Risk Factors Associated with Healing Chronic Diabetic Foot Ulcers: The Importance of Hyperglycemia
Diabetic foot ulcers (DFU) frequently precede limb loss1 and remain a difficult clinical problem to treat. Standard wound management protocols have not proven to result in rapid healing. Margolis et al2 reported the results of a meta-analysis of 10 prospective studies of DFU — the aggregate percentage of patients healed with standard care protocols was 24% at 12 weeks and 31% at 20 weeks.
Numerous treatment strategies have been devised in an attempt to speed the healing process, including the use of topical growth factors, living human skin equivalents, hyperbaric oxygen, and electrical stimulation. Some, including platelet-derived growth factor and two living human skin equivalents, studied in prospective multicenter clinical studies,3-5 yielded significant benefit regarding wound closure compared to controls. However, the improvement in the percentage of patients healed in these trials has been disappointing, with an increase of 10% to 20% over standard treatment in each trial. Coupled with the high cost of these active therapies, the relatively moderate benefit has led to questions concerning the appropriate use of growth factors and skin equivalents.
Dermagraft (Smith and Nephew, Inc, Largo, Fla) is a living human dermal substitute consisting of neonatal human dermal fibroblasts (HDF) seeded onto an absorbable polyglactin mesh.6 It was found to significantly increase the rate of wound closure compared to control in a multicenter randomized clinical study of 245 patients.5
The effect of hyperglycemia on the critical components of wound healing in animal models has been studied extensively.7-10 Inhibition of angiogenesis by chronic hyperglycemia has been reported at multiple levels; it appears that consistent hyperglycemia negatively affects the potential for new capillary growth and other key components of the wound healing process.8 Because subgroup analysis and risk factors associated with the incidence of healing were not reported in the human dermal substitute study5 and in order to provide further information on the use of this active treatment modality, a secondary analysis of the study data was conducted to determine risk factors related to wound closure in the HDF and control groups. Specifically, the relationship of HgbA1c levels to wound closure was studied.
Study Design and Analysis
Human dermal substitute structure. The HDF utilized is a cryopreserved, human fibroblast-derived dermal substitute. Human neonatal dermal fibroblasts are cultured in vitro onto a bioabsorbable mesh. The fibroblasts proliferate across the mesh, secreting human dermal collagen, matrix proteins, growth factors, and cytokines to create a three-dimensional human dermal substitute containing metabolically active, living cells.11
Human dermal substitute DFU study. The specific inclusion and exclusion criteria, treatment particulars, and follow-up of the HDF diabetic foot ulcer study have been described in detail.5 Briefly, the study was a prospective, multicenter, single-blind, randomized, controlled investigation that compared an HDF regimen to a control regimen (frequent debridement and wound dressing with saline-moistened gauze) in the treatment of chronic diabetic foot ulcers. Between December 1998 and March 2000, 314 were enrolled in the study. At the screening visit, before randomization, study ulcers received sharp debridement. Following randomization and screening, weekly application of HDF occurred for up to 8 weeks or until ulcer closure, defined as “full epithelialization of the wound with the absence of drainage.”
Patients in both the HDF and control groups were evaluated weekly until complete wound closure or until the study endpoint (12 weeks). At each visit, tracings of the wound margins were made to document changes in wound size using computer planimetry and photographs were taken for a visual record. An ulcer was considered healed only after closure was confirmed at the next weekly visit.
Eligible patients were adults with diabetes and a foot ulcer measuring between 1.0 cm2 and 20 cm2 on the plantar surface of the forefoot or heel at Day 0 that had been present for a minimum of 2 weeks. The ulcer could extend through the dermis and into subcutaneous tissue but without muscle, tendon, bone, or joint capsule exposure. The patient’s ankle-brachial index by Doppler was ≥0.7 on the study limb.
Results of the original study showed that HDF significantly increased the percentage of wounds closed after 12 weeks of treatment in patients with a history of foot ulceration of >6 weeks at the time of enrollment. For the purposes of risk factor analysis in the current study, the dataset has been limited to the 245 patients with ulcers of >6 weeks duration.
Variables analyzed. All variables collected in at least 80% of patients were analyzed, including the following information collected at baseline: age, race, gender, diabetes type, smoking history, alcohol use, ulcer duration, ulcer size, HgbA1c, serum albumin, magnesium, and creatinine. Additional variables, collected during patient follow-up, were evaluated for their relationship to ulcer closure: average HgbA1c (mean of initial and 12 week HgbA1c), change in HgbA1c over the 12-week study period (initial minus 12 week HgbA1c level), average hours of weight-bearing, and study ulcer infection.
Statistical analysis. Data management and analysis of safety and effectiveness data were provided by the Biostatistics and Clinical Data group of Advanced Tissue Sciences, Inc., La Jolla, Calif. Analysis of each of the potential clinical risk factors noted above was performed in relationship to wound closure time using the two-sided Cox’s proportional hazards model — positive factors were evaluated to determine relationship to the proportion of wounds healed using the one-sided Fisher’s exact test. Multivariate analysis was performed to determine whether the factors found to relate to healing were independent of the assigned treatment group. A P value of <0.05 was required for statistical significance. Risk ratios are presented along with 95% confidence limits.
Descriptive statistics and the median and standard deviation of continuous variables evaluated are listed for all 245 patients in Table 1. The majority of study participants (73.9%) were men (n =181) with type 2 diabetes (n = 186), an average ulcer duration of 53.3 weeks (± 46.8), and an average age of 55.6 (± 11.1) years. The following factors were not found to be associated with time to wound closure: age, race, diabetes type, ulcer duration, initial HgbA1c, average hours of weight-bearing, smoking or alcohol history, creatinine, magnesium, and albumin levels (see Table 2). Risk factors found to have a significant independent effect on time to wound closure included initial ulcer size, gender, infection, and change in HgbA1c (see Table 2).
The proportion of wounds closed with an initial ulcer size >2 cm2 (11 out of 90, 12%) was significantly lower than the proportion for smaller ulcers (49 out of 155, 32%, P = 0.02). Women were twice as likely to heal as men (P = 0.009). An episode of infection during the 12 weeks of treatment was associated with a 3.4 times increased risk of nonclosure (P <0.01).
Initial and terminal (value at 12 weeks of treatment) HgbA1c levels, obtained on 206 patients completing the protocol, showed a trend toward reduced proportion of wounds healed in patients with lower initial and terminal HgbA1c levels but this difference did not reach statistical significance (P = 0.09) (see Figure 1). To better determine the relationship between glucose control and wound healing, the change in HgbA1c between initial and 12-week levels was calculated. Patients with a decrease in this value were believed to have better glucose control on average than those with an increasing level. Of the 245 patients studied, 206 had complete Hgb A1C data; HgbA1C increased in 101 patients during the study. Only 21 (20.7%) of patients whose HgbA1c increased during the study achieved wound closure. Among the 105 patients whose HgbA1c levels decreased during the study, 38 (36.3%) experienced wound closure — a significant improvement (P <0.05). In the control group, no significant difference was noted in the rate of healing in those with improving (10 out of 45, 22%) compared to worsening (11 out of 48, 23%) HgbA1c. Also, the number healed in the HDF-treated group with worsening HgbA1c (10 out of 53, 19%) was similar to the control rates but the number healed in the HDF group with an improving HgbA1c (28 out of 60, 46.7%) was significantly higher than in the other three groups (P = 0.009) (see Figure 2).
Because diabetic foot ulcers are slow to heal when standard techniques are used, interest in alternative treatment modalities is growing. Innovative products approved for use in these cases include growth factors and living skin substitutes. Randomized trials of these products have reported significantly improved healing rates, but the level of improvement in general has been less than expected.3-5 In most cases, investigators have labored to design protocols emphasizing inclusion/exclusion criteria and standard treatment to allow fair comparison of the test product to standard therapy. Less effort has been expended to maximize the potential of the patient to heal with or without the test product.
This study reports the results of an extensive subset analysis of data collected during a study of a living skin substitute in the treatment of DFUs. Inclusion criteria in this study were designed to allow enrollment of a representative group of patients similar to the “real world” situation where most patients are ambulatory and may have relatively poor control of their diabetes. Therefore, no criteria were used to exclude patients with high HgbA1c levels and patients with random glucose levels as high as 450 were included. The liberal criteria provided an excellent opportunity to evaluate which of the numerous variables affect healing rates in the standard therapy group and in patients treated with living dermal fibroblasts.
Some of the risk factors found to relate to improved healing such as smaller wound size and lack of infection are not surprising; they have been reported in previous DFU studies.12,13 Several risk factors found to relate to healing in other DFU studies were not found to affect healing in this study. The duration of ulceration before study enrollment has been described as an important risk factor previously, with ulcers of longer duration having poor healing rates.13,14 However, only patients with ulcers of duration >6 weeks were enrolled, which may have affected analysis of this risk factor. Ulcer duration related to healing usually depends on the type of treatment the wound has received over time. An ulcer may have been present for years but treatment neglected pressure relief and debridement or involved the use of toxic agents. With proper therapy, healing may proceed rapidly.
Clinicians often believe that increasing amounts of ambulation will adversely affect healing rates of plantar diabetic foot ulcers. Patients are typically counseled to remain nonweight-bearing as much as possible.14
However, relevant clinical trials have not reported a correlation between weight-bearing and healing rates.3,4 Plus, time of ambulation is difficult to assess from patient-managed journals, such as were included in this study. Recently, Armstrong et al15 reported on the lack of compliance with study orthotics as assessed by pedometers. Patients in this study wore their orthotic for approximately 30% of their walking steps. It is unlikely that patients would routinely confess this level of noncompliance to study coordinators, rendering the literature inaccurate and bringing the impact of being “ambulatory” on healing into question.
Of significant interest was the finding that gender was a significant risk factor. Women appeared to have a significantly better chance of complete wound closure. Reasons for this difference are not obvious. Although it has been suggested that women may be more likely to reduce ambulation and follow specific wound care instructions, this has not been documented in a reliable study to date. The influence of androgen levels on wound healing has been studied in animal models. Ashcroft et al,16 using a mouse wound model, reported that castration of male mice resulted in acceleration of cutaneous wound healing, suggesting testosterone upregulates proinflammatory cytokine expression by macrophages and inhibits wound healing. In a previous multicenter study of 181 leg ulcers, van Rijswijk17 also found that men were less likely to heal than women.
To the authors’ knowledge, this is the first prospective DFU study to find an association between hyperglycemia and wound healing. It was interesting that the strongest association was only in the group of patients treated with the living dermal fibroblasts. This suggests that if the HgbA1c is worsening, indicating higher average glucose levels during the treatment phase, the wound may be less able to utilize the active fibroblasts after application.
Animal studies modeling the effect of hyperglycemia on the critical components of wound healing have found that angiogenesis is inhibited by chronic hyperglycemia at multiple levels. New capillary growth requires the degradation of the basement membrane for buds to emerge. A key step in basement membrane degradation involves increased plasminogen activator activity, resulting in conversion of plasminogen to plasmin and subsequently in degradation of fibronectin and laminin and activation of matrix metalloproteinases. Chronically elevated glucose levels have been reported to result in reduced levels of urokinase plasminogen activator and increased levels of plasminogen activator inhibitor.7 The necessary basement membrane degradation is thereby inhibited, reducing the potential for new capillary growth.
Increased production of vascular endothelial growth factor (VEGF) plays a central role in the upregulation of angiogenesis. Endothelium-derived nitric oxide (NO) release appears to be necessary for growth factor stimulation of angiogenesis. Jang et al8 reported that inhibition of NO synthase in a mouse angiogenesis model resulted in marked inhibition of angiogenesis. Nitric oxide synthase inhibitors have been reported to increase after meals in humans with diabetes mellitus, resulting in competitive inhibition of VEGF expression.9 Lerman et al10 reported that mouse fibroblasts exposed to chronic hyperglycemia expressed markedly reduced migration compared to nondiabetic fibroblasts. Lerman et al also found that VEGF expression in response to hypoxia was reduced 75% after chronic exposure to hyperglycemia. Although specific levels of hyperglycemia in the person with diabetes have not been defined as a threshold for impaired angiogenesis, it appears clear that consistent hyperglycemia would affect the potential for new capillary growth central to the wound healing process.
In a study of the effect of hyperglycemia on skin keratinocytes, Spravchikov and colleagues18 used a murine cell culture system to evaluate the ability of keratinocytes to proliferate. They reported that high glucose concentrations (20 mmol/L) resulted in decreased proliferation and differentiation of keratinocytes, in part due to reduced activity of insulin-like growth factor-I (IGF-I). Hehenberger and Hansson19 exposed normal human fibroblasts to hyperglycemia in vitro and found that glucose concentrations of 15.5 mM and above inhibited fibroblast proliferation. The cells were also resistant to increased proliferation when stimulated with growth factors including IGF-I and epidermal growth factor.
The application of a living human dermal fibroblast preparation is, in essence, an application of a source of multiple growth factors and other protein expression. The mechanism of action for wound stimulation is through delivery of these proteins to the host tissue, hopefully resulting in increased angiogenesis, fibroblast migration, and keratinocyte proliferation that will afford wound healing. Based on the previously reviewed pre-clinical study findings, this likely cannot occur in the presence of sustained hyperglycemia. In the present study, no benefit was noted in patients treated with living dermal fibroblasts who had poor glucose control. However, in those with the best control, the chance of healing was more than double the control group with good glucose control. It appears reasonable to expect that this effect also would be important for the use of other living dermal equivalents or growth factor preparations.
With this in mind, patients with DFU should be thoroughly prepared before active treatment modalities — in particular, expensive products such as living dermal substitutes — are initiated. This should involve not only preparation of the wound bed, including debridement, infection control, exudate control and offloading, but also preparation of the patient, addressing systemic issues such as hyperglycemia management and nutritional supplementation if necessary. Finally, education on the need for compliance and the objectives of the treatment plan will allow the maximal potential for benefit from DFU treatment.
Treating diabetic foot ulcers is more effective when clinicians understand and implement measures to manage both the wound and underlying factors such as uncontrolled HgbA1c levels that may impede successful outcomes. Use of adjunctive and often expensive therapies is best initiated after these factors have been addressed. Even though age, race, diabetes type, ulcer duration, initial HgbA1c, average hours of weight-bearing, a history of smoking or alcohol use, creatinine, magnesium, and albumin levels did not seem to have a significant effect on healing, initial ulcer size, gender, and infection were found to affect time to healing and proportion of ulcers healed; change in HgbA1c was found to increase the likelihood of healing after 12 weeks, especially in HDF-managed wounds. These and other factors warrant further study to better understand their role in HDF use specifically and the healing process in general.
The authors are grateful to Advanced Tissue Sciences, Inc (La Jolla, Calif) and Smith and Nephew, Inc (Largo, Fla) for the research grant received to support this study.
The Dermagraft Diabetic Foot Ulcer Study Group included the following clinicians: Marc Brenner, DPM, Glendale, NY; Leon Brill, DPM, and Jeff Stone, DO, Dallas, Tex; Andrew Carver, DPM, San Francisco, Calif; William Eaglstein, MD, and Anna Falabella, MD, Miami, Fla; Timothy Emhoff, MD, Springfield, Mass; David Fivenson, MD, Detroit, Mich; John Grady, DPM, Oak Lawn, Ill; David R. Greenberg, DO, San Diego, Calif; Douglas Hague, DPM, Sacramento, Calif; Maria Surprenant, DPM, South Miami, Fla; Larry Harkless, DPM, and John Steinberg, DPM, San Antonio, Tex; Jeffrey Jensen, DPM, Denver, Colo; Alan Kass, DPM, Livingston, NJ; Bruce Kraemer, MD, St. Louis, Mo; Steven Krych, DPM, Austin, Tex; Adam Landsman, DPM, Chicago, Ill; Michael Lerner, DPM, and Joel Lerner, DPM, Union, NJ; Scott Lipkin, DPM, Allentown, Pa; John Macdonald, MD, Fort Lauderdale, Fla; Robert Mendicino, DPM, Pittsburgh, Pa; Bruce Miller, MD, Portland, Oreg; Arshag Mooradian, MD, St. Louis, Mo; William R. Omlie, MD, Edina, Minn; Michael Pfeifer, MD, Greenville, NC; Riley Rees, MD, University of Michigan, Ann Arbor, Mich; George Rodeheaver, PhD, and David Drake, MD, Charlottesville, Va; Peter Sheehan, MD, Staten Island, NY; Jim Stavosky, DPM, Daly City, Calif; David L. Steed, MD, Pittsburgh, Pa; Scott Wyant, DPM, Mesa, Ariz; Brian Youn, MD, Fort Wayne, Ind; Gregg Young, DPM, Salt Lake City, Utah; Phil Acda, RN, Carla Ashby, Michelle Benson, Linda DiBenedetto, RN, Chris Dickinson, Kenneth R. Heilbrunn, MD, Scott Iwasaki, Ginger Lee, Adam Morgan, Shelly Taylor, RN, and Emelyn Vargas, Advanced Tissue Sciences Inc., La Jolla, Calif.