutaneous malignancies related to gene therapy and no systemic RCR infections were identified on serial serologic assays during follow-up. Two subjects developed cutaneous SCCs. However, all tumors were distant from graft sites, contained no detectable retroviral genome using PCR analysis when sampled, and arose when both subjects were in their 30 s. This strongly suggests that these malignancies developed due the subjects’ underlying RDEB, given the well-known increased risk of aggressive SCCs in adults with RDEB [
13], and not due to retroviral insertional mutagenesis or oncogenesis associated with EB-101 treatment.
The long-term risk for clinically significant autoimmunity to gene-corrected keratinocyte grafts was also low. No cytotoxic T cells with anti-C7 activity were identified on serial serologic analysis. Transient circulating anti-C7 antibodies were detected in two participants. Tissue-bound antibodies beyond trace staining were detected in four participants, the majority of which were transient or resolved by year one, and no subjects developed any concerning clinical symptoms related to these localized immune reactions. In subject 4, localized immunoreactants were observed at graft sites up to two years after treatment. As previously discussed [
10,
11], this participant was initially negative for pre-existing anti-C7 antibodies during enrollment screening with an immunofluorescence microscopy assay certified by the Clinical Laboratory Improvement Amendments, but was later found to have preexisting anti-C7 antibodies at baseline using a more sensitive Western blot assay. Subject 4 died five years after treatment; however, circulating anti-C7 antibodies were not thought to contribute to his death, as this participant never developed any clinical symptoms or sequelae concerning for a severe systemic immune reaction including fevers, increased generalized blistering outside of his graft sites, or anaphylaxis, and did not require any treatment for this immune response throughout five years of follow-up. Given this experience, however, enrollment criteria for the ongoing Phase 3 randomized controlled trial (ClinicalTrials.gov, NCT04227106) was revised to exclude participants with pre-existing anti-C7 antibodies at baseline as detected by either IIF or the more sensitive method of enzyme-linked immunosorbent assay (ELISA).
Wound healing varied between participants. In serial evaluations of 42 total grafted wounds, we observed that several wound and participant characteristics may affect graft uptake and contribute to poorer wound healing over time, such as the presence of persistent, localized anti-C7 antibodies. In subject 4, persistent expression of tissue-bound anti-C7 antibodies at graft sites may have led to early and ongoing degradation of full-length C7 expressed at grafts, resulting in reduced long-term molecular correction. However, the transient expression of localized antibodies in subjects 1, 2, and 3 did not appear to impede long-term wound healing.
Larger baseline wound size and anatomic location also impacted wound healing. In subject 6, all six grafts were placed contiguously on a large (> 200 cm
2), confluent wound bed on the mid and lower back (Fig.
2D). Starting at year one, subject 6 had < 50% wound healing at most of their graft sites (Fig.
1), which likely represents poor graft uptake and early graft loss due to mechanical trauma to the grafts as this location was more difficult to immobilize and protect from excess friction and pressure during the immediate postoperative period compared to other anatomic locations, such as the extremities. Given the poor graft uptake observed in subject 6, the postoperative care protocol was revised for the ongoing Phase 3 randomized controlled trial (ClinicalTrials.gov, NCT04227106) to optimize graft uptake, including standardizing the required postoperative inpatient admission period to at least one week with strict immobilization of graft sites and extensive graft bandaging, padding, and care by research staff to reduce trauma, pressure, and friction at graft sites.
However, other grafts placed on areas at risk for trauma or friction successfully demonstrated long-term wound healing. Subject 7 received four contiguous grafts on the upper back and posterior shoulder (Fig.
2C), and by year 5, sustained ≥ 75% wound closure at all treated sites (Fig.
1). The success of subject 7’s grafts may be due to the characteristics of her open wounds at baseline which, in contrast to subject 6’s wounds, were discrete (i.e., not confluent) and smaller, with an average open wound size of 25 cm
2. Similarly, subject 2 received a graft on a single wound on the central low back (Fig.
2A) which demonstrated ≥ 75% wound healing from years two onwards; the success of this graft site may also be attributed to its smaller baseline wound area. The long-term wound healing observed on graft sites on the back for subjects 2 and 7 suggests that successful, durable re-epithelialization of areas at risk for trauma and ulceration is possible, though may be influenced by wound-specific factors including smaller baseline wound size. These findings are consistent with prior work which identified that larger RDEB wounds are more difficult to close compared to smaller wounds [
14,
15], and highlight the need for careful consideration of wound characteristics including location and baseline size when selecting wounds for clinical trials.
Wound healing at individual graft sites also varied over time. Subject 2’s wound E, for example, showed < 50% wound healing at month six, but improved to ≥ 50% healing at year one, and demonstrated ≥ 75% healing from year two onwards as noted above. Mechanisms underlying these fluctuations in wound healing within a single graft site are multifactorial, and may include recent trauma, and bacterial colonization or infections [
16]. Nonetheless, the observed, new capacity for these previously chronic open wounds to repeatedly heal after grafting demonstrates that treatment with EB-101 may improve long-term skin durability and wound healing.
Notably, increased wound healing was significantly correlated with sustained reductions in pain and itch. Wounds with improved wound healing were significantly less painful and less pruritic than wounds with poorer wound healing, demonstrating that treatment with gene-corrected keratinocyte grafts is associated with long-term, clinically-significant benefit for patients with RDEB, even in the absence of complete wound closure. As much remains unknown about the natural history and progression of wounds in RDEB [
15,
17], the use of PROs including pain and itch in therapeutics targeting chronic wounds allows investigators to incorporate the participant’s own perspective and experiences of investigational treatments [
18], and to target drug development towards treatments that are clinically meaningful for patients.
These findings have several implications for clinical care, as chronic wounds are a major mediator of disease course in RDEB. The disrupted microenvironment and pathologic remodeling of chronic wounds leads to persistent inflammation and increased risk of bacterial colonization and infection, facilitating the development of sepsis, as well as severe, treatment-refractory anemia, and malnutrition due to increased metabolic demand from impaired wound healing [
16,
19]. This, in turn, can worsen disease trajectory by further impairing wound healing capabilities and physiologic reserve. Critically, aggressive SCCs, which are the leading cause of death in adult RDEB patients, frequently develop at chronic wounds [
13]. Chronic wounds are also associated with poorer psychosocial outcomes [
20] as chronic wounds are significantly larger and more painful than recurrent wounds [
15]. Chronic wounds are also associated with worse quality of life [
21]. Lastly, chronic wounds impose significant financial and time burdens on RDEB patients due to the need for extensive routine wound dressing changes [
19,
22]. Thus, treatment of chronic RDEB wounds may produce many important clinical benefits, including improvements in both disease course and outcomes, and quality of life.
A case report of transgenic epidermal grafting for junctional epidermolysis bullosa (JEB) recently reported long-term outcomes for one patient with JEB who received grafts expressing full-length, corrected
LAMB3, which encodes for laminin-332 [
23,
24]. Large, full-body grafts were placed on extensive wounds, and remained intact after five years. As detailed previously [
6], laminin-332 promotes keratinocyte stem cell maintenance and growth [
25], and directly mediates keratinocyte adhesion [
26]; C7, notably, lacks these qualities, which may have contributed to the more variable durability of
COL7A1-corrected grafts observed for RDEB in the present study. Future study of gene therapies for RDEB must consider these intrinsic characteristics of RDEB biology, including investigating methods to optimize transduction and enrichment of keratinocyte stem cells within grafts to improve long-term graft durability.
Our findings are limited by the small sample size, a common challenge in clinical trials for rare or orphan diseases, such as RDEB. This Phase 1/2a clinical trial focused on long-term safety and clinical efficacy; thus, serial skin biopsies to evaluate molecular expression of full-length C7 within grafts was not performed beyond year two [
11], which limited definitive assessment of long-term molecular correction of C7. Nonetheless, the observed, sustained wound healing and reduction in pain and itch years after grafting suggests that gene-corrected keratinocyte grafts may confer significant long-term benefits, and further evaluation of the specific molecular etiology of these benefits may be warranted. To reduce the risk of autoimmunity and graft rejection, all subjects in this trial were also required to demonstrate sufficient expression of the NC1 domain [
10,
11,
27]. This limits the generalizability of these results, particularly among patients with null
COL7A1 mutations who lack any C7 expression. Another limitation was the categorization of wound healing by IGA rather than quantification of wound healing using photography software as this was the outcome measure recommended by the FDA. Some observed fluctuations in wound healing at a single wound site over time may be attributed to the categorical nature of this scoring methodology; for example, Subject 5’s graft A was scored by IGA as ≥ 75% wound healing at month six, and 50–74% wound healing at year one. When wound healing at this site was evaluated on a continuous scale using the Canfield Vectra 3D photography system (Canfield Scientific, Parsippany, NJ), however, it was assessed as 96% wound healing at month six, and 70% wound healing at year one—a value which is close to the 75% wound healing cutoff used in the categorical IGA methodology. Lastly, this trial was initially designed with a focus on safety outcomes and a limited number of control wounds were selected. Thus, wound pairs were not randomized prior to treatment, and systematic comparisons of control and treated wounds could not be performed.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.