Topical Platelet-Derived Growth Factor vs Placebo Therapy of Diabetic Foot Ulcers Offloaded With Windowed Casts: A Randomized, Controlled Trial

Objective. This study sought to compare the efficacy of topical platelet derived growth factor (Regranex, Smith and Nephew, London, UK) (test group) to placebo (control group) in treating diabetic foot ulcers. All subjects had a short leg walking cast with a window fashioned in the cast over the site of the ulcer.   Methods. Forty-six subjects were randomized (double-blind) 1:1 to the test or control group and treated for up to 4 months. Subjects had Wagner grade I ulcers with wound area of 1 cm² to 16 cm² without severe peripheral arterial disease, osteomyelitis, or any infection requiring antibiotics. Study medication was applied daily and casts changed approximately every 14 days. Results. Of the 46 subjects randomized, 38 either healed or completed 16 weeks of therapy without healing. Eight subjects dropped out prior to 16 weeks. Based on intention-to-treat, 12 of 23 (52%) test group subjects healed before 16 weeks compared to 13 of 23 (57%) control group subjects (not significant). Regression analysis demonstrated that slower healing was associated with larger initial wound size (hazard radio [HR] = 0.997, 95% confidence interval [CI]: 0.995-1.00, P = 0.028) and excessive wound drainage (HR = 0.346, 95% CI: 0.126-0.948, P = 0.039). Excluding the patients who dropped out, 25 of 38 (66%) subjects healed by 4 months. Three additional subjects healed with casts that were worn longer than 4 months, for an overall rate of 74% at 9 months. Five subjects developed cast burns, and 3 patients required amputation. Conclusion. Topical platelet derived growth factor does not appear to significantly improve healing of Wagner grade I diabetic foot ulcers that are treated by offloading with a short leg walking cast. Excellent healing rates may be achieved with casting alone.

  A variety of factors influence healing of diabetic neuropathic ulcers including the presence of underlying deep infection such as osteomyelitis, impaired sensation, the adequacy of arterial circulation, and the presence of deformity. Decreasing the pressure exerted on these wounds via offloading of the affected foot is universally accepted as a critical part of the management of these wounds.^1-6 Unfortunately, many patients are not sufficiently compliant with standard offloading shoes and braces, so caring for these wounds often fails.⁴ One generally successful approach to offloading that overcomes the problem of noncompliance has been the use of below-knee plaster or fiberglass casts. Casting neuropathic ulcers originated with treatment of Hansen’s disease, also known as Leprosy, in India in the 1930s.¹ Subsequent designs were modified by limiting the amount of padding and the total contact cast (TCC) method was developed by Brand and colleagues.^2,3

  Casting methods are limited, however, as they preclude the use of topical medications that independently promote healing, and unless a window is fashioned in the cast, the wound cannot be inspected until the cast is changed. Windowing the cast overcomes both of these limitations, but despite reports of their safety and efficacy from a few centers internationally,^4,5 the method has not been widely adopted in the United States.

  Several years ago the authors experimented with treating diabetic foot ulcers with the combination of topical human recombinant platelet-derived growth factor (PDGF) (Regranex, Smith and Nephew, London, UK) and offloading via windowed casts. The subsequent retrospective analysis demonstrated that the combination of therapies achieved a very high level of healing—80% of wounds treated healed in an average interval of approximately 2 months.⁶ This study was not randomized or prospective, and a more thorough evaluation of this combination seemed justified. To evaluate the relative importance of PDGF for the remarkable efficacy found in this earlier study independent of offloading, the present study was designed as double-blinded and prospective.

  The study was approved by the institutional review boards at the 2 participating institutions, Veterans Affairs Long Beach Healthcare System (Long Branch, CA) and Veterans Affairs Greater Los Angeles Healthcare System (Los Angeles, CA), and was conducted between March 2007 and February 2010. Enrollment was ended at 46 patients based on a projection that this would provide sufficient power to detect an estimated 30% improvement in treatment efficacy over control assuming a healing rate of 40%.

  Eligible patients had to have Wagner grade 1⁷ diabetic foot wounds for at least 1 month with full-thickness skin involvement but no joint, tendon, or bone exposure (Wagner grade 2 or 3). Wound size had to be between 1 cm² and 16 cm². Osteomyelitis or infection requiring antibiotics at enrollment were exclusions, as were moderate to severe peripheral arterial disease (ABI < 0.7 or toe pressure < 0.6). Active neoplasia, concomitant steroid use, chronic alcohol or drug abuse, more than 2 full-thickness wounds on the involved extremity, and concomitant use of an antiplatelet agent (Pletal, Otsuka Pharmaceutical Co Ltd, Tokyo, Japan) were also exclusions. Patients were excluded if they had an unstable gait prior to casting or after cast application. There were 3 patients excluded because they had undergone previous foot-sparing amputation (eg, transmetarsal amputation) such that the resulting foot remnant was too small to accommodate the cast window while still providing adequate off-loading. Two were excluded as being fall risks when they first attempted ambulation after application of a cast. Morbid obesity was not an exclusion criterion and did not appear to present a problem with subjects tolerating casts.

  Recombinant human PDGF was purchased from Janssen Pharmaceuticals Inc (Titusville, NJ), and the medication was repackaged in unlabeled, aseptic 15 ml tubes. A placebo hydrogel lacking PDGF, but with the same ingredients, pH, and viscosity of the PDGF used in the test group was prepared by a compounding pharmacy (California Pharmacy, Newport Beach, CA) and packaged in identical-appearing tubes. The study drug was stored between 2˚ C and 8˚ C with coded labels so neither subjects nor investigators could tell which medication was dispensed. Subjects were taught to store drug supplies in a refrigerator and to keep tubes chilled during transport.

  All subjects were male and had diabetes. Demographic data and wound characteristics were recorded at each patient’s baseline screening. Sharp debridement was performed prior to wound photographs and tracings. The presence of moderate to severe deformity was assessed in the affected foot and limb, as well as the presence of excessive wound drainage. Hemoglobin A1c and erythrocyte sedimentation rate (ESR) were also determined at screening.

  Subjects who met the previously described inclusion criteria were then randomized on a 1:1 basis to placebo hydrogel (control group) or PDGF (test group) based on a predetermined randomized list, with investigators and subjects blinded to treatment allocation.

  At randomization, a fiberglass cast was applied with a window slightly larger than the study wound dimensions. The technique of casting was based on methods for creating well-padded short leg walking casts, not classic TCC methods. Blinded study medication was dispensed and each subject taught a uniform method for daily local care, including 1) daily cleansing of the wound with saline, 2) application of a thin coat of study medication to the wound, 3) covering the wound with a single layer of saline moistened gauze, 4) placement of a layer of sterile, nonadherent wound dressing (Telfa, Covidien, Mansfield, MA) over the moist gauze, and 5) replacement of the fiberglass plug of cast material and securing the plug with tape.

  The duration of study therapy was 4 months after randomization, with a final follow-up visit scheduled 6 months after healing or 10 months after randomization if the wound was not healed at 4 months. Subjects returned 1 week after randomization for an initial cast check, and, if necessary, to make modifications. Photographs and tracings of the wounds were performed at follow-up visits every 2 weeks after randomization, up to 4 months. Sharp debridement was performed when clinically indicated to remove callus, slough, or devitalized tissue. Complete epithelialization was used to define healing. Once healing was observed, further therapy and offloading was at the discretion of the treating physician.

  Windowed casting. The casting method employed in this study was based on the standard well-padded short leg walking cast.⁸ The windowed casts had 5 layers. Starting with the skin these were 1) a thick hydrocolloid protective layer applied directly to the skin around the wound (Duoderm CGF, ConvaTec, Inc, Bridgewater, NJ), 2) a cotton tubular roll gauze (Stockinette, Alba Health Products, Inc, Boynton Beach, FL), 3) water resistant 1/8 inch foam padding (Delta Terry-Net Adhesive Cloth/Foam Padding, BSN Medical, Inc, Charlotte, NC), 4) cotton roll cast padding (Webril, Covidien, Mansfield, MA), and 5) four-inch fiberglass (Scotchcast Plus, 3M, St. Paul, MN). In most cases (89%) when no previous forefoot amputation had been performed, a toe plate type cast was used so that the toes were exposed, otherwise the forefoot was sealed completely within the cast. In a few instances (eg, Charcot deformity or transmetatarsal amputation), rubber plantar bars were fixed to the bottom of the cast, but in most cases only a standard cast shoe was dispensed to protect the bottom of the cast.

  Routine application includes the following steps:

    1. Clean the skin to be incorporated in the cast with warm water, dry the skin, and apply petrolatum from knee to ankle.

    2. Apply a thick hydrocolloid protective layer starting close to the wound perimeter and extending approximately 1.5 inches away from the wound. The area of the wound is exposed via a cut-out in the hydrocolloid protective layer. This step is aimed at protecting the surrounding skin from moisture maceration. In wounds with excessive drainage, application of the hydrocolloid protective layer several inches away from the wound may be necessary to sufficiently protect a larger area of skin.

    3. Use 3-inch or 4-inch cotton tubular roll gauze to cover the foot, ankle, and leg with an additional 6 inches to 8 inches extending distally that is subsequently rolled back proximally to create 2 layers over the foot. Incise the gauze anteriorly with a transverse cut to accommodate ankle dorsiflexion with overlapping of the cut material.

    4. Place the foam padding over bony prominences including, but not limited to, the heel, the malleoli, and the anterior aspect of the tibia. This padding is intended to protect the skin from shearing and cast burns over boney prominences.

    5. To precisely locate the wound under the gauze, first cut a square piece of foam padding approximately 0.5 inches x 0.5 inches larger than the wound. Then make a hole in the gauze slightly larger than the wound and place the foam square directly over the center of the wound with attachment to the thick hydrocolloid protective layer previously placed around the wound. This step completes a picture frame centered on the wound comprising an inner layer of the hydrocolloid layer and an outer layer of foam padding.

    6. Place a pre-made square of padded foam with a sterilized 1-inch, large-head anodized roofing nail sticking up out of the padded foam in the center of the wound with the nail spike pointed perpendicular to the wound surface. The nail spike will serve later as a marker buoy to guide cutting of the fiberglass plug.

    7. Apply 3 layers of cotton roll cast padding extending from the heel to the dorsum of the foot, starting at the distal end, wrapping the foot and ankle. Make sure the nail spike is exposed outside the padding. Apply an additional 3 layers of padding from the dorsum of the foot to the proximal end of the gauze. Continue to wrap the leg to create generous padding, using 50% overlap of the padding.

    8. Moisten and apply the first roll of fiberglass starting on the lateral aspect of the foot. Apply 2 layers around the medial and lateral aspects. Wrap the dorsum of the foot and ankle. Apply the second roll of fiberglass starting at the distal end of the cast, wrapping the foot again, and advancing to the intended proximal margin at the middle of the leg.

    9. Position the ankle and foot in the desired position (usually neutral) by molding the plantar surface of the foot and all other landmarks of the ankle and leg. Start the third roll of fiberglass by folding 4 to 5 layers back and forth upon each other like a fan to create a reinforcing plantar fiberglass splint from heel to toe. Secure the fiberglass splint with the remainder of the third roll starting at the heel, wrapping the fiberglass in a spiral fashion all the way to the proximal margin of the cast.

    10. Recheck the molds and make adjustments as needed. Check the molding of the leg and fold excess gauze and padding proximally and distally (if a toe plate type cast is constructed) over the fiberglass.

    11. Take care to keep the marker nail spike exposed outside the padding and fiberglass during the steps 7-10.

    12. Trim the fiberglass and padding toe plate area so that it is a half to three-quarters of an inch longer than the hallux, and trim the distal end to close to the fifth toe. Fashion an even line to maintain a smooth and well-shaped finish cut. Secure the folded edges of padding and gauze with tape to the fiberglass. Fold the proximal end of the cast down to make a soft circumferential edge. Apply a final layer of fiberglass to complete the cast.

    13. Use a felt-tip pen to mark the cast material for removal centered on the protruding nail spike. Cut out a squared window (cast plug) with a cast saw. Carefully pull on the nail to remove the cast plug. Remove the nail from the cast plug. Remove or trim any excess padding to expose the wound. It is important that when the cast plug is reinserted after dressing the wound that it not protrude above the level of the cast, particularly if the wound is over a weight bearing surface, as this will counteract offloading.

    14. Apply study medication and dress the wound with saline-moistened 2 inch x 2 inch or 4 inch x 4 inch gauze (depending on wound size) and then a piece of nonadherent dressing. The surface of the dressing should be flush with the surface of the padding layer of the cast. Place the cast plug back into the window and secure with tape.

    15. Fit the patient with an appropriate cast shoe.

  Figure 1 shows a casted foot with toes exposed and a window over an ulcer.

  Demographic data and wound characteristics were tabulated. Differences in these variables between the 2 treatment groups were examined using 2-sample t tests for continuous variables and chi-square tests for categorical variables. All analyses were on an intention-to-treat basis. Kaplan-Meier (KM) curves of wound healing were plotted for the test and control groups, which were compared using log-rank test. In addition, separate Cox Proportional Hazards (Cox PH) regressions were performed to assess associations between wound healing and various patient and wound characteristics. These include the initial wound size, ESR, presence or absence of severe deformity, A1c hemoglobin, and the presence of high levels of wound drainage. The final multifactor Cox PH regression model included the treatment indicator and the factors with significance level of 0.05 or less from the separate regressions. All analyses were conducted using SAS 9.2 software (SAS Institute, Cary, NC), and the graph was generated using the publicly available statistical software R, developed by the R Development Core Team.

  Demographic and wound characteristics. Demographic and wound characteristics of the patients are shown in Table 1. A total of 46 subjects were enrolled, and 23 were assigned randomly to each treatment group. The age, body mass index (BMI), A1c hemoglobin levels, current smoking status, and proportion of subjects taking insulin were comparable between the 2 groups. There was significantly higher mean ESR in the test group (32.3 ± 16.6 mm/hour vs 54.6 ± 24.5 mm/hour for control vs test, respectively; P = 0.004). In addition, the wound duration prior to treatment was longer in the test group (18.5 ± 22.2 compared to 9.8 ± 11.6 weeks for the control group, P = 0.11) although this did not achieve statistical significance.

  Treatment effect on healing and factors related to healing. Based on intention to treat, without excluding subjects who dropped out, there was no significant difference in healing rates at 4 months between patients in the control group and the test group (57% vs 52%, respectively; chi-square P = 0.77). Figure 2 shows the Kaplan-Meier curves of wound healing during 4 months of monitored therapy by the test group. No significant difference was observed in time to wound healing between the test and control groups (median time to healing: 97 vs 91 days, respectively, log-rank P = 0.70). Results from the separate regressions indicate that only 2 factors were significantly associated with longer time to wound healing, initial wound size (hazard ratio [HR] = 0.997, 95% confidence interval [CI]: 0.995-1.00, P = 0.028), and excessive drainage (HR = 0.346, 95% CI: 0.126-0.948, P = 0.039). A multifactor Cox PH regression model demonstrated that slower healing was associated with larger initial wound size (HR = 0.998, 95% CI: 0.996-1.00, P = 0.017) as well as excessive wound drainage, (HR = 0.370, 95% CI: 0.132-1.036, P = 0.059) though the latter did not achieve statistical significance.

  Dropouts and complications. During the course of the study, 8 subjects (3 in the test group and 5 in the control group) dropped out. One subject in the test group died unexpectedly of unrelated causes during the first week after randomization. The other 7 subjects dropped out due to anxiety (1), inability to tolerate the cast (1), job change (1), cast-related pain (2), inconvenience of the cast (1), and hospitalization for unrelated causes (1).

  There were no other deaths during the treatment and follow-up intervals. If dropouts are excluded, 66% (25/38) of subjects who were treated with casting for 4 months experienced complete wound healing. Three subjects, 2 from the control group and 1 from the test group, who had partial wound healing during the first 4 months continued with casting and healed during the follow-up phase at 5, 6, and 9 months after randomization, respectively. When these 3 subjects are considered, complete healing of the study ulcers was observed in 74% (28/38) of subjects treated with cast therapy up to 9 months after study enrollment.

  New cast-related ulcers, or cast burns, developed in areas of skin undamaged at study entry in 5 subjects. Three were in the test group and 2 in the control group. Three of these patients were successfully treated by continued casting with a second window cut out over the new ulcer until the cast burn healed. One patient developed a cast burn that never healed. Lastly, 1 patient healed his study wound over the fifth metatarsal head, but developed a cast burn over the midfoot, which eventually required transmetatarsal amputation.

  Two other subjects enrolled in the study required amputation. The first was a patient in the control group whose wound worsened despite off-loading, which eventually required transmetatarsal amputation 6 months after enrollment. The other patient was in the test group. His wound never healed, and after casting was discontinued at 4 months, he developed infection requiring below-knee amputation at 14 months after enrollment.

  It was found that when effective off-loading was employed, equivalent rates of wound healing were seen when a topical PDGF was compared in a blinded, prospective fashion to a placebo hydrogel. The simplest explanation for this finding, given that baseline characteristics in the 2 treatment groups were for the most part very similar, is that the efficacy of the off-loading regimen is much more important than the topical treatment regimen, and when off-loading is close to optimal, differences in the efficacy of the topical agents employed have minimal impact on healing rates.

  A similar randomized prospective trial in India compared once-daily application of 0.01% recombinant human PDGF (Plermin, Dr. Reddy’s Laboratories Inc, Hyderbad, India) to saline-moistened gauze dressing, both applied through a cast window.⁹ Ten subjects were in each group. All wounds healed by the end of the study, but mean time to healing was shorter by 41.8% in the growth factor group (50 ± 23 days compared to 86 ± 31 days, P = 0.02). The divergence of these results from the findings in the current study may be related to the superior efficacy of the hydrogel vehicle for PDGF-BB compared to saline gauze. In support of this, one of the studies performed to secure US Food and Drug Administration approval of topical recombinant human PDGF in the United States compared saline-moistened gauze to placebo hydrogel for treatment of diabetic foot ulcers, which showed 22% healing for the gauze vs 36% for the hydrogel (P = 0.078, chi-square).¹⁰

  Given the small number of subjects in the present study, it is possible the inherent efficacy of the topical PDGF was masked by the study being underpowered to detect the approximate 30%-35% greater healing in diabetic foot ulcers reported in previous and larger randomized trials.^10-12 Moreover, there were 2 differences in the treatment groups that may have confounded the analysis by favorably influencing the control group; namely, higher ESR and longer duration of the wounds in the test group. Erythrocyte sedimentation rate was measured to detect previously undiagnosed bone infection, and in general, higher ESR is not associated with worse prognosis when osteomyelitis is absent, as was the case here. On the other hand, previous studies have shown that wound chronicity is a significant prognostic factor for healing.¹³ This disparity could have overshadowed a modest positive effect of the topical PDGF on healing. It seems likely, however, that even if a positive effect with topical PDGF was missed, the magnitude of the effect, when casting is employed, is not as large as has been observed with less stringent off-loading.

  This study’s overall healing rates, particularly the finding that 66% of the subjects who participated for 4 months experienced complete healing, is better than that found in a meta-analysis of the reported healing rates for conventional off-loading regimens for diabetic foot ulcers.¹⁴ The low rate of serious adverse events associated with therapy in both groups (ie, 5 cast burns, 1 leading to metatarsal amputation) and the relatively high level of acceptance (only 7 of 46 subjects dropped out voluntarily) indicates that the method is safe and could be widely employed.

  One important aspect of the current study is that the windowed casts were essentially modified versions of well-padded short leg walking casts. The TCC method is technically more challenging to employ, and based on past experiences, the authors suspect it may have a higher risk for cast burns. A wide range of healing rates have been reported with the TCC method including those much higher and those about the same as the current study.^4,5,15-21 Results from studies with windowed casts and TCC are shown in Table 2. Although this study did not compare the 2 casting methods, the authors speculate that similar results would be achieved with both methods, although a prospective study is required to verify this point. If healing rates are in fact similar with both methods, windowed casts would have significant advantages: 1) windowing allows much more frequent observation of the wound, 2) the cast change interval can be extended (in the authors previous retrospective study⁶ casts were changed monthly) and changes every 2 or 3 weeks are practical at a significant cost-savings compared to TCC, 3) in contrast to TCC, this technique is easier to learn and use, and 4) if there were a topical medication whose effect was not overshadowed by effective off-loading, it could be combined with windowed casts.

  Most of the randomized controlled trials evaluating casting do not report the percentage of patients for whom casting is appropriate and well-tolerated.^4,5,18,21,22 Similarly, it is difficult to discern from the available literature what proportion of patients with diabetic ulcers would be exposed to an unacceptable risk for falls due to gait disturbances or balance problems. The authors’ experience, both in this study and in their practice, is that the large majority of patients (at least 80%) with diabetic foot ulcers can safely be casted either with short leg casts, such as those employed in this study, or with TCC methods. It should be appreciated, however, that many patients are significantly inconvenienced by casting, particularly in relation to the operation of motor vehicles, mobility at work, and limitations with using public transport. Based on their experience, the authors think these factors are just as likely to preclude casting as the risk for falls. Further, given how important patient compliance is for successful casting, it has been suggested that, in part, the efficacy of casting compared to other modalities reflects better compliance in subjects amenable to casting due to their willingness to tolerate the associated inconveniences.^15,23

  Although casting is regarded as a gold standard for treatment of diabetic foot ulcers, its utility is limited by other factors in addition to risks for falls, limited mobility, and inconvenience. Additional disadvantages are that applying a cast is time and labor intensive, it requires trained staff, and it is associated with cast-induced pressure-ulcers.²³ Alternative off-loading treatment options for those found unsuitable for, or unable to tolerate casting, include removable cast walkers and therapeutic shoes, as well as surgical correction of deformities. The limitation of all removable shoe gear is clearly compliance—removable shoe gear is limited by compliance, as several studies have shown,^4,24 but not clearly precluded, as some patients cannot tolerate casting—as patients with diabetes are compliant with wearing off-loading footwear less than 50% of the time.^4,24 Although many approved and widely employed topical agents, such as engineered tissue substitutes²⁵ or collagenase,²⁶ have been shown to improve diabetic ulcer healing, no randomized studies have compared an advanced local regimen combined with casting for off-loading in a manner similar to the current study. In particular, negative pressure wound therapy has been clearly shown to improve wound healing for diabetic foot infections,²⁷ but its combination with casting has never been studied. Similarly, as far as the authors know, none of the other advanced agents, including silver products, foams, and collagen matrices have been rigorously studied in combination with casting methods. Although the authors presume that in some circumstances such combinations should lead to better results than 1 modality alone, this was not the case in the current study. To what degree advanced local methods partly ameliorate the impact of suboptimal off-loading, but fail to improve the results when optimal off-loading (eg, bed rest, crutches, and casts) is employed, remains an open question.

Christine Ma, MD; and Michael A. Hernandez, OT are from the Department of Surgery, Veterans Affairs Long Beach Healthcare System, Long Beach, CA. Vincent E. Kirkpatrick, MD; and Ian L. Gordon, MD, PhD are from the Department of Surgery, Veterans Affairs Long Beach Healthcare System, Long Beach, CA; and Department of Surgery, University of California Irvine School of Medicine, Orange, CA. Li-Jung Liang, PhD; and Aksone L. Nouvong, DPM are from Veterans Affairs Greater Los Angeles Healthcare System, Los Angeles, CA; and University of California Los Angeles, David Geffen School of Medicine, Los Angeles, CA. Aksone L. Nouvong, DPM is also from Western University of Health Sciences, Pomona, CA.

Address correspondence to:
Vincent E. Kirkpatrick, MD
University of California Irvine Medical Center
Department of Surgery
333 City Blvd W, Suite 700
Orange, CA 92868
kirkpatrickv@uthscsa.edu

Disclosure: The authors disclose this study was funded by the Heritage Medical Research Institute of Heritage Provider Network, Inc, Northridge, CA, a health maintenance organization whose research arm supports medical research. The sponsor had no influence on the collection and interpretation of data or the writing of the manuscript.