Treatment of chronic diabetic foot ulcers with adipose‐derived stromal vascular fraction cell injections: Safety and evidence of efficacy at 1 year

Abstract Diabetes affects multiple systems in complex manners. Diabetic foot ulcers (DFUs) are a result of diabetes‐induced microarterial vessel disease and peripheral neuropathy. The presence of arteriosclerosis‐induced macroarterial disease can further complicate DFU pathophysiology. Recent studies suggest that mesenchymal stromal cell therapies can enhance tissue regeneration. This phase I study was designed to determine the safety and explore the efficacy of local injections of autologous adipose‐derived stromal vascular fraction (SVF) cells to treat nonhealing DFUs greater than 3 cm in diameter. Sixty‐three patients with type 2 diabetes with chronic DFU—all amputation candidates—were treated with 30 × 106 SVF cells injected in the ulcer bed and periphery and along the pedal arteries. Patients were seen at 6 and 12 months to evaluate ulcer closure. Doppler ultrasounds were performed in a subset of subjects to determine vascular structural parameters. No intervention‐related serious adverse events were reported. At 6 months, 51 subjects had 100% DFU closure, and 8 subjects had ≥75% closure. Three subjects had early amputations, and one subject died. At 12 months, 50 subjects had 100% DFU healing and 4 subjects had ≥85% healing. Five subjects died between the 6‐ and 12‐month follow‐up visits. No deaths were intervention related. Doppler studies in 11 subjects revealed increases in peak systolic velocity and pulsatility index in 33 of 33 arteries, consistent with enhanced distal arterial runoff. These results indicate that SVF can be safely used to treat chronic DFU, with evidence of efficacy (wound healing) and mechanisms of action that include vascular repair and/or angiogenesis.

healing. Five subjects died between the 6-and 12-month follow-up visits. No deaths were intervention related. Doppler studies in 11 subjects revealed increases in peak systolic velocity and pulsatility index in 33 of 33 arteries, consistent with enhanced distal arterial runoff. These results indicate that SVF can be safely used to treat chronic DFU, with evidence of efficacy (wound healing) and mechanisms of action that include vascular repair and/or angiogenesis. infection, and hospitalization, often ending in amputation. 2,3 DFUs in the context of combined neuroischemic disease exhibit worse outcomes. 4 In the United States, 54% of all amputations are diabetes related, and in 85% of cases the precipitating factor is a DFU, resulting in a cost of $9 to $13 billion annually. 5 Peripheral vascular disease (PVD) is a known cause of ischemic ulcers and is also an aggravating condition for DFU. PVD, either alone or in combination with diabetes, often culminates in recurrent, nonhealing ulcers and amputations. 6 Approximately 50% of patients with DFU have concurrent vascular disease. 7 As surgical revascularization is not always feasible in these patients, an urgent need exists for the development of alternative therapies capable of improving blood supply to the ischemic foot.
Cell-based therapies have gained attention as viable options to provide the required elements to help restore damaged vessels while inducing the formation of new ones. 8 Cell products may contain endothelial progenitor cells (EPCs) and/or mesenchymal stem/stromal cells (MSCs), both critical during vascular repair and formation given the structural participation of the former and the documented proangiogenic activity of the latter. 9 Based on the individual cell-type capabilities documented for EPCs and MSCs, the use of a combinatorial cell approach in the same product constitutes an interesting alternative to treat vascular disease.
Multiple small clinical studies have used autologous or allogenic bonemarrow mononuclear cells, either directly after bone marrow harvest or after tissue culture, to treat critical limb ischemia. In general, therapy resulted in improved symptoms (decreased pain) 10 and in some studies improved ankle/brachial index and/or tissue oxygenation. 11,12 Adipose tissue-derived stromal vascular fraction (SVF) stands as a viable option to treat vascular disease, given its EPC enrichment and higher titers of MSCs when compared with other sources (eg, bone marrow). [13][14][15] Logistical advantages complement this key multiphenotypic display, as SVF cells can be obtained from a same-day processing of readily accessed and harvested adipose tissue without the need of a good manufacturing practice (GMP) processing facility to manufacture an MSC-based product, thus making SVF a "point-of-care" therapy.
It is difficult to treat vascular disease and chronic ulcers caused by PVD and/or diabetes in resource-poor countries such as

Lessons learned
• Stromal vascular fraction (SVF) is capable of inducing new blood vessels under ischemic conditions.
• Local administration of SVF produces accelerated wound healing in wounds.
• The rate of wound healing does not correlate with wound size.
• SVF administration distal to the terminal arteries of the foot (tibialis anterior and tibialis posterior) is associated with changes in wave form and flow velocity consistent with new blood vessel induction and reduced distal resistance.

Significance statement
Microvascular disease seen in diabetes has no effective form of surgical treatment. This article reports the possibility of a novel therapy for diabetic wounds based on blood vessel induction in situ that can alleviate the intractable pain and/or infection associated with these chronic wounds, conditions that are amenable only to amputation. By using stromal vascular fraction (SVF) injection, surgeons can prevent limb loss, an outcome with devastating socioeconomic consequences for both the patient and society. The SVF protocol presented is easy to execute and can be carried out quickly and safely as an ambulatory procedure, under conditions in the developing world. The primary inclusion criteria were active type 2 diabetes requiring treatment, a nonhealing ischemic ulcer of the lower extremity ≥3 cm 2 , the ulcer having longer than 3 months duration, and clinically approaching the need for amputation. Criteria for exclusion were age < 30 years, unstable cardiovascular disease at the moment of enrollment, smoking and/or the presence of chronic pulmonary disease, ongoing infection and/or sepsis, and uncontrolled diabetes.
Because subjects having multiple ulcers at the time of enrollment was not addressed by either the inclusion or the exclusion criteria, three subjects were enrolled having two concomitant ulcers on the same foot. The largest ulcer was chosen as the study ulcer. Usual care for these diabetic patients with DFU prior to entry into the study consisted of saline gauze changed twice a day and follow-up with their local health care provider/clinic. Patients were advised to avoid weight bearing on the affected limb to the extent possible.
The primary endpoint of this clinical study was the safety of the GID SVF-2 device processing and associated SVF injection therapy.
Safety was determined based upon reported adverse effects secondary to the adipose tissue harvesting procedure (hematoma, infection, profuse bleeding), the injection of the cell product (vessel puncture, intravascular administration of the cell preparation with potential phlebitis and/or distal embolism), and the therapy itself (secondary local infection requiring antibiotics). The secondary endpoint was to explore efficacy as percent wound closure at 1 year. Wound closure was defined as intact epithelial coverage without need for further dressing changes. An exploratory question was to evaluate if DFU size limited the efficacy of SVF as measured by wound closure. In addition, in a subset of patients at one site, the effect of SVF injection on pedal artery supply was studied via spectral color Doppler ultrasound.

| Isolation of SVF (liposuction and cell processing)
The surgical procedure consisted of preparation of the harvest site with local anesthesia and tumescent solution of Ringer's lactate followed by lipoaspiration into a closed adipose tissue processing system-the GID SVF-2 device (The GID Group). The lipoaspirate collected in the SVF-2 device was then processed by (a) serial washing to restore pH and remove oil, leukocytes, and erythrocytes, thus producing "dry fat" free of fluids; (b) enzymatic digestion with GMP-grade collagenase; (c) neutralization by repeat washing with lactated Ringer's; (d) aspiration of the SVF pellet from the cell collecting chamber of the device; (e) resuspension of the SVF cells; and (f ) cell count. Details of these procedures are provided in our original report. 16

| SVF cell injection procedure
Clinical application of the cells was carried out after obtaining a cell count of the SVF in the form of total nucleated cells and with a cell viability greater than 85%. The target foot was injected with a total dose of 30 × 10 6 SVF cells. The cells were delivered in a total volume of 60 cc of lactated Ringer's distributed as follows ( Figure 1A Patients were discharged with instructions to continue the same regimen. No antibiotics were prescribed. A wound care follow-up was done at 7 to 10 days.

| Study follow-up
A focused clinical examination was performed at 1, 3, 6, and 12 months post-treatment. The wounds were evaluated and photographed at 6 and 12 months. Questions were asked regarding symptoms, changes in medications, and any potential adverse event. Dressing materials were provided and instructions for home care reinforced.

| Ultrasound analysis
In order to assess evidence of neovascularization, color Doppler ultrasound was used at one site (Matagalpa Hospital) to document 1-year changes in wave form, blood flow, and elasticity in the dorsal circulation (tibialis anterior and dorsalis pedis) and the plantar circulation (tibialis posterior). Peak systolic velocity (PSV) corresponds to the measured maximum height of the velocity tracing as seen in the spectral window. Pulsatility index (PI) is a calculated flow parameter used to assess the resistance in a pulsatile vascular system consistent with adequate perfusion. It is derived from measurement of three frequency shifts through the cardiac cycle, maximum, minimum, and mean, by the following equation: (peak systolic velocity − minimal diastolic velocity)/mean velocity. Normal vs pathological values for PI are follows: femoral artery, 4 to 6 vs ≤4; popliteal artery, 6 to 12 vs <6; tibialis anterior and tibialis posterior, 7 to 12 vs <7. 20

| Statistical analysis
Descriptive statistics (means, frequencies) were used to summarize demographic characteristics of the study population. Correlation coefficient was calculated with the Spearman rank correlation test.   Table S1.

| Safety
No serious adverse events were observed/reported related to the liposuction procedure, the SVF-2 device, or the accompanying SVF injections. Before the 6-month time point, three patients underwent amputation within the first 2 weeks of the procedure by a nonprotocol surgeon based on previously scheduled surgery, and one additional patient passed away from cardiac causes <2 months after the intervention. Over the course of 1 year, in this high-risk population with advanced diabetes, six more patients died from cardiac causes between 6 and 12 months. Study team evaluation concluded that these serious adverse events (deaths) were unrelated to the SVF treatment as they were all ascribed to the underlying disease. No concerns were raised regarding these adverse events by the Ministry of Health.

| Wounds
As described in patient characteristics, the 63 patients had wounds ranging in size from 9 cm 2 (3 cm × 3 cm) to 120 cm 2 (15 cm × 8 cm) with an average wound size of 32.9 cm 2 ( Figure 2A shows the frequency distribution of ulcer size). At 6 months, 59 of the 63 subjects enrolled were evaluable for closure (as above, three subjects had amputation, and a fourth subject died less than 2 months after the intervention from unrelated causes). Of the evaluable subjects, 51 achieved closure (ie, 86% of subjects had complete closure; 95% confidence interval = 0.745-0.936). Of the remaining eight subjects, three achieved 95% closure, three achieved 85% closure, and two achieved 75% closure. At 12 months, 54 of the enrolled 63 subjects were evaluable for closure (four additional deaths occurred between 6 and 12 months).    (Table S2).

| DISCUSSION
Arteriosclerosis and diabetes contribute to the pathophysiology of DFU. The presence of ischemia because of underlying PVD negatively affects the outcomes of DFU, evidenced in lower probability and longer duration to heal, ulcer recurrence, and risk of amputations. 4,6 F I G U R E 4 Peak systolic velocity and pulsatility index. Peak systolic velocity (A) and pulsatility index (B) assessed in tibialis anterior, dorsalis pedis, and tibialis posterior by Doppler spectral color ultrasound, comparing pre-and postinjection results, and evidencing statistically significant differences in all arteries examined. PSV, peak systolic velocity Consequently, therapeutic efforts must be directed at preventing or reversing ischemic conditions in the foot.
Adipose-derived SVF is a heterogeneous cell product composed of different endothelial cell populations, including progenitor and mature cohorts discriminated by the expression of CD34, added to hematopoietic and other cells with perivascular and MSC phenotypes (ie, pericytes and supra-adventitial cells, respectively) and monocytes/ macrophages. 23 The angiogenic and vasculogenic potential of SVF has been documented both in vitro 24 and in vivo in models of ischemic limb 25 and refractory wound healing. 21 In addition, Han et al demonstrated complete diabetic wound healing in a 26-patient group treated with SVF with no reported adverse events. 21 For this phase I safety study, we focused on chronic, nonhealing DFU (>3 cm 2 ) in a population of patients with type 2 diabetes and underlying microangiopathy. In our published 2017 study, the SVF dose used to treat PVD ranged from 30 × 10 6 to 158 × 10 6 SVF cells. 16 Based upon the observed clinical responses even for the lowest dose and taking into consideration the smaller injecting area in the current study, we used a fixed cell dose of 30 × 10 6 SVF cells for this study. This dose was shown to be safe and demonstrated clear efficacy with closure response rates among evaluable patients of between 86% and 93% at the 6-and 12-month endpoints. The wound healing process was observed to heal by two different directions: from the periphery, as expected, but also by upward proliferation from the ulcer bed. In several cases, newly developed tissue was capable of covering previously exposed tendons. Furthermore, even among ulcers greater than 10 cm 2 , virtually all patients achieved 85% closure or better by 6 months. No correlation between ulcer size and closure was observed.
Although it was not a specified endpoint, this study had an amputation rate of 4.7%. In comparison to reported DFU "standard-of-care" healing, this study generally showed improved ulcer healing with SVF treatment. Riaz  Health Administration (VHA) and non-VHA care settings. 28 Overall, the VHA rate of healing was 76%, with a time to heal of 10.9 ± 10.5 weeks. The amputation rate was 18.3%. Non-VHA rate of healing was 85%, with a time to heal of 10.6 ± 13.6 weeks. The amputation rate was 10.6%. The differences in these populations were not found to be statistically significant.
Multiple small studies have examined the use of adipose stem cells to treat chronic ulcers. 29 These studies include adipose stem cells obtained by different methods, different routes and treatment doses, and different causes of the chronic ulcers. Although the time points for healing evaluation also vary between studies, the reported healing rates are between 67% and 100%, and where control groups were included, the adipose stem cell treated group had a higher healing rate than the control group. The results of this study are comparable to the other adipose stem cell treatment healing rates.
The SVF cells in this study were locally administered along the vascular trajectories distally feeding the foot, in an effort to "concentrate" the cell product around the diseased arteries, instead of the intramuscular route used in our previous PVD study. In terms of the imaging used to document the neovascularity responses, MRI-based angiography offers a direct visualization of the arterial tree as used in our pilot study. However, with the larger number of patients in this study, this technology proved impractical as a research tool in Nicaragua's health system. Spectral color Doppler ultrasound was chosen as it demonstrates physiologic evidence for distal blood vessel formation, in the form of dampening patterns in the wave form and in the augmentation of blood flow as measured by the peak systolic velocity (cm/s). Moreover, improved elasticity of the arterial walls was evidenced by changes in the PI. 30 This study provides further evidence for physical changes in the arterial walls caused by the angiogenic effects of SVF cells based upon the increases in blood flow observed in the pedal arteries as well as changes in the vessel walls positively affecting arterial compliance.
A potential drawback to this study was the lack of a treatment control group, important to determine therapeutic efficacy. However, given the state of these chronic nonhealing DFU and impending amputations as the last therapeutic alternative, the local ethics committee determined that including a "standard-of-care" control group was not appropriate. Additionally, this phase I study was to confirm the safety of the SVF-2 device and the accompanying SVF injection procedures. Evidence of efficacy was a secondary endpoint. It was recognized that subsequent randomized, controlled, larger trials would be needed after meeting the required ethical considerations. In addition, vascular studies should include both limbs for comparison of treated/untreated distal blood vessel changes. This would permit assessment of a prophylactic effect in the nonulcerated diabetic foot.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.