Safety and Wound Outcomes Following Genetically Corrected Autologous Epidermal Grafts in Patients With Recessive Dystrophic Epidermolysis Bullosa

Key Points

Question  Can gene therapy be safely used to restore type VII collagen expression to recessive dystrophic epidermolysis bullosa (RDEB) wounds?

Findings  In this phase 1 clinical trial, autologous RDEB keratinocytes transduced with a retroviral vector containing full-length human COL7A1 were assembled into epidermal sheet grafts and applied to 6 wounds on 4 patients. All grafts were well tolerated without serious adverse events. Some but not all grafts showed improved wound healing as well as type VII collagen expression in anchoring fibrils at the dermal-epidermal basement membrane.

Meaning  Autologous type VII collagen gene–corrected grafts may have potential in the molecular and clinical correction of RDEB skin.

Importance  Recessive dystrophic epidermolysis bullosa (RDEB) is a devastating, often fatal, inherited blistering disorder caused by mutations in the COL7A1 gene encoding type VII collagen. Support and palliation are the only current therapies.

Objective  To evaluate the safety and wound outcomes following genetically corrected autologous epidermal grafts in patients with RDEB.

Design, Setting, and Participants  Single-center phase 1 clinical trial conducted in the United States of 4 patients with severe RDEB with a measured area of wounds suitable for grafting of at least 100 cm². Patients with undetectable type VII collagen keratinocyte expression were excluded.

Interventions  Autologous keratinocytes isolated from biopsy samples collected from 4 patients with RDEB were transduced with good manufacturing practice–grade retrovirus carrying full-length human COL7A1 and assembled into epidermal sheet grafts. Type VII collagen gene–corrected grafts (approximately 35 cm²) were transplanted onto 6 wounds in each of the patients (n = 24 grafts).

Main Outcomes and Measures  The primary safety outcomes were recombination competent retrovirus, cancer, and autoimmune reaction. Molecular correction was assessed as type VII collagen expression measured by immunofluorescence and immunoelectron microscopy. Wound healing was assessed using serial photographs taken at 3, 6, and 12 months after grafting.

Results  The 4 patients (mean age, 23 years [range, 18-32 years]) were all male with an estimated body surface area affected with RDEB of 4% to 30%. All 24 grafts were well tolerated without serious adverse events. Type VII collagen expression at the dermal-epidermal junction was demonstrated on the graft sites by immunofluorescence microscopy in 9 of 10 biopsy samples (90%) at 3 months, in 8 of 12 samples (66%) at 6 months, and in 5 of 12 samples (42%) at 12 months, including correct type VII collagen localization to anchoring fibrils. Wounds with recombinant type VII collagen graft sites displayed 75% or greater healing at 3 months (21 intact graft sites of 24 wound sites; 87%), 6 months (16/24; 67%), and 12 months (12/24; 50%) compared with baseline wound sites.

Conclusions and Relevance  In this preliminary study of 4 patients with RDEB, there was wound healing in some type VII collagen gene–corrected grafts, but the response was variable among patients and among grafted sites and generally declined over 1 year. Long-term follow-up is necessary for these patients, and controlled trials are needed with a broader range of patients to better understand the potential long-term efficacy of genetically corrected autologous epidermal grafts.

Trial Registration  clinicaltrials.gov Identifier: NCT01263379

Recessive dystrophic epidermolysis bullosa (RDEB) is a severe inherited blistering disease characterized by painful erosions, debilitating scarring, and aggressive squamous cell carcinoma during early adulthood. In 1999, the prevalence and incidence of RDEB were estimated to be 0.92 and 2.04, respectively, per 1 million live births.¹ Specific therapy is lacking and care is purely supportive and palliative.²

Recessive dystrophic epidermolysis bullosa is caused by loss of function mutations in COL7A1 (NCBI Accession M96984 and D13694), the gene coding for type VII collagen. Type VII collagen localizes to anchoring fibrils, which stabilize the dermal-epidermal basement membrane. Type VII collagen contains a noncollagenous domain 1, a central collagenous domain, and a noncollagenous domain 2 that catalyze the anchoring fibril assembly.³

Allogeneic fibroblast injection,⁴ bone marrow transplantation,⁵ mesenchymal cells,^6,7 and skin substitutes^8-10 have shown variable efficacy and safety for patients with RDEB. Gene therapy holds potential for the treatment of monogenic diseases similar to RDEB; however, concerns have been raised regarding the viral insertion of cancer-promoting therapeutic genes in patients with severe combined immunodeficiency and Wiskott-Aldrich syndrome.^11,12 Cutaneous gene therapy is advantageous because superficial graft sites can be clinically evaluated and assessed for neoplasms.

Long-term type VII collagen expression in regenerated human RDEB epidermal xenografts was previously established using a Moloney leukemia virus–derived retroviral vector (LZRSE) containing the COL7A1 gene (LZRSE-COL7A1).¹³ A retrovirus was chosen because of the large size of the COL7A1 gene (9 kb) and the high level of viral titers achieved. Moreover, LZRSE-modified epidermis previously transplanted to 1 patient with junctional epidermolysis bullosa showed a long duration of expression without safety issues.¹⁴ In light of this background, this study sought to evaluate ex vivo gene transfer of recombinant type VII collagen grafts onto wounds in patients with RDEB.

Single-center phase 1 clinical trial of patients with RDEB grafted with type VII collagen gene–corrected autologous keratinocytes that were transduced with a retroviral vector containing the full-length COL7A1 coding sequence. This open-label trial was approved by the Recombinant DNA Committee in 2007 and later by the US Food and Drug Administration and the Stanford University institutional review board (protocol appears in Supplement 1).

Patients were aged 18 years or older and clinically diagnosed with RDEB with a measured area of wounds suitable for grafting of at least 100 cm². Patients were selected from review of Stanford University records and other clinician referrals using a screening protocol¹⁵ approved by the Stanford University institutional review board. The presence of RDEB was confirmed via genetic testing (GeneDx). Type VII collagen noncollagenous domain 1 expression was assessed by Western blot of conditioned patient keratinocyte medium and indirect immunofluorescence microscopy of skin biopsy samples. The type VII collagen and anchoring fibril deficiencies in skin biopsy samples were confirmed by immunofluorescence microscopy and immunoelectron microscopy using the LH24 antibody specific to noncollagenous domain 2 (eFigure 1 in Supplement 2).

Circulating and tissue-bound immunoreactants (IgG, IgA, IgM, and C3) were analyzed via immunofluorescence microscopy by placing a serum sample on a primate esophagus and via direct immunofluorescence of patient biopsy sections, respectively (eMethods in Supplement 2), and by Western blot. Patients with medical complications, including human immunodeficiency virus, hepatitis, systemic infection, or cardiac abnormalities, were excluded. Clinically significant anemia was treated prior to grafting. Of 38 patients screened, 8 gave consent for the study between October 2013 and February 2015. Of those patients, 3 did not fulfill a clinical or molecular phenotype of severe generalized RDEB, and 1 patient died of metastatic squamous cell carcinoma prior to grafting. Data on 4 patients with at least 12 months of follow-up are presented.

Primary autologous patient keratinocytes were cultured from two 8-mm diameter skin biopsy samples obtained from unwounded, unscarred skin and transduced with COL7A1-containing retrovirus (LZRSE-COL7A1). The full-length COL7A1 complementary DNA expression was controlled using the Moloney leukemia virus long terminal repeat promoter and then the virus was pseudotyped with a gibbon ape leukemia virus xenotropic envelope protein prior to grade stock production using current good manufacturing practices. Retroviral infection efficiency was determined by counting the number of type VII collagen expressing cells (mean and range of 4 fields).

Proviral genome per cell was assayed using quantitative polymerase chain reaction (3 replicates shown) and illustrated using GraphPad. Eight type VII collagen gene–corrected grafts were produced from transduced patient keratinocytes. After approximately 25 days, patients returned for grafting under general anesthesia. Six grafts were applied to uninfected, eroded, or scarred wound sites that lacked clinical evidence of squamous cell carcinoma. Graft sites were selected based on the patient’s choice for quality of life, ease of immobilization after grafting, and accessibility during general anesthesia.

Wound beds were cauterized to minimize retained epidermal stem cells. Five wounds and 1 induced wound (site Z) were grafted on each patient. The induced wounds (site Z) were created by rubbing normal-appearing unscarred skin. Type VII collagen gene–corrected grafts were affixed to the wound beds via dissolvable sutures after preparation of the wound bed. Patients 3 and 4 consented to have small india ink tattoos placed at each graft corner to help track graft location. Grafts were covered with nonadhesive dressings and topical mupirocin for 5 to 7 days after grafting.

The primary safety outcomes were recombinant viral infection, cancer, and autoimmune reaction. Presence of recombinant retrovirus in serum samples was evaluated at 3, 6, and 12 months (clinical protocol for stopping guidelines appear in eMethods in Supplement 2) using quantitative polymerase chain reaction for the gibbon ape leukemia virus sequence (Indiana University Vector Production Facility), with a detection limit of 10 copies per 0.2 µg of DNA. Cytotoxic T cells specific for type VII collagen were assayed using counts of T_h1 (IFN-γ) and T_h2 (IL-4) (ELISPOT assay).

At each study visit, blood was collected for complete blood count, complete metabolic panel, and evaluation of circulating basement membrane zone–reactive antibodies by immunofluorescence microscopy. Immune response to type VII collagen was assessed by immunofluorescence microscopy by placing a serum sample on a primate esophagus and via direct immunofluorescence of patient biopsy sections to detect circulating and tissue-bound immunoreactants (IgG, IgA, IgM, and C3), respectively (eMethods), and by Western blot. All adverse events were characterized according to version 3.0 of the National Cancer Institute common terminology criteria for adverse events.

Type VII collagen expression was assessed by immunofluorescence microscopy or immunoelectron microscopy on graft biopsy samples, and by level of wound healing in graft sites at 3, 6, and 12 months after transplantation using an investigator global score. Wounds were scored by rating the wound site as 100% to 75% healed, 74% to 50% healed, 49% to 25% healed, or 24% or less healed compared with baseline by 2 independent investigators and with the Canfield Vectra camera (eMethods in Supplement 2).

Secondary outcomes included laboratory abnormalities, investigator assessment of biopsy sites, and graft dimensions. Representative graft sites were biopsied at each visit to evaluate type VII collagen expression by immunofluorescence microscopy and type VII collagen ultrastructural localization by immunoelectron microscopy (eMethods in Supplement 2). The timing, order, and patient assignment for the assessed photographs were not blinded. This was a deviation from the protocol to detect early squamous cell carcinoma and other adverse events at the graft site.

All patients were male, with a mean age of 23 years (range, 18-32 years), with an estimated body surface area affected with RDEB ranging from 4% to 30%. Each patient was diagnosed clinically as having generalized severe (Hallopeau-Siemens type) RDEB and demonstrated hallmarks of this RDEB subtype, including anemia, esophageal strictures, and pseudosyndactyly (Table 1).

Four adult patients with RDEB received transplants with type VII collagen gene–corrected grafts (eFigure 2 in Supplement 2). Patients carried various compound heterozygous COL7A1 mutations causing truncated type VII collagen expression (noncollagenous domain 1) as detected in a keratinocyte medium by Western blot (eFigure 3), but not in tissue by immunofluorescence microscopy (Figure 1 and Figure 2 and eFigure 4).

The biopsy sites used for graft production completely healed in all instances without complications. Primary RDEB keratinocytes were transduced with the LZRSE-COL7A1 retroviral vector containing full-length COL7A1 sequence and the long terminal repeat promoter (representative images from patient 1 in Figure 3) with a mean of 70% efficiency and 0.8 proviral genome copies per cell (shown for all 4 patients in Figure 4). Wounds were not induced in any of the biopsy or harvest sites. Most grafted chronic wounds were present for more than 5 years (Figure 5). All 24 type VII collagen gene–corrected grafts were serially monitored for the percentage of wound healing, infection, pain, and pruritus. Patients received standard of care treatments as needed during this trial (eg, transfusions, esophageal dilations).

No serious adverse events were reported. Graft site pruritus (n = 3) and increased graft site drainage (n = 2) were the most common adverse events (grade 1 or 2), and no clinical signs of malignancy were noted. Recombinant retrovirus and cytotoxic T-cell assays were negative for the majority of time points (Table 2); a minority were undetermined. At 6 months, site Z for patient 3 had wound colonization (grade 2) and was considered a graft failure.

Type VII collagen immune responses were monitored, and no patient showed systemic autoimmune symptoms or increased blistering outside grafted areas. To reduce the number of biopsies and to preserve graft tissues, not all wound sites were biopsied during every visit. In patient 1, no circulating or tissue-bound antibodies were detected (Table 2). Patient 2 showed transient tissue-bound antibodies (mild; >1 subclass of IgG and IgM at the dermal-epidermal junction) without complement fixation at 3 months or circulating type VII collagen antibodies. Patient 3 showed transient tissue-bound antibodies without circulating type VII collagen antibodies.

In contrast, patient 4 demonstrated a 1:160 titer dermal-epidermal junction binding IgG antibodies at 1 and 3 months and 1 to greater than 2 subclasses of IgG, IgA, C3, and IgM tissue–bound antibody staining at 3 months (Table 2). However, no systemic autoimmune symptoms or increased blistering outside the graft sites were noted and the type VII collagen-specific cytotoxic T-cell assay results were negative. At 6 months, circulating antibodies and complement levels were reduced (1:40) in patient 4, with no tissue-bound immune complexes detected within the graft sites. At 12 months, no circulating or tissue-bound antibodies were seen in patient 4.

After detection of dermal-epidermal junction–bound immunoreactants in patient 4, baseline serum samples of anti–type VII collagen antibodies were reassessed using Western blot. In contrast to the negative baseline immunofluorescence microscopy results, Western blot analysis showed that serum samples from both baseline (1:300) and 3 months after grafting (1:1000) were reactive to purified type VII collagen, indicating patient 4 was sensitized to exogenous type VII collagen prior to graft placement (eFigure 5 in Supplement 2). Patient 4’s serum anti–type VII collagen antibody reactivity was confirmed within a previously characterized carboxyl terminal–type VII collagen pepsin fragment¹⁶ both before and after graft placement.

Type VII collagen expression was detected in 9 of 10 biopsy samples (90%) at 3 months, in 8 of 12 samples (66%) at 6 months, and in 5 of 12 samples (40%) at 12 months. Immunofluorescence microscopy analysis of graft sites showed type VII collagen localization at the dermal-epidermal junction (Figure 1 and Figure 2 and eFigure 4B in Supplement 2) in contrast to uncorrected skin controls. Gene-corrected graft sites showed fully differentiated epidermis with spinous and granular layers, which were positive for epidermal markers keratin 14, keratin 1, and loricrin resembling normal skin (eFigure 4B). This demonstrated that type VII collagen overexpression did not inhibit epidermal differentiation.

Among the negative samples, type VII collagen was undetectable in patient 2’s biopsy samples at 6 and 12 months; however, anchoring fibrils were present in a parallel biopsy sample obtained at 6 months (Figure 1). In patient 3 at 12 months and in patient 4 at 6 and 12 months, type VII collagen was identified using noncollagenous domain 1–specific antibodies (Figure 2).

At 3 months, 5 of 7 biopsy samples (71%) showed a morphologically normal appearance and a frequency of anchoring fibrils that were reactive with type VII collagen noncollagenous domain 2 antibody. At 6 months, anchoring fibrils were detected in 4 of 12 biopsy samples (33%), but were not detected in biopsy samples obtained from patient 4. At 12 months, 3 of 12 biopsy samples displayed anchoring fibrils; however, no anchoring fibrils were detected in biopsy samples from patients 2 and 3.

Gene-corrected graft sites showed improved wound healing compared with baseline and representative photographs from each patient appear in Figure 1 and Figure 2 and in eFigure 4 in Supplement 2. In contrast, untreated wounds in patient 4 did not heal (Figure 2).

Figure 5 shows the clinical response per wound graft compared with baseline. At 1 month after transplantation, there was 75% or greater wound healing in 20 of 24 graft sites compared with baseline and 4 of 24 graft sites had improvement with 50% to 74% healing. At 3 months, 21 of 24 graft sites (87%) had improvement with 75% or greater wound healing and 3 of 24 graft sites (13%) had 50% to 74% wound healing. At 6 months, 16 of 24 graft sites (67%) had improvement with 75% or greater wound healing and 5 of 24 graft sites (21%) had 50% to 74% wound healing. However, 3 of 24 graft sites (13%) displayed extensive blisters or erosions and were considered graft failures (only 0%-49% wound healing). At 12 months, 12 of 24 graft sites (50%) had improvement with 75% or greater healing and only 4 of 24 (17%) graft sites had healing of less than 49%. If the clinical response was based on per-patient response (rather than per graft), all 4 participants would have a response rate of greater than 50% at month 12 (patient 1: 5 of 5 chronic wound sites ≥50% healed; patient 2: 3 of 5 sites ≥50% healed; patient 3: 5 of 5 sites ≥75% healed; patient 4: 3 of 5 sites ≥50% healed).

One main challenge to RDEB genetic correction has been the efficient delivery of the large COL7A1 complementary DNA transgene, encompassing greater than 9 kb. To our knowledge, this study represents the first human trial of cutaneous gene therapy using autologous type VII collagen gene–corrected grafts for wound healing. These findings suggest acceptable safety in 4 patients with severe generalized RDEB. Genetically corrected keratinocytes regenerated functional, self-renewing epidermis in 67% of graft sites tested at 6 months after grafting with type VII collagen or morphologically normal anchoring fibrils detectable up to 1 year for 3 patients (Figure 1 and Figure 2). These results are an improvement compared with previous studies of nongenetically modified allogenic keratinocyte grafts in which only 22% (2/9) chronic wounds showed healing at 18 weeks.⁹

Neither allogeneic fibroblast injections^4,17 (despite some initial improvement in wound healing) nor intradermal or intravenous injections of bone marrow–derived mesenchymal stem cells^6,7 showed long-term differences compared with placebo. A case report of a similar gene therapy using similar methods for 1 patient with junctional epidermolysis bullosa and palm-derived keratinocytes (as a rich stem cell source) demonstrated a 6-year gene correction.^14,18

In the current trial, 12 months of continued type VII collagen expression was detected in gene-corrected graft sites, corresponding to approximately 10 epidermal turnover cycles. This suggests the current intervention successfully targeted keratinocyte stem cells; however, a general decrease in wound healing at graft sites was observed at 1 year. This could be attributed to the reduced number of stem cells in biopsy samples from patients with RDEB due to unavailability of unscarred palm skin for biopsy (due to pseudosyndactyly) or uncorrected cells in the wound bed may have competed with corrected cells within the graft.

In this study, wound healing was accompanied by full-length type VII collagen expression in the dermal-epidermal basement membrane. Wound healing was also associated with the formation of anchoring fibrils that were not present prior to grafting. During the critical first few days of graft placement, some graft sites were difficult to immobilize, including the back and posterior shoulder of patient 2. This patient also had the shortest period of immobilization after transplantation compared with the other study participants (Table 1). It is possible these factors could have reduced wound healing overall and may explain the absence of detectable type VII collagen by indirect immunofluorescence microscopy in biopsies sampled at 6 and 12 months in patient 2. Patients 1 and 2 were brothers, further implicating external rather than genetic factors in their graft outcome differences.

Few adverse safety events were reported. No evidence of recombinant retrovirus in blood or squamous cell carcinoma at graft sites were seen to date; however, long-term monitoring will continue for 5 years. All molecular replacement approaches pose the risk of unwanted immune responses against the therapeutic product, particularly in patients with null mutations. In this trial, all patients expressed a nonfunctional truncated type VII collagen molecule containing the noncollagenous domain 1, which is believed to be the most antigenic portion of the protein,¹⁹ and may be responsible for minimizing the risk of potential immune reaction. No evidence of type VII collagen–associated cytotoxic T-cell activity was seen in any patients during the study.

However, immunofluorescence microscopy and direct immunofluorescence studies in patient 4 revealed transient tissue-bound antibodies and circulating antibodies (Table 2). Patient 4 was later found to have had anti-type VII collagen antibodies prior to transplantation (as detected using a more sensitive Western blot analysis) that were not identified during the screening process with a immunofluorescence microscopy assay certified by the Clinical Laboratory Improvement Amendments, suggesting that more sensitive standardized methods should be developed to assess baseline immune cross-reactivity in future therapeutic studies. Altering collagen isoform balance in the skin remains a theoretical concern; however, extensive dermatological experience at injecting collagen I and III into human skin over many years has not proven deleterious in this regard.

Limitations of this phase 1 trial included results on only 4 patients and a 12-month follow-up period. Wounds were measured using a categorical assessment of wound healing as determined by 2 independent investigators. Although there were no disagreements between the investigators scoring the level of wound healing, this RDEB global score has not been formally validated. Because this study focused on safety, photographic assessments of wounds were not blinded to detect possible adverse events, such as squamous cell carcinoma, and there was not a systematic comparison with untreated wounds due to a limited number of available patients with paired chronic wounds.

Clinical outcomes were based on a per-graft response rather than a per-patient response because there were intraindividual characteristics, such as graft location, that affected the success of grafting. Attachment and protection of the graft during the initial 2 weeks has been shown to be the most important factor in graft survival and wound healing.

Another limitation is that the trial included only a subset of patients with RDEB who had noncollagenous domain 1 and it is not known what the response would be in a broader group. In general, clinical wound healing accompanied type VII collagen expression; however, only a small portion (3- to 4-mm biopsy) of each graft was sampled serially at study visits to preserve the graft. In addition, off-target genomic analysis was not assessed; however, there was no evidence of squamous cell carcinoma in any of the 24 graft sites.

Given the greater than 1-year follow-up for 4 patients without significant adverse events, the Food and Drug Administration has allowed initiation of a phase 2A trial (NCT01263379) that is enrolling adolescents aged 13 years or older and will focus on quantitative wound outcome measurements compared with untreated wounds and other clinical outcomes, such as wound pain and itch.

In this preliminary study of 4 patients with RDEB, there was wound healing in some type VII collagen gene–corrected grafts, but the response was variable among patients and among grafted sites and generally declined over 1 year. Long-term follow-up is necessary for these patients, and controlled trials are needed with a broader range of patients to better understand the potential long-term efficacy of genetically corrected autologous epidermal grafts.

Corresponding Authors: M. Peter Marinkovich, MD, and Jean Y. Tang, MD, PhD, Department of Dermatology, Stanford University School of Medicine, 269 Campus Dr, Stanford, CA 94305 and 450 Broadway St, Redwood City, CA 94063 (mpm@stanford.edu and tangy@stanford.edu).

Author Contributions: Drs Tang and Marinkovich had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Tang and Marinkovich contributed equally.

Concept and design: Siprashvili, Furukawa, Khavari, Lane, Marinkovich.

Acquisition, analysis, or interpretation of data: Siprashvili, Nguyen, Gorell, Loutit, Khuu, Lorenz, Leung, Keene, Reiger, Khavari, Lane, Tang, Marinkovich.

Drafting of the manuscript: Siprashvili, Nguyen, Gorell, Loutit, Lane, Tang, Marinkovich.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Gorell, Keene, Tang, Marinkovich.

Administrative, technical, or material support: Nguyen, Gorell, Loutit, Furukawa, Lorenz, Leung, Keene, Reiger, Khavari, Lane, Tang, Marinkovich.

Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Siprashvili reported having a US patent pending. Dr Lane reported having a US patent pending. Dr Tang reported having a US patent pending and being a clinical investigator for Scioderm. Dr Marinkovich reported having a US patent pending and being a clinical investigator for Fibrocell. No other disclosures were reported.

Funding/Support: This work was supported by grant R01 AR055914 from the National Institutes of Health and by funding from the Epidermolysis Bullosa Medical Research Foundation and the Epidermolysis Bullosa Research Partnership. Drs Khavari and Marinkovich were supported by funding from the Office of Research and Development, Palo Alto VA Medical Center. Dr Tang was supported by a Damon Runyon Clinical Investigator Award.

Role of the Funder/Sponsor: The funding sources had no influence on the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We thank the patients for granting permission to publish this information. We gratefully acknowledge the study staff, all the physicians, nurses, and other personnel at Stanford University and the Lucille Packard Children’s Hospital.