Inhibition of PAI-1 Via PAI-039 Improves Dermal Wound Closure in Diabetes

Diabetes is a family of metabolic disorders characterized by elevated blood glucose levels and impaired insulin signaling. Currently, it is estimated that >29 million individuals in the U.S. and >347 million individuals worldwide have diabetes (1,2). By 2030, the worldwide prevalence of diabetes will approach 8% of the world’s population (3). Individuals with diabetes are at a significantly elevated risk for a number of comorbidities, including nephropathy, neuropathy, peripheral artery disease, stroke, and retinopathy. Another major complication associated with diabetes is nonhealing dermal wounds. These wounds are especially common in the lower distal extremities, namely, diabetic foot ulcerations. Foot ulcers are the leading cause of hospital admissions for persons with diabetes, are estimated to occur in 15% of all patients with diabetes, and precede 85% of all diabetic lower leg amputations (4,5). In the U.S., the average cost of treating one single infected diabetic foot ulcer is $17,000, and amputation costs approach $45,000 per amputation (5,6). Approximately 72,000 nontraumatic diabetic lower-limb amputations are performed in North America each year (5). Despite significant medical advancements in the treatment of diabetic wounds, these statistics have not faltered significantly in the past 30 years, highlighting the necessity to develop effective strategies to expedite diabetic wound healing in order to avoid amputation (6).

Wound healing in individuals without diabetes is a multifactorial process that follows a basic series of overlapping processes: 1) hemostasis (clotting); 2) inflammation (clean out debris and bacteria); 3) proliferation (rebuild wound site); and 4) maturation (regeneration of damaged tissue and vessels). In patients with diabetes, it is reported (4) that there are >100 known physiological factors that contribute to the deficits in wound healing. In general, the impairments in diabetic wound healing appear to be concentrated in the inflammatory and proliferative phases, and include the prolonged presence of neutrophils and macrophages, an impaired angiogenic response, decreased migration and proliferation of fibroblasts and keratinocytes, decreased quantity/quality of granulation tissue, and altered growth factor/cytokine expression (4).

Though diabetes is characterized by altered blood glucose regulation and dysregulation of insulin (and/or insulin signaling), numerous other endocrine factors are also known to be differentially regulated. One hormone of particular interest is plasminogen activator inhibitor 1 (PAI-1), a member of the serine protease inhibitor family. PAI-1 has actions that are largely determined through two distinct, yet interrelated, cascades. First, PAI-1 is involved in fibrinolysis via the inhibition of plasminogen activators; and second, in cell migration, survival, and proliferation, through binding to urokinase plasminogen activator, the urokinase plasminogen activator receptor vitronectin, and low-density lipoprotein-related protein. Plasma PAI-1 levels are significantly elevated in individuals with diabetes, obesity, and insulin resistance, and have been implicated in the development of vulnerable atherosclerotic plaques and nephropathy (7–9).

Based on the work of Chan et al. (10), who demonstrated that PAI-1–deficient mice display accelerated wound healing, we hypothesized that elevations in PAI-1 levels were contributing to the impairments in dermal wound healing in diabetes mellitus. Furthermore, we hypothesized that pharmacologically decreasing PAI-1 levels may serve as a useful therapeutic strategy to improve wound healing in diabetes mellitus. The results of the present investigation demonstrate that PAI-1 inhibition significantly improves epidermal closure in diabetic wounds, a process that is critical for the formation of a protective barrier atop the wound site, preventing bacterial entry and infection, and allowing the expedition of the healing process. These findings support the use of PAI-1 inhibitors as a novel therapeutic agent to improve diabetic cutaneous wound repair.

Male C57BL/6J mice (The Jackson Laboratory, Bar Harbor, ME) were provided enrichment material, chow (OpenSource Diets D12450K; Research Diets, New Brunswick, NJ), and water ad libitum. Animal housing conditions were maintained at 21°C, 50% humidity, and a 12-h/12-h light-dark cycle. Experimentation was approved by the McMaster University Animal Research Ethics Board, in accordance with the guidelines of the Canadian Council for Animal Care.

At 10–12 weeks of age, animals were randomly assigned to streptozotocin (STZ)-induced diabetic (STZ-diabetic) or control (wild-type [WT]) groups. Cohorts were then subdivided into groups with aggravated wounds and minimally handled wounds.

STZ-diabetic animals received three daily injections of STZ anomer (Streptozocin; Sigma-Aldrich, Oakville, ON, Canada) dissolved in sterile saline at 50 mg/kg, and one final injection at 200 mg/kg.

STZ (Calbiochem, Gibbstown, NJ) was dissolved in sodium citrate buffer, pH 4.5, and one injection at 150 mg/kg was administered. No significant differences were observed between groups in body mass (P = 0.49) or blood glucose values as a result of the two distinct STZ protocols (Supplementary Fig. 1).

Six weeks after diabetic onset (blood glucose >14 mmol/L), punch biopsies were conducted. Hair removal took place 2 days prior to punch biopsy. On the day of the biopsy, all mice were anesthetized via isoflurane. After surgical preparation (a sequential application of 0.5% proviodine scrub, 70% ethanol, and 1% proviodine solution), each animal received two 6-mm-diameter full-thickness wounds via punch biopsy (Miltex, York, PA) in their dorsal scapular region. Oral analgesic (Tempra; Bristol-Myers Squibb Canada, Montreal, QC, Canada) was administered 2 h prior to biopsy, and topical analgesic (Emla Cream; AstraZeneca, Mississauga, ON, Canada) was applied prior to surgical preparation. Oral analgesic was also provided to all animals every 4–6 h for the first 48 h postbiopsy.

Changes in PAI-1 levels were confirmed using the mRNA expression levels observed in cardiac tissue (11). Briefly, cDNA was prepared from cardiac tissue of WT and STZ-diabetic mice. Semiquantitative PCR was performed using sequence-specific primers for PAI-1 (ACGTTGTGGAACTGCCCTAC and GCCAGGGTTGCACTAAACAT) and β-2 microglobulin (ATCCAAATGCTGAAGAACGGG and CATGCTTAACTCTGCAGGCG), and the findings were quantified using CareStream Imager software. Consistent with the findings of others (11), PAI-1 levels were significantly elevated in STZ-diabetic mice (0.55 ± 0.03 WT vs. 0.76 ± 0.03 STZ, P < 0.01).

To determine whether elevations in PAI-1 levels were contributing to impaired cutaneous wound repair, PAI-039 (Axon Medchem, Groningen, the Netherlands), an orally effective PAI-1 inhibitor, was administered. STZ-diabetic and WT mice were randomly assigned to receive treatment with vehicle (V) (0.5% methylcellulose and 2% Tween-80 in sterile H₂O) or PAI-039 (2 mg/kg PAI-039 in V) at both 11:00 a.m. and 3:00 p.m. daily, beginning on the day of wounding and terminating on the day of harvest, 10 days postbiopsy. This PAI-039 administration protocol has previously been demonstrated to reduce PAI-1 levels to those of nondiabetic mice (12).

Animals received one of the following two variations of wound care: aggravation or minimal handling. In order to model improper wound care, wounds were manually grasped and held during treatment administration via oral gavage, twice daily. In order to model the proper care of diabetic wounds and remove wound site aggravation (13), treatment was combined with sugar-free cherry-flavored Tempra (Bristol-Myers Squibb Canada), a solution that allowed simple oral administration via dropper without handling of the dorsal scapular wounds.

Ten days postbiopsy, animals were euthanized via cervical dislocation; and wounds were isolated, bisected, fixed in 4% paraformaldehyde, processed (TP 1020 Tissue Processor; Leica Biosystems, Wetzlar, Germany), and paraffin embedded. All macroscopic images were taken with a Powershot SX 200 IX Camera (Canon, Tokyo, Japan) at a standardized height and magnification. All microscopic images were obtained with a 90i Eclipse Microscope (Nikon, Inc., Melville, NY). All analysis was completed using NIS Elements Analysis Software (Nikon, Inc.).

Healing was assessed by imaging and quantifying eschar size throughout the 10-day healing period. Animals were briefly anesthetized via isoflurane, and images were taken at a standardized height at 0, 2, 4, 6, 8, and 10 days postwounding. To obtain the percent decrease in eschar size at each time point, the eschar area was measured and transformed via Formula A (the highest and lowest values were removed).

where A_x represents the eschar area measurement taken at day 2, 4, 6, 8, or 10, and A₀ represents the wound area measurement at day 0.

The 6-μm paraffin sections were air dried overnight, deparaffinized, and rehydrated. Hematoxylin-eosin (H-E) and Masson trichrome staining was performed using standard protocols. H-E–stained sections taken from the center of the excised wounds were analyzed. If the leading epidermal edges were fully fused, wounds were considered to be fully closed. The number of fully closed wounds per group was tallied and graphed accordingly. Masson trichrome–stained sections taken from the center of the excised wounds were used to determine epidermal thickness differences between wound edges and unwounded tissue (Formula B, below), the total presence of collagen in the wound bed dermis (signal threshold settings were used as the detection method), and the dermal depth of the wound bed (six evenly spaced depth measurements provided an average wound depth):

Formula B represents the difference in thickness between the wound edges and the unwounded epidermis. The average of four cross-sectional measurements of unwounded epidermis (U₄) was subtracted from the average of eight epidermal cross-sectional measurements from the wound edges (E₈; four measurements were made on each side of the wound).

The 6-μm sections were air dried overnight, deparaffinized, and rehydrated. Heat-mediated antigen retrieval of sections took place in citrate buffer (pH 6) at 65°C for 30 min.

Sections were incubated in 0.2% Triton X-100 for 30 min, 5% normal goat serum for 40 min, and CD206 antibody (1:2,500) (Abcam, Cambridge, MA), or CD86 and NOS2 antibody (1:50 each) (Santa Cruz Biotechnology, Dallas, TX) for 2 h at room temperature. To visualize CD206, sections were incubated in goat-anti-rabbit Alexa Fluor 488 (1:250) (Abcam) for 60 min. To visualize NOS2 and CD86, sections were incubated in goat-anti-rabbit Alexa Fluor 594 or goat-anti-mouse Alexa Fluor 488, respectively (1:250) (Abcam) for 60 min. DAPI (1:10,000) was used to stain and identify nuclei. Analysis included the determination of CD206-, NOS2-, and CD86-positive areas via manual quantification. If a macrophage stained via the CD86/NOS2 costain produced a positive signal for both antibodies, it was classified as an M1 macrophage; however, if a macrophage stained positive only for CD86, it was classified as an M2b macrophage.

Sections were incubated with CD31 antibody (1:50), anti-collagen I antibody (1:200), or anti-collagen III antibody (1:500) (Abcam) for 1 h at 37°C, and biotinylated anti-rabbit IgG (1:1,000) (Vector Laboratories, Burlingame, CA) for 40 min; and were counterstained with H-E. Analysis included the determination of CD31-, collagen I-, or collagen III-positive areas using signal threshold settings as the detection method.

All statistical analyses were performed using Prism 6 (GraphPad Software, La Jolla, CA). For all analyses, apart from the analysis of epidermal closure, statistical significance was determined using a two-way ANOVA followed by Tukey multiple-comparison post hoc test. Statistical significance for the analysis of epidermal closure was determined using a Pearson χ² test. Statistical significance was defined as P ≤ 0.05.

Figure 1 highlights improvements in eschar size from the day of wounding (day 0) to the day of harvest (day 10). Significant differences in the eschar size of aggravated wounds (Fig. 1A and C) between groups at 10 days postwounding was noted. Significant differences in the percent eschar area between groups at days 2, 4, 6, and 8 were revealed in minimally handled wounds (Fig. 1B and D). Wound aggravation delayed reductions in eschar size regardless of treatment (Fig. 1A and C). The coupling of minimal handling (Fig. 1B and D) with PAI-039 treatment proved most effective in reducing eschar size. In order to glean information on the cellular processes in the dermis and epidermis of the wound, further microscopic analysis was conducted.

In order to observe the effects of PAI-039 treatment on the structure and composition of the subeschar layers of tissue, histochemical/immunohistochemical analysis was conducted on excised wound tissue 10 days postwounding. Given the role of PAI-1 in extracellular matrix remodeling, it was critical to investigate the role of PAI-039 treatment in the remodeling and restoration of the dermis. Collagen content (Fig. 2A and B) was analyzed; however, no significant differences in total collagen content were revealed as a result of PAI-039 treatment. Further collagen analysis (Supplementary Fig. 2) revealed no significant differences in either collagen I or collagen III content as a result of PAI-039 treatment. Small but statistically significant differences were noted in collagen III content in aggravated and minimally handled wounds as a result of diabetes. Information regarding the depth of the dermal portion of the wound bed (Fig. 2C and D) was gathered; however, no significant differences were revealed as a result of treatment. These results indicate that the administration of PAI-039 did not facilitate repair in the dermis.

To assess angiogenesis, immunofluorescent staining for CD31, an endothelial cell marker, was conducted on both aggravated (Fig. 3A) and minimally handled (Fig. 3B) wounds 10 days postinjury. In both aggravated and minimally handled WT wounds, PAI-039 significantly reduced CD31 content, indicating a negative effect on wound angiogenesis. In both aggravated and minimally handled wounds, STZ-V–treated animals displayed lower levels of CD31 than WT-V–treated animals, indicating a negative effect of diabetes on angiogenesis. This reached statistical significance in minimally handled wounds (P = 0.0038) but not in aggravated wounds. As expected, given its effect on wounds in WT mice, PAI-039 treatment was not seen to restore CD31 content in either aggravated or minimally handled wounds in STZ-diabetic mice.

A costain of CD86 and NOS2, which was used to visualize M1 macrophages, revealed no significant differences in either aggravated or minimally handled wounds (Fig. 4A and B). CD206 staining, which was used to visualize M2a macrophages, revealed greater amounts of CD206 in aggravated the wound beds of diabetic mice compared with those of WT mice (Fig. 4C) and were notably greater than in minimally handled (Fig. 4D) wounds, regardless of diabetic state. CD206 levels were elevated in STZ-PAI-039–aggravated wounds, and, although a significant main effect of treatment (PAI-039; P = 0.0283) was revealed, no significant interactions were observed in minimally handled wounds (Fig. 4D). CD86, which was used to visualize M2b macrophages, revealed no significant differences in aggravated wounds (Fig. 4E), and, although a significant main effect of PAI-039 (P = 0.0049) was revealed, no significant interactions were observed in minimally handled wounds (Fig. 4F). A sum total of all aforementioned macrophage quantification was calculated (M1 + M2a + M2b), and no significant differences between groups were found (Fig. 4G and H). In an effort to identify differences between groups with aggravated and minimally handled wounds in terms of the total number of macrophages, we undertook a series of unpaired t tests. While no differences between WT groups were noted, this analysis revealed wound aggravation to exacerbate the presence of macrophages, namely, as a significant increase in the sum total macrophage content in aggravated wounds in both STZ-diabetic mice and STZ-diabetic mice treated with PAI-039 (P = 0.01 and P = 0.05, respectively) compared with minimally handled wounds.

Wounds were observed to have either an uninterrupted epidermal layer, and were classified as “fully closed,” or to have an interrupted epidermal layer over the wound bed, and were classified as “not fully closed.” In aggravated wounds of WT-V–treated mice, 75% of wounds were fully closed 10 days postwounding, compared with only 17% of STZ-V–treated wounds (Fig. 5A). This dramatic decrease in the number of fully closed wounds is consistent with the negative effect of diabetes on wound repair (for review, see Greenhalgh [14]). In animals with aggravated wounds, the administration of PAI-039 to WT mice resulted in all wounds being fully closed 10 days postwounding. Importantly, a significant 300% increase in the number of fully closed wounds was observed in STZ-treated mice administered PAI-039. These data demonstrate that systemic reduction of PAI-1 levels markedly improved epidermal closure, thus, significantly improving diabetic wound healing in the presence of aggravation. In the group with minimally handled wounds (Fig. 5B), 100% of wounds in WT mice treated with V and 80% of wounds in STZ-diabetic mice treated with V were fully closed. This finding emphasizes the importance of proper wound care in diabetes but still illustrates the decreased wound-healing ability of the diabetic population. Coupled with minimal handling, the administration of PAI-039 to STZ-diabetic animals allowed for closure of 100% of wounds at day 10 postbiopsy, a 20% improvement over wounds in STZ-diabetic animals treated with V.

It has been demonstrated that diabetes negatively impacts cellular migration (4,14–17). It was hypothesized that the negative impact on cellular migration, particularly epidermal migration, would result in “thicker” wound edges, and that the increase in epidermal closure with PAI-039 treatment was the result of improved epidermal migration. As can be seen in Fig. 6, diabetes significantly impaired cell migration, resulting in a thicker epidermis at the wound edges in both aggravated and minimally handled wounds. PAI-039 treatment reversed this defect, resulting in wound edge thicknesses comparable to those of the WT cohort, regardless of the level of handling.

One of the most prominent diabetic complications is the impaired ability to heal wounds (14,18–21). In the current investigation, decreases in epidermal closure and CD31 content, as well as an increase in epidermal edge thickness exemplify the impairments in healing associated with diabetes. In 2001, Chan et al. (10) reported an accelerated rate of cutaneous wound healing in nondiabetic PAI-1–deficient mice. As previously mentioned, diabetic patients experience elevated levels of PAI-1 regardless of their level of glycemic control (10,22). Due to the increased expression of PAI-1 in the diabetic population, as well as the aforementioned improvements in WT dermal wound repair observed as a result of genetically mediated knockout of PAI-1 (10,22), investigations into the pharmacological reduction of PAI-1 levels for the restoration of diabetic wound healing were executed. To the best of our knowledge, this is the first study to analyze the effects of PAI-1 inhibition on wound repair in the diabetic state.

Epidermal migration prompts wound re-epithelialization and is consequently responsible for initiating the process of wound repair (23). Therefore, epidermal closure is the most important step in wound healing, as all other cellular proliferative and migratory events in the healing cascade follow epidermal closure (24). After wound closure, newly formed epidermis around the wound edges becomes hyperplastic due to the accumulation of cells that will migrate inward to close the wound gap (23). The thickness of this hyperplastic epidermal edge is expected to return to “nearly normal” by 5 to 7 days postinjury in a nondiabetic environment (25). Consistent with the literature (14–17), impairments in cellular proliferation and migration associated with diabetic wound healing, namely, attenuation of epidermal closure and increased epidermal thickness, have been shown in the current study. Importantly, the administration of PAI-039 to diabetic mice accelerated epidermal cellular migration, facilitating the restoration of epidermal thickness and total wound closure to a level consistent with that in wounds of WT mice.

Given the prominent role of PAI-1 in regulating the fibrinolytic pathway and in the remodeling of the extracellular matrix, a significant impact of PAI-1 inhibition on fibrosis in the regenerating wounds was expected. This expectation was consistent with the previously reported attenuation of skeletal muscle collagen accumulation in diabetic mice postinjury after PAI-039 treatment (12). What we found, however, was that PAI-039 treatment had no effect on dermal collagen content postwounding, suggesting that the mechanism of PAI-039 action does not include mediating changes in the wound dermis.

Increases in PAI-1 expression, as well as alterations in the inflammatory response (including prolonged inflammation and defective macrophage function) are associated with impaired wound healing in diabetes (10,17,19,22,24,26–28). In a genetically induced diabetic mouse model (Akita), elevated PAI-1 levels attenuated collagen turnover and, ultimately, impaired macrophage infiltration into damaged skeletal muscle. This impairment in infiltration was restored with the systemic administration of PAI-039 (29). Similarly, previous findings indicate an increase in the number of lung exudate macrophages in the presence of PAI-1, while, contrarily, the administration of a small-molecule PAI-1 inhibitor attenuated renal macrophage migration in nondiabetic animals (30,31). In the current investigation, PAI-039 treatment consistently decreased macrophage content (M2a and M2b) in minimally handled wounds. This effect was not consistently observed in aggravated wounds; however, differences between aggravated and minimally handled wounds were also observed in other ways. A lesser overall amount of CD206, a marker of M2a macrophages with a wound-healing phenotype, was observed in minimally handled wounds. As CD206 is indicative of repair and regeneration, these data, most notably in diabetic wounds, suggests that minimally handled wounds may have progressed further in the wound-healing process, and thus have fewer CD206 macrophages. Furthermore, when the sum total of all macrophages was compared between diabetic groups, aggravated wounds displayed significantly more macrophages, regardless of treatment, than their minimally handled counterparts. Taken together, however, as both aggravated and minimally handled wounds still displayed great improvements in epidermal closure in the presence of these macrophages, it is likely that the changes in dermal macrophage content did not hinder the ability of PAI-039 to facilitate wound closure.

Peripheral vasculature complications are prevalent in patients with diabetes and pose as a significant hurdle to wound healing, as the formation of vasculature in the wound bed is critical for sustaining migrating cells and newly formed granulation tissue (32,33). Elevations in PAI-1 are seen to increase angiogenesis in cancerous tissue (34,35). In noncancerous tissue, however, thickening of the basement membrane due to matrix metalloproteinase inhibition and a subsequent decrease in extracellular matrix turnover would be expected to impair the ability of endothelial cells to infiltrate the wound site. Mice overexpressing PAI-1 display significantly increased numbers of venous occlusions, resulting in necrotic tails and swollen limbs (36), and it has been known for some time that there is a connection between depressed fibrinolysis (through elevated PAI-1) and venous thrombosis (37). Collectively, it is evident that PAI-1 is involved in the regulation and dysregulation of angiogenesis. In the current study, PAI-039 treatment failed to restore CD31 content and improve angiogenesis; however, epidermal wound closure was not hindered as a result.

Although there is no animal model that fully reflects all aspects of diabetes and its complications in humans, STZ is the most widely used rodent diabetogenic chemical agent, and has been so since it was first described in 1963 (38). It is for this reason that we chose the STZ-diabetic model, as it is the most widely studied and would allow our results to be compared with the hundreds of published research studies investigating wound healing in STZ-diabetic animals. Clearly, future studies investigating other animal models of diabetes (such as the db/db or NONcNZO10 polygenic mice [39]) with and without PAI-039 would prove to be fruitful in further validating the efficacy of this inhibitor. Given the lack of concomitant vascular disease in STZ-diabetic mice, further experimentation using other models of diabetes that more closely mimic the vascular aspects of diabetes may provide further insight into the impact of PAI-039 administration on angiogenesis.

An important consideration of the current study is the fact that two models of wound care were undertaken. The continuous disruption of the wounds with routine handling would be reflective of the repeated aggravation of foot ulcers in diabetic humans who continue to ambulate and wear improper footwear after experiencing a dermal wound. In contrast, the minimal handling model exemplifies offloading, a critical component of diabetic wound care. Resting the wound site and minimizing pressure on the lesion or ulcer is, to date, the most effective intervention to promote diabetic wound healing, and is the first step in the treatment of a neuropathic diabetic injury (13,40). To avoid wound chronicity, activities that may lead to wound site aggravation should be avoided until proper healing has occurred (13). Consistent with these recommendations, the wound site aggravation used in this study exacerbated the wound-healing deficit observed in STZ-diabetic mice. This is, to our knowledge, the first histological investigation that combines pharmacological treatment with variations in wound care, highlighting the benefit to wound closure as a result of a reduction in wound site aggravation. This study also identifies the effectiveness of PAI-039 treatment in promoting wound closure to nondiabetic levels even in the presence of wound aggravation. In fact, the greatest impact of PAI-1 inhibition on epidermal wound closure was observed in aggravated wounds. These findings are of particular clinical importance, as it is estimated that more than half of diabetic individuals with foot ulcers do not follow the recommended guidelines for wound monitoring and maintenance (41).

Coupled with the absence of apparent changes in the wound dermis, the restoration of epidermal closure and epidermal edge thickness in STZ-diabetic mice as a result of PAI-039 treatment reveals the mechanism of PAI-039 action to be primarily epidermal in origin. PAI-039 treatment facilitates the migration of epidermal cells, accelerating the initiation of epidermal migratory edges and facilitating wound closure, allowing the proceeding steps in the healing cascade to commence in the presence of diabetes.

Funding. This work was supported by a Canadian Institutes of Health Research Doctoral Fellowship (D.M.D.), the Natural Sciences and Engineering Research Council of Canada (T.J.H.), and the Canadian Foundation for Innovation (T.J.H.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. I.A.R. designed the study; interpreted the results; performed animal care, sample collection, and data analysis; and wrote the initial draft of the manuscript. M.J.R. performed sample collection and data analysis, interpreted the results, and edited the manuscript. D.M.D., S.K.C., and A.N.R. performed sample collection and data analysis and edited the manuscript. T.J.H. designed the study, interpreted the results, performed animal care, and edited the manuscript. All authors contributed to the final version of the manuscript. T.J.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Keystone Symposia on Complications of Diabetes, Whistler, BC, Canada, 23–28 March 2014.