Therapeutic potential for mesenchymal stem cell transplantation in critical limb ischemia

The therapeutic potential of mesenchymal stem cell (MSC) transplantation for the treatment of ischemic conditions such as coronary artery disease, peripheral arterial disease, and stroke has been explored in animal models and early-phase clinical trials. A substantial database documents the safety profile of MSC administration to humans in a large number of disease states. The mechanism of the therapeutic effect of MSC transplantation in ischemic disease has been postulated to be due to paracrine, immunomodulatory, and differentiation effects. This review provides an overview of the potential role of MSC-based therapy for critical limb ischemia (CLI), the comparison of MSC cellular therapy with angiogenesis gene therapy in CLI, and the proposed mechanism of action of MSC therapy. Preclinical efficacy data in animal models of hindlimb ischemia, current early-phase human trial data, and considerations for future MSC-based therapy in CLI will also be discussed.


Introduction
Mesenchymal stem cell (MSC) transplantation has been proposed as a novel treatment approach for tissue engineer ing and regenerative medicine for various disease states. MSC-based therapy has been explored in preclinical animal models and recently has been used in early clinical trials for ischemic disorders, including stroke, coronary artery disease and peripheral arterial disease (PAD). In this review, we discuss the comparison of MSC cellular therapy with angiogenesis gene therapy in critical limb ischemia (CLI), the possible mechanism of action and safety profile of MSC therapy. We also highlight the potential role of MSC for the management of patients with CLI by describing the relevant preclinical and early clinical trial results. We conclude by discussing the several practical considerations for future clinical trials.

Peripheral arterial disease: unmet clinical need
Up to 10% of the population in the Western world suffers from PAD and this represents a major health problem [1]. The prevalence of PAD has increased exponentially due to the increase in the prevalence of diabetes mellitus (DM) and an aging population. The increasing prevalence of PAD has resulted in a substantial increase in the consumption of health-care costs [2]. DM is prevalent in patients with PAD. In fact, DM itself increases the risk of lower-extremity PAD by two-to fourfold [3]. Poor glycemic control is associated with an increased risk of PAD independently of other known cardiovascular risk factors [4]. Individuals with poor glycemic control (A1c >7.5%) are five times more likely to develop intermittent claudication and be hospitalized for PAD as compared with those with better glycemic control (A1c <6%) [4]. In fact, 1% increment in hemo globin A1c in patients with Type 2 DM correlates with a 28% increase in the risk of PAD [5].
Patients with PAD may be asymptomatic or suffer from intermittent claudication, ischemic ulceration, rest pain or limb loss. CLI is the most advanced clinical stage of PAD. It is defined as rest pain or impending limb loss secondary to an objectively proven arterial occlusive disease for more than two weeks. The current treatment options aim at improving distal arterial perfusion by endovascular or surgical approaches or a combination of the two [1]. However, amputation is often inevitable in the majority of patients because of co-morbidities or unsuitable vasculature, and since these patients have no alternative therapeutic options, they have been termed no-option patients. These patients also had a 20% mortality within six months [1]. Hence, this condition represents an unmet clinical need. Cell trans plantation has been suggested as a possible approach for the treatment of CLI. A variety of cell types have been proposed. Currently, clinical trials using cells from both the autologous and allogeneic sources for the treatment of CLI either have been completed or are under way.
There is a particular interest in the use of both fractionated and unfractionated bone marrow cells as well as MSCs derived from various sources. Table 1 shows the current registered clinical trials at ClinicalTrials.gov [6] for various stem cell therapies for CLI.
Prior to the era of cellular therapy, gene therapy was proposed as a therapeutic option for CLI. Several phase 2 gene therapy clinical trials -with vascular endo thelial growth factor (VEGF), del1, hypoxia-inducible factor 1a (HIF1a)/VP16, hepatocyte growth factor, and fibroblast growth factor 1 (FGF1) -have been completed and demonstrated their safety and feasibility in patients with PAD [7]. A phase 3 trial with FGF1 was recently completed but did not reach the combined primary outcome of reduction in major amputation or death [8]. The comparison between MSC cellular therapy and angiogenesis gene therapy for patients with CLI is summarized in Table 2.

Mesenchymal stem cells
MSCs are multipotent non-hematopoetic, fibroblast-like plastic adherent cells that can be isolated from various tissue sources, including the bone marrow, adipose tissue, placenta, and umbilical cord blood [9]. They are capable of differentiating into different cell types such as bone, fat, cartilage, and muscle and demonstrate specific surface antigen expression [9]. In addition to possibly under going cell differentiation, MSC transplantation may exert its therapeutic effects via secretion of paracrine factors that may have anti-inflammatory and immunomodu latory effects [10,11]. MSCs are relatively resistant to apoptosis induced by conditions such as serum starvation and hypoxia [12]. In fact, MSCs exposed to hypoxia produce more VEGF in vitro, and transplantation of hypoxic preconditioned MSCs into murine ischemic limbs has been reported to lead to an increase in vessel density [13]. MSCs can also home to hypoxic muscle following intravenous administration [14]. Even though MSCs represent a very small fraction of the total population of nucleated cells in the bone marrow (0.001% to 0.01%), they can be culture-expanded readily to yield a large number of cells. These cells can be stored for both autologous and allogeneic use. The latter may be possible due to the immune-privileged status of these cells. This property offers substantial practical advantage in the clinical setting, especially when autologous cell transplan tation would be ineffective because of diseaseinduced cell dysfunction [15]. Furthermore, MSCs can be programmed to become a specific differentiated cell type prior to transplantation, thereby decreasing the likelihood of aberrant differentiation of MSCs after transplantation.

A. Paracrine effects
MSCs can be isolated from a variety of sources, including bone marrow, adipose tissue, and induced pluripotent stem cells (iPSCs) [16,17]. The conditioned media from the undifferentiated MSCs promote in vitro angiogenesis and migration [18,19]. As compared with bone marrowderived MSCs, adipose tissue-derived MSCs secrete more VEGF, hepatocyte growth factor, and transforming growth factor-beta [20], and the iPSC-derived MSCs secrete more stromal-derived factor 1 alpha (SDF-1alpha), hepatocyte growth factor, stem cell factor, basic nerve growth factor, basic fibroblast growth factor, and VEGF [16]. The conditioned media from adipose tissuederived MSCs also have higher matrix metallo proteinase 3 (MMP3) and MMP9 and showed enhanced in vitro tube formation in comparison with that from bone marrow-derived MSCs [17]. The MSC conditioned media can augment in vivo angiogenesis [21]. Under hypoxic conditions, bone marrow-derived MSCs produced more VEGF in comparison with normoxic conditions, and transplantation of these MSCs into murine ischemic limbs led to an increase in vessel density [13]. The MSC conditioned media can also enhance proliferation of endothelial cells and smooth muscle cells in a dosedependent manner [22]. MSCs stimulate capillary morpho genesis via distinct proteolytic mechanisms [23]. Cross-talk occurs between MSCs and endothelial cells. MSCs attenuate myocardial ischemic reperfusion injury by secreting exosomes [24]. Under hypoxic conditions, bone marrow-derived MSCs secreted higher amounts of VEGF with enhanced proliferative capacity as compared with fibroblasts [13]. Co-culturing of MSCs with  [23,25]. Bone marrow-derived MSCs have also been shown to be attracted to in vitro formed vascular structures [26]. These studies supported the notion that MSCs exert the angiogenic properties via paracrine and autocrine effects, and the intensity is dependent on the MSC source.

B. Differentiation effect
There is considerable doubt that the mode of action of MSCs occurs via in vivo differentiation given that only small numbers of cells engraft at the site of ischemia following intravenous administration and those that do are rapidly lost from the site. In the mouse models of myocardial infarction, the majority (83%) of the xenogeneic MSCs were found in the lung at one hour after intravenous administration [27]. Another study showed that, after intravenous administration, allogeneic MSCs were detected for up to 20 days and completely undetectable at 40 days because of graft rejection but that syngeneic MSCs survived up to 40 days [28]. This result suggests that an allogeneic immune response may occur after MSC transplantation. On the other hand, xenogeneic MSCs administered intramuscularly to nonischemic thigh muscle remain confined to the site of injection; the highest level was detected after one day, and cells were detectable for up to three weeks [29]. These results suggested that small numbers of MSCs engraft in the ischemic site following intravenous administration, that the beneficial effect of allogeneic transplantation may be attenuated by graft rejection, and that local administration may be the optimal approach for MSC-based therapy for CLI. However, there is evidence that MSCs can acquire myogenic and endothelial properties.

i. Myogenic differentiation
MSCs derived from adipose tissue, bone marrow, and synovial membrane are capable of inducing skeletal muscle regeneration, and adipose tissue-derived MSCs are the most efficient [30]. Adipose tissue-derived MSCs can enhance muscle regeneration even in dystrophindeficient mice following intramuscular administration [31]. Hypoxia preconditioned MSCs can further enhance skeletal muscle regeneration [32]. MSCs can also differentiate into cardiomyocytes and these prediffer en tiated MSCs further augment cardiac regeneration more efficiently than undifferentiated cells [33]. MSCs derived from bone marrow, iPSCs, and adipose tissue were also capable of differentiating into smooth muscle cells [16,34,35]. However, iPSC-derived MSCs differ entiate into smooth muscle more efficiently than bone marrow-derived MSCs [16]. MSC-derived smooth muscle cells have been used to engineer small-diameter vessel wall grafts [35]. In a dog model of peripheral and coronary artery bypass, canine grafts were able to maintain their patency and prevent dilatation, calcification, and intimal hyperplasia [36]. Furthermore, human engineered grafts were successfully made and tested in a baboon model of arteriovenous access for hemodialysis [36]. Currently, a functional urinary bladder tissue is being engineered by using a combination of smooth muscle cells and urothelium-like cells derived from human bone marrow-derived MSCs [37].

ii. Endothelial differentiation
MSCs derived from bone marrow, iPSCs, dental pulp, amniotic fluid, and adipose tissue have been reported to undergo endothelial differentiation [16,34,38,39]. Direct com pari son of umbilical cord-derived MSCs and bone marrow-derived MSCs showed that both cell types expressed MSC-specific markers and demonstrated trilineage differ entiation ability with the ability to take up low-density lipoprotein following endothelial differentiation. However, umbilical cord-derived MSCs had higher proliferative potential and higher expression of the endothelial-specific factors and were able to form more capillary networks than bone marrow-derived MSCs [40].
Endothelial differentiation of MSCs can be stimulated by growth factors, including VEGF, and shear force [38,39,41]. We have shown that overexpression of Ephrin-B2 in MSCs resulted in an earlier endothelial differen tiation with simultaneous reduction of osteogenic potential [42]. Endothelial differentiated MSCs have diminished capacity to differentiate into adipocytes, and subcutaneous implantation of these cells in collagen plugs in immunodeficient mice resulted in the formation of functional blood vessels incorporating these cells [43]. In response to hypoxia, these differentiated cells secrete angiogenic factors (VEGF, placental growth factor, and hepatocyte growth factor) [38].
There is a complex cross-talk between MSCs and endothelial cells. MSCs increase endothelial cell proliferation and migration, promoting early events of angiogenesis and decrease endothelial cell monolayer permeability. In direct co-culture with endothelial cells, MSCs increase the persistence of pre-existing vessels in a time-and dose-dependent manner, and complex vessels remain stable for more than 10 days [44]. The conditioned media from MSCs also stimulate the proliferation of the local endothelial cells [19]. MSCs exposed to epidermal growth factor enhance adhesion and migration on cultured endothelial cells [45]. Co-culturing of bone marrowderived endothelial progeni tor cells (EPCs) and MSCs upregulate angiogenesis-related transcripts and result in the formation of elon gated structures after three days, even with serum starvation and the absence of growth factors [46]. Besides exhibiting direct contact, these cells exhibited vesicle transport phenomena [46]. Both MSCs and EPCs contributed to these tubule structures [46]. Coculture of MSCs with macro-vascular endothelial cells led to an increase in expression of both endothelial and smooth muscle cell markers. On the other hand, coculture with micro-vascular endothelial cells increases the expression of endothelial cell markers only [47]. Secreted frizzled-related protein-1 enhances MSC function in angiogenesis and contributes to the maturation of new vessels [48].

C. Immunomodulatory effect of mesenchymal stem cells
A large body of literature supports the immune-modulatory properties of MSCs. It is outside the scope of the present article to review this in detail, but the topic was recently reviewed [49]. In a rat model of acute myocarditis, allogeneic administration of fetal membranederived MSCs attenuated the host cell-mediated immune response [50]. In a rat model of intra-cerebral hemorrhage, intra-cerebral administration of umbilical cordderived MSCs attenuated inflammation and promoted angiogenesis, leading to earlier neurological function recovery [10]. In a mouse model of hind-limb ischemia, xenogeneic intramuscular administration of MSCs attenu ated the local oxidative stress and endothelial inflammation [51]. Furthermore, intramuscular injection of adipose tissue-derived MSCs can reduce local inflammation in the dystrophin-deficient mice [31]. Intravenous administration of MSCs to brain-injured rodents reduces injury-induced enhanced blood-brain barrier permeability, thereby reducing the associated inflammatory response [52]. The immunosuppressive role of MSCs is promoted by CD14 + monocytes [53].
Despite the wealth of evidence that these cells have immune-modulatory properties, a recent publication demonstrated that allogeneic MSC administration was not completely immune-privileged as compared with syngeneic MSCs [28]. Of note, the optimal time for functional benefit of MSC transplantation after myocardial infarction was one week given that the absence of scar formation and the reduction in inflammation at this time point facilitate integration of transplanted cells, leading to functional recovery [54]. This result is supported by another study, in which MSC transplantation improved cardiac function, reduced the apoptosis of cardio myocytes, and increased vessel density much better when administered at one week but not within one hour or after two weeks [55]. In a mouse model of hindlimb ischemia, administration of syngeneic MSCs or conditioned media immediately after induction of hindlimb ischemia did not improve revascularization but did do so when administered one day after induction of ischemia [22]. These studies suggest that local inflammatory processes can impede the therapeutic efficacy of MSC transplantation and that the optimal timing of administration is crucial.

Current safety profile of mesenchymal stem cells
A substantial amount of evidence supports the safety of administration of human MSCs in a variety of disease states, and this has been reviewed by Ankrum and Karp [56]. Selected examples demonstrating safety of MSC administration will be reviewed here. The study with the longest period of follow-up assessed the autologous intra-articular bone marrow-derived MSC transplantation for cartilage repair for up to 11 years and 5 months [57]. No tumors or infections were reported in this cohort of 41 patients with MSC administration to 45 joints [57].
Yamout and colleagues [15] used intrathecal administration of ex vivo-expanded autologous bone marrowderived MSCs for the treatment of multiple sclerosis. This phase 1 trial showed that intrathecal administration of a mean dose of 3 to 5 × 10 7 MSCs per patient was safe and feasible. Concomitant intrathecal and intravenous administration of MSCs with a mean dose of 63.2 × 10 6 in patients with amyotrophic lateral sclerosis with a followup period of up to 25 months was also safe and feasible [58].
A five-year follow-up study of intravenous autologous administration of two doses of 5 × 10 7 culture-expanded autologous bone marrow-derived MSCs into 16 patients with severe middle cerebral artery territory infarction has been completed. A significant clinical improvement was demonstrated in the MSC-treated group and this improvement correlated with serum SDF-1 levels and the extent of the stroke. No serious adverse effects or increase in the incidence of seizures or recurrent vascular events was observed [59].
Intravenous administration of two doses of cultureexpanded autologous bone marrow-derived MSCs at a dose of 1 to 2 × 10 6 MSCs/kg seven days apart in patients with refractory Crohn's disease appeared to be safe and feasible [60]. In addition, intravenous administration of allogeneic MSCs with a dose escalation of 0.5, 1.6, and 5.0 × 10 6 bone marrow-derived MSCs/kg body weight was shown to be safe and feasible in patients with acute myocardial infarction [61]. Interestingly, MSC therapy also led to an improvement in pulmonary function and a lower incidence of arrhythmias in this cohort [61].
in mouse/rat models of hind-limb ischemia ( Table 2). Each paper will not be reviewed in detail, but the accumulated evidence suggests that MSCs represent an attractive target to advance to clinical trials in humans. Although these studies demonstrate the efficacy of MSCs, there is a suggestion that MSCs derived from different sources may have variable in vivo therapeutic effects. iPSC-and adipose tissue-derived MSCs were more efficacious in therapeutic revascularization than bone marrow-derived MSCs [16,17]. On the other hand, the fetal membrane-derived and bone marrow-derived MSCs demonstrated comparable efficacy for improvement in blood perfusion and capillary density [11].
In addition to the efficacy of transplantation of unmodified MSCs described above, the effect of modified MSCs, such as the exposure to the cells to hypoxia prior to transplantation, has been explored. Rosova and colleagues [62] demonstrated that inter-ventricular adminis tration of normoxic and hypoxic preconditioned human bone marrow-derived MSCs restored blood flow following induction of hind-limb ischemia and that earlier improvement was observed in the hypoxic preconditioned group. However, in a subsequent study using intramuscular administration to a similar model of hind-limb ischemia, hypoxic preconditioned MSCs were superior, and no difference between the MSCs cultured in normoxia and controls was detected, suggesting that non-preconditioned MSCs do not improve blood flow [63]. Various approaches used to augment the therapeutic efficacy of MSCs are listed in Table 3 [11,12,16,17,20,21,32,34,45,51,.
A critical factor in the design of human trials is that of the cell dose to be administered. To date, various doses of MSCs derived from different sources were administered to rodent models of hind-limb ischemia and resulted in significant therapeutic revascularization. A dose of one million MSCs has been the most commonly used in these preclinical studies. Another key consideration for human translation is the route of administration. Most preclinical studies reported to date have transplanted the cells via the intramuscular route.
The timing of administration in relation to the induction of ischemia and the number and site (or sites) of administration also varied in different studies. The cells were administered most commonly 24 hours after induction of hind-limb ischemia. The sites of intramuscular injections vary from the medial thigh alone or a combination of different sites such as the gastrocnemius, tibialis anterior, hamstring, and the adductor muscle groups (Table 3).
In addition to cell dose, cell type, and timing and route of administration, the endpoints to be assessed are crucial. Most of these studies used a combination of laser Doppler perfusion imaging, in vivo functional assessment (which included the ambulatory score and necrotic score), and histological assessment. However, in aggregate, these preclinical studies have demonstrated the proof of principle that MSCs derived from various sources are therapeutically effective in models of hindlimb ischemia in many different rodent species and strains. In addition, there appears to be a substantial preclinical safety profile. This provides a substantial impetus for the progression of MSC-based therapy into clinical trials for patients with CLI.

Clinical data
The preclinical efficacy and toxicology data reviewed above have provided a platform for the initiation of clinical trials of MSCs in CLI. The first reported human study using intramuscular administration of allogeneic human umbilical cord-derived MSCs was conducted in four patients with Buerger's disease. Allogeneic umbilical cord-derived MSCs improved ulcer recovery time, enhanced limb perfusion, and relieved the symptoms of rest pain [82]. Dash and colleagues [84] have shown that intramuscular administration of autologous bone marrowderived MSCs to patients with non-healing ulcers accelerated ulcer healing and improved pain-free walking distance. In this cohort, nine patients with Buerger's disease and three with diabetic foot ulcers were included.
Whereas the two reports above used intramuscular cell delivery, the first study using intravenous administration of autologous bone marrow-derived MSCs was recently published. That report was of a single case of a patient with systemic sclerosis who developed acute gangrene of the upper and lower limbs, and three intravenous pulses of autologous bone marrow-derived MSCs were administered. The areas of necrotic skin were reduced following the first infusion. Following the third infusion, the revasculari za tion of the patient's extremities was confirmed by angiography. The angiogenic role of MSC therapy was confirmed microscopically by using histological analysis of the skin section, which showed cell clusters with tube-like structures with high expression of multiple angiogenic factors [85].
Lasala and colleagues [86,87] assessed 40 intramuscular injections per patient of a combination of up to 30 × 10 6 bone marrow-derived MSCs and 30 × 10 8 bone marrowderived mononuclear cells. Walking time, ankle brachial pressure, and quality of life were improved with no reported adverse events after a mean follow-up period of 10 months. The angiogenic effect of MSCs was confirmed by both the angiographic and 99mTc-TF perfusion scintigraphy scores. These patients had diabetes mellitus with moderate to severe PAD (Fontaine class IIb to IV).
Lu and colleagues [88] compared the therapeutic effect of autologous intramuscular administration of bone marrow-derived MSCs with bone marrow-derived mononuclear cells in 20 patients with diabetes and severe PAD (Fontaine class IV). The authors showed that the ulcer healing rate was significantly higher in the bone marrow-derived MSC group than the bone marrowderived mononuclear cell group at six weeks. Furthermore, the bone marrow-derived MSC group achieved complete ulcer healing four weeks earlier than the bone marrow-derived mononuclear cell group. In addition, the bone marrow-derived MSC group demonstrated a significant improvement in pain-free walking time, ankle brachial index, transcutaneous oxygen pressure, and magnetic resonance angiography analysis after 24 weeks of follow-up. It is important to point out that there was no significant difference among the groups in terms of pain relief and amputation. Of note, neither cell type resulted in any adverse effects. Lee and colleagues [89] later demonstrated that autologous adipose tissue-derived MSC transplantation in patients with Buerger's disease and diabetic foot (a total of 3 × 10 8 cells) was feasible and safe. It improved claudication walking distance, collateral vessel formation, wound healing, and clinical symptoms, especially pain relief. There was a trend toward an improvement in maximal walking distance. However, there was no change in ankle brachial index [89]. The details of these published human studies are summarized in Table 4 [82,[84][85][86][87][88][89].
From the commercial perspective, Stempeutics Research Pvt. Ltd. (Bangalore, India) has completed a phase 1/2 clinical trial using intramuscular administration of offthe-shelf allogeneic bone marrow-derived MSCs into patients with CLI. The company has reported that the MSCs were well tolerated with no adverse events or rejection. A positive efficacy trend toward improvement in ankle brachial pressure index and a reduction in the number of ulcers were demonstrated. No significant increase in amputation rate was observed. The efficacy of allogeneic bone marrow-derived MSCs is currently being assessed in phase 2/3 clinical trials [90].
In parallel to the trial by Stempeutics Research Pvt. Ltd., two phase 1 trials using intramuscular

Considerations for future clinical trials using mesenchymal stem cell therapy
Although using MSCs to treat CLI is a rather novel therapeutic concept, abundant data are available from the preclinical studies and recent early clinical trials to draw conclusions on the beneficial effect of MSC therapy, beyond the recurring safety profile and feasibility evaluation. However, there is a general lack of consensus on several crucial issues on the recent early clinical trials, rendering the direct comparison among these studies impossible. These issues include the patient type, cell dosing, relevant clinical endpoints, and long-term followup (Table 4).

a. Patient type
Whereas atherosclerosis is the commonest cause of peripheral vascular disease, Buerger's disease or thromboangiitis obliterans is a less common but important cause. The latter is an inflammatory disorder, a distinct form of vascular occlusive disease that afflicts the peripheral arteries of young smokers. It is often characterized by an inexorable downhill course, even in patients who discontinue smoking, once a stage of CLI associated with ulceration or gangrene is reached. Regardless of these two causes of PAD, the early clinical trials that included both group of patients have demonstrated the safety and feasibility of the therapy with suggestion of efficacy. Currently, there is no evidence that either condition may respond better to MSC-based therapy. This needs to be confirmed in larger trials.

b. Cell dosing
As in preclinical studies, the site of administration and the total cell number appeared not to affect efficacy, and the current regimes (either single or multiple doses and either intramuscularly or intra venously administered) used in early clinical trials were safe and seemed to be efficacious (Table 4). However, these parameters need to be standardized to allow direct comparison with other trials.

c. Relevant clinical endpoint
Future clinical trials should include relevant clinical endpoints beyond measurement of ankle brachial index, walking time, and quality of life. Specific objective performance goals (OPGs) have been developed by the Society for Vascular Surgery to define the therapeutic benchmarks for revascularization therapies in CLI [92]. These OPGs for both the safety and efficacy endpoints are: 1) major adverse limb event (MALE), 2) MALE and peri-operative death (POD), 3) major adverse cardiovascular events (MACEs), 4) above-ankle amputation of the index limb, 5) amputation-free survival (AFS), 6) any re-intervention of above-ankle amputation of the index limb (RAO), 7) any re-inter vention, aboveankle amputation of the index limb or stenosis (RAS), and 8) all-cause mortality [92]. These OPGs are crucial, particularly to allow direct comparison with other trials. Table 1 showed that the duration for post-administration follow-up ranged from three months to one year. According to the Society for Vascular Surgery, the assessment of safety endpoints, including the MACE, MALE, and amputation within 30 days was considered the standard duration for post-procedural events for new devices [92]. On the other hand, the minimal exposure time for relevant clinical efficacy is one year: MALE and POD are the primary efficacy endpoints and amputationfree survival is the secondary efficacy endpoint [92]. These OPGs should be adopted in clinical trials involving patients with CLI to allow direct comparison among trials.

Conclusions
MSCs have been shown to be effective in multiple reports in preclinical models of CLI. In addition, there is a substantial amount of evidence on the safety of MSC administration to humans. So far, there has been no evidence of toxicity in terms of either aberrant differentiation or tumorigenesis noted in human studies. Larger studies with longer follow-up will be required to confirm the safety demonstrated by recent studies. The published human data reviewed in this article have enrolled small numbers of patients with relatively short follow-up periods. Since CLI represents the most severe form of PAD, it may also reduce the likelihood of demonstrating efficacy given the severity of the disorder. This is, however, the easiest regulatory pathway to the clinic. Once additional safety data are collected, it may be reasonable to progress to studies to patients with intermittent claudication who represent the majority of patients with PAD and in whom therapeutic efficacy may be easier to demonstrate. This review did not focus on good manufacturing practice (GMP) production of cells or issues surrounding the need to scale up manufacture to generate therapeutic product with predicted efficacy. The challenge for the field remains to undertake clinical trials that progress from phase 1 to 3 while using cells manufactured under GMP conditions. The issue of whether to use autologous or allogeneic 'off the shelf ' cells will also need to be addressed.

Competing interests
TO is a founder and director of and equity holder in Orbsen Therapeutics Ltd.
(Galway, Ireland). AL declares that he has no competing interests.