Treatment of Chronic Ulcers in Diabetic Patients with a Topical Metalloproteinase Inhibitor, Doxycycline

Disclosure: Supported in part by Veterans Administration Merit Type II grant awarded to Dr. Chin. Introduction It was estimated in 1998 that there were over two million, and perhaps up to five million, chronic wounds annually in the US alone.[1] Diabetic foot ulcers account for five percent of these ulcers and 50 to 80 percent of all lower-extremity amputations. Optimal conventional treatments for chronic wounds are based on the concepts of wound bed preparation, which include eliminating necrotic tissue and fibrinous exudate, controlling infection, establishing moisture balance, and optimizing the epidermal margin.[2] However, despite applying the principles of wound bed preparation, some chronic wounds fail to heal in a timely fashion and are candidates for advanced interventions, including topical growth factors,[3,4] bioengineered skin substitutes,[5] or surgical intervention for closure. Although advanced technologies have been proven to be effective in controlled clinical studies, they are expensive and are not universally available to all patients. Thus, there is a need for an inexpensive, readily available, simple therapy that can be added to the concept of wound bed preparation. Multiple studies have established that the molecular and cellular environments of chronic skin wounds differ dramatically from the acute wound environment. In general, chronic wounds typically contain increased levels of pro-inflammatory cytokines and increased levels of proteases that are able to degrade essential mitogenic factors.[6–9] These observations led to the hypothesis that correcting these molecular abnormalities would promote healing of chronic wounds.[10] Studies from Trengove and colleagues[6] support this hypothesis by demonstrating that elevated levels of pro-inflammatory cytokines and proteases decreased in chronic venous stasis ulcers as healing progressed. These results suggest that an inexpensive, simple-to-administer, topical agent that reduced inflammation and protease activities might be an effective adjunctive therapy for wound bed preparation. Doxycycline, an antibiotic of the tetracycline family of drugs, is an inhibitor of matrix metalloproteinases (MMPs),[11] and several animal studies reported that treatment with doxycycline or other tetracycline analogues improved healing parameters. For example, systemic treatment of rats with doxycycline increased tensile strength of rat intestinal anastomoses on Day 3 after surgery,[12] and systemic tetracycline reduced the incidence of ulceration in alkali-injured rabbit eyes.[13] Topical application of a chemically modified tetracycline, which lacks antibiotic activity but retains MMP inhibitor activity, increased hydroxyproline levels and decreased collagenase levels in skin wounds of streptozotocin-induced diabetic rats.[14] Thus, members of the tetracycline family of molecules, such as doxycycline, appear to reduce MMP activities in various animal wound models and improve parameters of healing. Another major difference in the molecular environments between acute, healing wounds and chronic wounds is elevated levels of pro-inflammatory TH1 cytokines, including TNFa, and interleukin-1 (IL-1). There are several important relationships between proteases and pro-inflammatory cytokines that appear to influence wound healing. For example, TNFa induced synthesis of IL-1 in human dermal fibroblasts,[15] and both TNFa and IL-1 stimulated synthesis of collagenase (MMP-1) by dermal fibroblasts.[16] Furthermore, TNFa inhibited synthesis of collagen mRNA and protein in cultured human fibroblasts.[17] Thus, chronically elevated levels of TNFa in wounds would be expected to impair healing by reducing synthesis of collagen and increasing synthesis of MMPs. Previous experiments showed that TNFa is synthesized as a 26kDa transmembrane precursor protein that is processed to a 17kDa soluble protein and secreted from cells after proteolytic cleavage by a metalloproteinase enzyme, named TNFa converting enzyme (TACE).[18–20] The proteolytic activity of TACE was inhibited in vitro by various synthetic MMP inhibitors.[18,21–23] In addition, systemic treatment of mice with these inhibitors, including doxycycline, prevented endotoxin-induced lethality by preventing secretion of TNFa into serum.[18,21–24] To better understand the possible effects of doxycycline on the environment of chronic wounds, we investigated the in-vitro effects of doxycycline on inhibiting the release of TNFa from cultured macrophages and on inhibiting protease activities in a sample of human chronic wound fluid. In addition, we performed a pilot, randomized, controlled trial to evaluate the safety and efficacy of topical one-percent doxycycline treatment on healing of chronic ulcers in diabetic patients. Methods Collection of chronic wound fluid. Wound fluid was collected from a patient with a nonhealing pressure ulcer of greater than six months duration using a previously described procedure.[6] Briefly, the ulcer was covered with an occlusive dressing for one hour, and the fluid that spontaneously collected was aspirated, centrifuged at 10,000 x g for 15 minutes, and the supernatant solution was stored in multiple aliquots at -80 degrees C. Effect of doxycycline on TNFa converting enzyme activity. The effect of doxycycline on TACE activity was assessed by measuring the release of soluble TNFa from a murine macrophage cell line (RAW) following stimulation by lipopolysaccharide.[21] Briefly, RAW cells were plated at 5,000 cells per well in a 24-well microtiter plate in RPMI 1640 medium supplemented with 10-percent fetal calf serum, 2mM glutamine, and 50mg/mL gentamicin, and cultured at 37 degrees C in five-percent carbon dioxide. Doxycycline (Sigma Chemical Co., St. Louis, Missouri) was added to give final concentrations of 0, 2, 20, 200, and 2,000mg/mL in the culture medium. RAW cells were then stimulated with 1mg/mL of Escherichia coli (serotype 0127:B8) lipopolysaccharide, and cell culture medium was harvested eight hours later. Levels of TNFa protein in the samples of RAW cells culture medium were measured using an enzyme-linked immunosorbent assay (ELISA) (R&D Systems, Minneapolis, Minnesota). Effect of doxycycline, tetracycline, and ilomastat on protease activities of chronic wound fluid. The effect of doxycycline, tetracycline (Sigma Chemical Co., St. Louis, Missouri), and ilomastat (Elastin Products Company, Owensville, Missouri) on protease activity in chronic wound fluid was measured using minor modifications of a previously described assay that utilizes azocoll (Sigma Chemical Co., St. Louis, Missouri) as the substrate.[25] Azocoll is insoluble bovine hide that is covalently coupled with dye molecules. Proteolysis of the insoluble azocoll substrate generates soluble colored peptide fragments. Since the hide is composed predominately of intact collagen molecules, proteolytic hydrolysis of the azocoll substrate preferentially measures the combined activities of MMPs, including collagenases (MMP-1, MMP-8, and MMP-13) and gelatinases (MMP-2 and MMP-9). The rate of hydrolysis of azocoll by non-MMPs, such as neutrophil elastase, is relatively slow. Azocoll was washed and suspended in 50mM Tris-HCl, 1mM CaCl2, pH 7.8, at a final concentration of 5mg/mL. Samples (50mL) of wound fluid, buffer, or inhibitors (50mL) were mixed with 900mL of azocoll suspension in a 1.5mL conical microcentrifuge tube. Ilomastat, a potent synthetic inhibitor of MMPs, was included as a positive control for MMP inhibition.[26] Reaction tubes were placed in a mixer that inverted the tubes 30 times per minute. After six hours incubation at 37 degrees C, the reactions were centrifuged at 10,000 x g for five minutes, and the absorbance of the supernatant solution was measured at 520nm with a spectrophotometer. Levels of protease activity in samples were calculated by interpolation from a standard curve generated by crude collagenase from Clostridium histoliticum (Sigma Chemical Co., St. Louis, Missouri) and were expressed as mg of collagenase equivalents per mL of wound fluid. Protease levels in samples were calculated as the average of three replicate assays and were compared for statistical significance by ANOVA with Tukey’s post-hoc test. Randomized, controlled trial of topical doxycycline treatment of chronic diabetic lower-extremity ulcers. Enrollment criteria. Seven diabetic patients with full-thickness, lower-extremity ulcers were enrolled in a randomized, double-blinded, placebo-controlled study comparing the effect on healing of once-daily, topical one-percent doxycycline gel treatment with vehicle treatment and standardized good wound care. Following a thorough history and physical exam at the screening visit, patients who met the following criteria were enrolled: ulcer duration of four weeks to two years; ulcer dimensions between 1.7 to 12cm2; bacterial counts from punch biopsy of =30mmHg; ankle brachial index (ABI) >=0.8; and blood analyses within normal limits, including complete blood count (CBC), metabolic panel, and glycosylated hemoglobin A 1C (HbA1C) of up to six percent. Patients were randomly assigned to one of the treatment arms by the pharmacy at the time of enrollment. Study drugs. Patients were provided with an opaque 60mL syringe equipped with a screw plunger that contained either the doxycycline hydrogel (1% doxycycline [w/v] neutralized with sodium hydroxide in a 2.7% [w/v] sodium carboxymethylcellulose [CMC] hydrogel with 1.7% [v/v] simethicone and 0.56% [v/v] polysorbate 20 antifoaming agents) or the vehicle hydrogel (2.7% CMC hydrogel with antifoaming agents). The contents of the syringes were unknown to the patient and the physician. The color of the vehicle hydrogel was matched to the light brown color of the doxycycline hydrogel with food dyes. The doxycycline hydrogel was stable for four weeks at 4 degrees C, although patients usually received new study drugs every two weeks. Wound treatment and assessment. Until the patients were enrolled, their wounds were dressed with moist-to-dry saline gauze dressings that were changed twice daily. Patients were trained to place the hydrogel on their wounds to a thickness of about 2mm (approximately the thickness of a dime), cover their wounds with dry gauze pads, and secure their dressings with soft outer wraps. Patients were fitted with an offloading shoe and were treated with the hydrogels until the ulcer healed, or for up to 20 weeks. If the ulcer was not healed after 20 weeks of treatment, the patient could elect to receive doxycycline treatment for an additional 12 weeks of open-label use. Wounds were clinically evaluated by the physician every two weeks, photographed, and the wound edge was traced on an acetate sheet. Wound areas were calculated by multiplying the longest perpendicular length and width dimensions of the wound, which were made at each biweekly clinic visit. The effect of doxycycline treatment on healing was compared to vehicle treatment using Chi-squared analysis of healed or not healed wounds at 20 weeks. At each clinic visit, patients were also assessed for any adverse events of treatment. Results Effect of doxycycline on TNFa converting enzyme activity. As shown in Figure 1, the level of TNFa in the conditioned medium of RAW cells that were not stimulated by lipopolysaccharide (LPS) was very low (1.7±1.1ng/mL, mean ± standard error). Addition of 0.01-percent LPS to the RAW cells increased TNFa concentration approximately 160 fold (276±12ng/mL). Addition of doxycycline at 2mg/mL concentration (4mMolar) had no effect on TNFa levels in the conditioned medium of the LPS-stimulated RAW cells (273±11ng/mL). In contrast, addition of doxycycline at concentrations of 20, 200, and 2,000mg/mL to the LPS-stimulated RAW cells caused significant (pEffect of doxycycline, tetracycline, and ilomastat on protease activities of chronic wound fluid. As shown in Figure 2, the chronic wound fluid contained a high level of protease activity, 180±20ng/mL. For comparison, we reported previously that acute wound fluid from mastectomy drains contained about 1ng/mL of protease activity, while the average level of protease activity from 40 chronic wounds was approximately 60ng/mL.6 Addition of doxycycline at 50, 500, and 5,000mM significantly (pRandomized controlled trial of topical doxycycline treatment of chronic, diabetic, lower-extremity ulcers. The small sample sizes in the two treatment arms in this initial pilot study limit generalization of the results. Nevertheless, as seen in Figure 3, all four of the chronic wounds treated with doxycycline healed (times to healing were 4, 7, 24, and 30 weeks). In contrast, only one of the three patients treated with vehicle healed during the initial 20 week treatment period (patient V3 at week 4, Figure 4). Patient V2’s wound showed no trend for healing after 20 weeks of treatment with vehicle hydrogel. However, after 12 weeks of open label treatment with doxycycline, the ulcer had reduced in size by about 60%. Patient V1’s wound remained essentially unchanged during the 20 week treatment period, and he did not choose to receive doxycycline treatment. Chi squared analysis of the healing outcome of the seven patients at 34 weeks indicated that topical doxycycline treatment significantly (p = 0.05) increased healing of the ulcers compared to treatment with vehicle hydrogel. In this pilot study, another major objective was to determine safety of topical doxycycline treatment. No adverse events were noted that were attributable to either the doxycycline treatment or to the vehicle treatment. Multiple confounding variables are known to influence healing of diabetic lower-extremity ulcers. These include the initial ulcer size, the duration of the ulcer, the age of the patient, co-morbidity conditions, such as renal or hepatic diseases, smoking history, and compliance with treatment protocol, especially offloading of plantar ulcers. It is difficult in this initial pilot study to adequately control for all these confounding variables. Nevertheless, as shown in Table 1, there was no obvious segregation of confounding variables to one of the treatment arms. For example, the wound areas at the time of enrollment were not significantly different. In fact, the average area for doxycycline-treated wounds was about 2cm2 larger than the average area for vehicle-treated wounds (5.33±4.59cm2 versus 3.47±3.48cm2, respectively, mean ± standard deviation). Six of the seven patients were men, reflecting the demographics of the VA study population. The average age of the four patients treated with doxycycline was 57 years (range from 46–68 years), while the average age of the three patients treated with vehicle was 70 years (range from 64–78 years). There was no notable difference in smoking history or in the average number of years patients had diabetes (10 years in doxycycline-treated and 11 years in vehicle-treated) in the two treatment arms. The most common co-morbid condition was hypertension, followed by coronary artery disease (CAD), and congestive heart failure (CHF). Ulcer duration in the doxycycline-treated group averaged three months, with a range from 2 to 5 months. Patients V2 and V3 had ulcers for six and seven months before enrollment, while patient V1 had an ulcer for 24 months before enrollment. Discussion The concept of treating chronic lower-extremity ulcers in diabetic patients with topical inhibitors of MMPs and TACE is based on multiple reports of elevated levels of inflammatory cytokines and proteases in wound fluids and biopsies, including diabetic foot ulcers,[7] and on the ability of the proteases to degrade growth factors and matrix proteins that are essential for wound healing.8 Therefore, it is logical to hypothesize that reducing elevated levels of TNFa and MMPs in chronic wounds should promote healing.[10] Clinical evidence supporting this hypothesis was provided by Trengove and colleagues6 who reported that levels of cytokines and proteases decreased as chronic venous ulcers began to heal, and by Ladwig and colleagues29 who reported that optimal healing of chronic pressure ulcers correlated with low values of the ratio of MMP-9/TIMP-1. Fortunately, both TACE and MMPs are members of the metalloproteinase superfamily of proteinases due to the presence of a zinc ion in the active center of the enzymes. Thus, an optimal approach would be to use a single drug that can inhibit both TACE and MMP activities. As seen in Figures 1 and 2, doxycycline is able to significantly inhibit both TACE and MMP activities in a dose dependent manner in vitro. However, the data also reveal that doxycycline is not a very potent inhibitor of TACE or MMPs, since the IC50 values are in the range of mg/mL concentrations. Fortunately, doxycycline can be formulated in physiologically compatible vehicles at concentrations in the range of 10mg/mL (1% w/v), which is 100 to 1,000 times higher than the IC50 values determined by the in-vitro reactions. Serine proteases, especially neutrophil elastase, have also been reported to be elevated in many chronic wounds, suggesting that optimal healing of chronic wounds may require the reduction of both serine proteases and metalloproteinases.[6] Although doxycycline does not directly inhibit neutrophil elastase, it may indirectly inhibit elastase activity by preventing the destruction alpha-1 protease inhibitor (A1-PI), the major natural inhibitor of elastase, by MMPs.[30] Thus, topical doxycycline may favorably alter the activities of three proteases in chronic wounds: TACE, MMPs, and neutrophil elastase. Another approach to reducing protease activities in chronic wounds is the use of dressings that combine absorbent materials, such as oxidized regenerated cellulose (ORC) or alginate, which contain high concentrations of substrates for wound proteases, such as collagen fibers. As wound fluid is absorbed into the dressing, it is proposed that the proteases bind and degrade the collagen substrate molecules in the dressing, rather than diffusing back into the wound bed and breaking down the collagen molecules, growth factors, and receptors that are essential for healing.[31] Thus, these dressings act as a sink for proteases in wound fluids. Results from a randomized controlled trial of chronic diabetic plantar ulcers treated with Promogran® collagen/ORC dressing (Johnson & Johnson Wound Management, Somerville, New Jersey) suggested a trend for increased healing.[32] A major objective of the pilot clinical study was to evaluate the safety of topical one-percent doxycycline hydrogel treatment of chronic lower-extremity ulcers in diabetic patients. Since MMPs play important roles in migration of epidermal cells and in neovascularization, it is possible that topical doxycycline might retard epithelial healing or impair formation of granulation tissue in chronic wounds.[33–35] Topical treatment of suction blisters in human volunteers with the potent MMP inhibitor, ilomastat, retarded epithelial healing about 30 percent,[36] and minocycline, a tetracycline family member, reduced angiogenesis in a rabbit corneal angiogenesis model.[37] However, all four chronic wounds treated with doxycycline formed granulation tissue and epithelialized within 30 weeks of treatment. Another concern of long-term topical treatment with one-percent doxycycline was the possibility of developing doxycycline-resistant bacteria. However, development of antibiotic resistance is favored by long-term use of an antibiotic at low levels that approach the minimal inhibitor concentration for 50-percent reduction of growth (MIC50). The MIC50 for doxycycline with susceptible bacteria is approximately 0.1mg/mL, and for intermediate sensitivity bacteria is 8mg/mL (doxycycline hyclate package insert, Mutual Pharmaceuticals, Philadelphia, Pennsylvania). The concentration of one-percent doxycycline is 10,000mg/mL, which is substantially higher than these MIC50 values, and no infections developed in the doxycycline-treated wounds. A second major objective of the pilot clinical study was to evaluate the efficacy of topical one-percent doxycycline in a CMC hydrogel on healing of chronic lower-extremity ulcers in diabetic patients. The Chi-squared analysis indicated that topical one-percent doxycycline treatment healed the chronic diabetic ulcers better than the vehicle treatment. Comparison of healing percentages in the doxycycline arm (75%) and control arm (30%) after 20 weeks of treatment were in general agreement with healing rates reported for placebo (30%) and PDGF (50%) arms in the initial becaplermin (Regranex®, Johnson & Johnson Wound Management) study of 61 patients.[38] The apparent beneficial effect of topical one-percent doxycycline treatment on healing of the chronic ulcers was probably not due to the antibiotic action of doxycycline, since bacterial levels at the time of enrollment of all the chronic wounds were below the level that is associated with impaired healing. The authors speculate that the beneficial effect on healing was due to reducing levels of TNFa, MMPs, and elastase activities in the chronic wound environment, which improved the actions of endogenous growth factors. Obviously, a much larger randomized clinical trial must be conducted to conclusively establish the safety and efficacy of topical one-percent doxycycline on healing of chronic lower extremity ulcers in diabetic patients, but this pilot study suggests that topical one-percent doxycycline is sufficiently safe and efficacious to justify further investigations.