ReviewStem Cells

Locoregional delivery of stem cell–based therapies

See allHide authors and affiliations

Science Translational Medicine  10 Jun 2020:
Vol. 12, Issue 547, eaba4564
DOI: 10.1126/scitranslmed.aba4564


Interventional regenerative medicine (IRM) uses image-guided, minimally invasive procedures for the targeted delivery of stem cell–based therapies to regenerate, replace, or repair damaged organs. Although many cellular therapies have shown promise in the preclinical setting, clinical results have been suboptimal. Most intravenously delivered cells become trapped in the lungs and reticuloendothelial system, resulting in little therapy reaching target tissues. IRM aims to increase the efficacy of cell-based therapies by locoregional stem cell delivery via endovascular, endoluminal, or direct injection into tissues. This review highlights routes of delivery, disease states, and mechanisms of action involved in the targeted delivery of stem cells.


Stem cells have shown great promise in facilitating the functional restoration of tissues impaired by age, injury, or disease via their ability to either differentiate into a variety of different cell types or modulate the microenvironment to make it conducive to tissue repair and regeneration. Modulation of the microenvironment occurs because of the stimulation of endogenous pathways through direct cell-cell contact and/or the release molecular cargo from stem cells (exosomes, microRNA, growth factors, and transcription factors) (1). Stem cells can be derived from embryonic, fetal, perinatal, and adult sources and can be used in allogenic, autologous, and xenogenic interventions to treat multiple conditions (25).

There are currently 14 U.S. Food and Drug Administration (FDA)–approved cellular products based on terminally differentiated or stem cells (Table 1). Of these products, eight involve allogeneic umbilical cord blood–derived hematopoietic stem cells (UCB-HSCs) used for hematopoietic system disorders, and one uses autologous chondrocytes for knee cartilage defects. In the European Union, darvadstrocel (NCT01541579; TiGenix), which uses allogeneic adipose-derived mesenchymal stem cells (AD-MSCs), recently became the first stem cell product approved for Crohn’s disease, whereas remestemcel-L (NCT02336230; Mesoblast, Ltd.), involving allogeneic bone marrow–derived MSCs (BM-MSCs) for steroid-refractory acute graft-versus-host disease (GVHD), is currently in phase 3 trials and is in the process of filing a biologic license application with the FDA (6). Although there are currently no approved therapies indicated for diseases affecting a broader patient demographic, there are several stem cell–based therapies in phase 2/3 trials that show considerable promise for conditions such as peripheral arterial disease (PAD; NCT03042572, phase 2/3), cardiac disease (CD; NCT03404063, NCT02323620, and NCT01652209, many in phase 3), liver disease (NCT03109236, phase 3), and stroke (NCT03545607, phase 3) (Table 2).

Table 1 U.S. Food and Drug Administration–approved cellular products.

HPC, hematopoietic progenitor cell; UCB-HSCs, umbilical cord blood–derived hematopoietic stem cells.

View this table:
Table 2 Currently active/recruiting stem cell clinical trials exploring locoregional and systemic delivery routes.

ACPs, angiogenic cell precursors; AD, Alzheimer’s disease; ADAS-Cog, Alzheimer’s Disease Assessment Scale–Cognitive Subscale; AD-cSVF, adipose-derived cellular stromal vascular fraction; AD-MSCs, adipose-derived mesenchymal stem cells; ALB, albumin; ALDHbr-SCs, aldehyde dehydrogenase bright stem cells; ALP, serum alkaline phosphatase; ALT, serum alanine aminotransferase; AQOL, Assessment of Quality of Life questionnaire; BM-ECs, bone marrow–derived endothelial cells; BM-MNCs, bone marrow–derived mononuclear cells; BM-MSCs, bone marrow–derived MSCs; BUN, blood urea nitrogen; CDCs, cardiosphere-derived cells; CHF, congestive heart failure; CLI, critical limb ischemia; CLIF-SOFA, chronic liver failure–sequential organ failure assessment; CM, cardiomyopathy; CSCs, cardiac stem cells; DASH, Disabilities of Arm, Shoulder, and Hand questionnaire; DB, direct bilirubin; EF, ejection fraction; EPCs, endothelial progenitor cells; GFR, glomerular filtration rate; GvHD, graft-versus-host disease; HF, heart failure; hiPSC-CMs, human induced pluripotent stem cell–derived cardiomyocytes; HLHS, hypoplastic left heart syndrome; HOMA-IR, homeostatic model assessment of insulin resistance; HOMA-β, homeostatic model assessment of β cell function; IHD, ischemic heart disease; LF, liver failure; LV, left ventricle/ventricular; LVEF, left ventricular EF; LVESV, left ventricular end-systolic volume; MELD, Model for End-stage Liver Disease; MELD/Na, MELD/sodium; MI, myocardial infarction; MMSE, Mini Mental State Examination; MR, magnetic resonance; mRS, modified Rankin scale; NSCs, neural progenitor/stem cells; PA, pre-albumin; PAD, peripheral arterial disease; PAOD, peripheral arterial occlusive disease; PB-HSCs, peripheral blood–derived HSCs; PB-iNSCs, peripheral blood–derived induced neural stem cells; PB-MNCs, peripheral blood–derived MNCs; PCB-MNCs, placental cord blood–derived MNCs; PRA, panel-reactive antibody; PVD, peripheral vascular disease; RVEF, right ventricular EF; SC, stem cell; Scr, serum creatinine; SHED, stem cells from human exfoliated deciduous teeth; S-MSCs, skin-derived MSCs; T2DM, type 2 diabetes mellitus; TB, total bilirubin; TIMI, thrombolysis in myocardial infarction; TMP, TIMI myocardial perfusion; UA, uric acid; UCB-MNCs, umbilical cord blood–derived MNCs; UC-MSCs, umbilical cord–derived MSCs; WJ-MSCs, Wharton’s Jelly–derived MSCs; N/A, not applicable; mo., month(s); y, year(s); wk, week(s).

View this table:

To date, the translation of many stem cell therapies has not reached its full clinical potential. One considerable, but often overlooked, challenge has been the route of administration of stem cells; indeed, studies have shown substantially different therapeutic outcomes when the same stem cell therapy is given by different routes of administration (610). Conventional intravenous injection results in the systemic delivery of cells, and this has resulted in mixed success, considering that the majority of cells become trapped in the lungs and organs of the reticuloendothelial system (spleen and liver) (8, 11). However, a growing body of evidence now suggests that directly delivering cells to a target tissue via endovascular (intra-arterial), intracavity (intraperitoneal and intranasal), or direct tissue (intramuscular and intraparenchymal) injections can increase their therapeutic effects (6, 8, 1214).

Advances in image-guided therapies have enabled interventional physicians, such as interventional radiologists, neuroradiologists, cardiologists, gastroenterologists, and surgeons, to reliably deliver therapies, such as stem cells, directly to target tissues using minimally invasive procedures. Accordingly, a new clinical subsubspecialty is starting to emerge that we propose will be called interventional regenerative medicine (IRM). The impetus underpinning IRM parallels that previously observed in interventional oncology (15). Traditionally, chemotherapy for several solid cancers was given by intravenous injection; however, this often resulted in systemic distribution of the drug associated with severe toxicity/side effects (due to off-target delivery) and poor therapeutic effects (from limited concentration of the active drug reaching target tumors and first-pass metabolism of the drug). To address this, image-guided minimally invasive procedures were developed to deliver these toxic, but potent, drugs directly into tumors via different locoregional approaches. One example is the delivery of chemotherapy directly into liver tumors via their arterial blood supply, transarterial chemoembolization, which has been shown to result in significantly improved clinical outcomes and a reduced need for surgical resections or solid organ transplantation (15). Considering the vast knowledge of routes of access into different organs in the body, technical ability to access these routes (which can include working with the blood supply to different organs or accessing organs directly via percutaneous or minimally invasive techniques), and technology such as microcatheters, needles, and imaging equipment with complementary navigation software developed by interventional physicians over the past few decades, IRM is therefore well poised to build a similar foundation and infrastructure with cellular therapy for organ regeneration. In this review, we will (i) highlight the emerging concepts involved in targeted (locoregional) stem cell delivery, (ii) evaluate stem cell sources and routes of administration, and (iii) outline the barriers to translation of promising preclinical regenerative therapies.

Commonly used stem cell sources

In preclinical and clinical studies, MSCs and mononuclear cells (MNCs) have been widely used to mediate regenerative and reparative effects in diseases arising in different organs. We briefly define two of the most commonly used stem cell sources for clinical cell-based regenerative therapies.

Mesenchymal stem cells. MSCs are a multipotent cell source that can be harvested from adult tissues, such as bone marrow (BM-MSCs), adipose (AD-MSCs), peripheral blood (PB-MSCs), and from perinatal umbilical cord (UC-MSCs) and placental tissues (16, 17). Within the umbilical cord, MSCs can be harvested either from the whole umbilical cord via explant or enzymatic digestion or from specific compartments within the umbilical cord, namely, from umbilical cord blood (UCB-MSCs), such as umbilical arteries or veins (UCA-MSCs and UCV-MSCs), or from the Wharton’s Jelly connective tissue (WJ-MSCs) (17). MSCs can differentiate into tissues of mesodermal lineage (osteocytes, chondrocytes, and adipocytes) but have also been shown to transdifferentiate into cell types of ectodermal lineage (neurons) and endodermal lineage (hepatocytes and pancreatic β cells) (16).

In response to inflammation and tissue injury, MSCs have been suggested to migrate to the injury site either through the circulation (blood and lymphatics) or within the tissue stroma and microcapillaries via growth factors, chemokines, and cytokines such as C-X-C motif chemokine 12 (CXCL12), CXCR4, stromal cell–derived factor 1 (SDF-1), granulocyte-colony stimulating factor, and stem cell factor (18). At the target site, MSCs participate in tissue repair and regeneration by secreting local factors that modulate host adaptive and innate immunological responses; promote angiogenesis; and regulate extracellular matrix and connective tissue deposition (18). Given their ability to exert paracrine actions for a variety of disease states and their low immunogenicity (19), MSCs are the most widely studied stem cell source for cell-based regenerative therapies.

Mononuclear cells. MNCs are isolated from bone marrow (BM-MNCs) or peripheral blood (PB-MNCs) and contain a heterogeneous mixture of adult cells including lymphocytes, monocytes, MSCs, endothelial progenitor cells (EPCs), and HSCs (20). The isolation of MNCs from peripheral blood is less invasive than BM-MNC isolation and does not require the use of anesthesia (20). In contrast to MSCs, which require in vitro expansion before their use because of their low frequency in the tissue of origin, large numbers of BM- and PB-MNCs can easily be fractionated through apheresis and density centrifugation (20). In addition, after isolation, MNCs can be further refined for specific cell types using methods such as magnetic bead–activated cell sorting to select for specific markers (CD133 for EPCs, CD14 for monocytes, and CD34 for HSCs) (20).

Similar to MSCs, MNCs have been shown to exert their therapeutic effects via paracrine signaling in the local host tissues, but studies have also reported their ability to differentiate into other cell types, such as blood, endothelial, hepatic, bone, and neuronal cells, to mediate the regeneration and repair of damaged tissue (21). The exact therapeutic effect of specific cell types within the MNC fractions, along with how the regenerative effects of MNCs differ from MSCs, has not yet been fully determined and may differ according to disease state (2224).


Several stem cell therapies using autologous transplantation of BM-MNCs and PB-MNCs have been explored as treatment strategies to revascularize and hence improve the perfusion of ischemic limbs affected by PAD (2, 2527). BM-MNC injections are proposed to play a paracrine role by augmenting the host neovascularization response, increasing the local concentration of angiogenic factors [vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and angiopoietin-1] and cytokines [tumor necrosis factor–α (TNF-α) and interleukin-1β (IL-1β)] without incorporating or differentiating into newly formed vascular endothelium (23). Clinical studies have reported improvements in multiple PAD outcomes (2831); however, the therapeutic benefit of these cellular therapies on other measures has been questioned (26). For instance, although injections of BM-MNCs or PB-MNCs improved secondary outcomes of PAD, they did not reduce amputation rates in patients with severe limb ischemia (31). In contrast, a meta-analysis of 23 randomized controlled trials (RCTs) using MNCs reported decreased amputation rates and an increased probability of ulcer healing (27). Whether greater clinical improvements were observed in therapies using PB-MNCs or BM-MNCs has also not yet been determined. Huang et al. (32) observed no significant difference in amputation rates between PB-MNC versus BM-MNC administration, whereas others have reported improved amputation rates after PB-MNC administration (26, 27). Because MNCs are composed of a mixture of different cell types, a more effective approach may involve isolating proangiogenic CD34+ cells within the mononuclear fraction, which has shown promising results in some clinical studies (33, 34).

Cellular therapies for PAD have typically been administered intramuscularly into the gastrocnemius, or intra-arterially via the femoral artery, without clear consensus on which route is superior (Fig. 1) (2). The transient deposition of cells into the ischemic tissue from intramuscular injections has been suggested to facilitate paracrine signaling and cell incorporation into the host neovasculature (25). In contrast, intra-arterial delivery may allow transplanted stem cells to access oxygen- and nutrient-rich peri-ischemic zones that better facilitate their cellular activity (2). Clinical studies incorporating intramuscular, intra-arterial, or both routes of administration have demonstrated improvements in PAD (2831); some studies show no difference between intramuscular or intra-arterial delivery (35), whereas others show that intramuscular delivery was more effective than intra-arterial (26, 27). Because these approaches have not been directly compared in larger RCTs, additional trials accompanied by imaging studies aimed to assess the biodistribution, integration, and prolonged survival of transplanted cells will be required to validate these findings (27). Currently, 9 of the 11 active and recruiting PAD clinical trials have specified a local route of delivery (Table 2).

Fig. 1 The locoregional delivery of stem cell–based therapies in interventional regenerative medicine.

Locoregional routes of administration of stem cells to treat liver disease have commonly involved portal vein (intraportal) or intra-arterial injections into the hepatic artery. For stroke and Alzheimer’s disease, stem cells have been delivered through intra-arterial (carotid artery), intracerebral, intranasal, and intrathecal routes of administration. In kidney injury, stem cells have been delivered intra-arterially into the suprarenal aorta or renal artery. In peripheral arterial disease, stem cells have been administered intramuscularly, into the gastrocnemius, and intra-arterially, into the femoral artery. In cardiac disease, stem cells have been delivered using intra-arterial (coronary artery) and intramyocardial injections. In type 2 diabetes mellitus, stem cells have been administered via intra-arterial injections into the gastroduodenal, superior pancreaticoduodenal, splenic, celiac, superior mesenteric, or dorsal pancreatic arteries.



In addition to MNCs used in PAD, cellular therapies in CD have also adopted the use of cardiac stem cells (CSCs), MSCs, and human induced pluripotent stem cell (hiPSC)–derived cardiomyocytes (3). When administered into ischemic myocardium, MNCs and MSCs have been reported to secrete paracrine factors [VEGF, IL-1β, platelet-derived growth factor (PDGF), and insulin-like growth factor 1] that, in turn, control reactive hypertrophy, attenuate T cell recruitment, reduce collagen deposition, and inhibit cardiomyocyte apoptosis in the host tissue (36, 37). Similar to MNCs, MSCs can also secrete VEGF, bFGF, and hepatocyte growth factor (HGF), growth factors that play a role in neovascularization of infarcted myocardium (38). The therapeutic effect of BM-MNCs is largely inconclusive, with some studies demonstrating improvement in the primary outcome left ventricular ejection fraction (LVEF) (39, 40) whereas others demonstrating no significant change (41, 42). However, MSCs have been suggested to be more effective than BM-MNCs (43). In the context of acute myocardial infarction and ischemic heart failure, MSC administration led to significant improvements in left ventricular function and stroke volume (44).

CSCs, which are multipotent progenitors harvested from myocardial biopsies, have also shown encouraging results for CD (45). Isolated CSCs can be cultured in vitro to yield self-assembling cardiospheres, which can then be dissociated into multipotent cardiosphere-derived cells (CDCs) (46). The administration of CSCs/CDCs is distinct in that it has been suggested to increase the proliferation and expression of cardiac-specific proteins in endogenous CSCs through paracrine signaling (47). When administered through the coronary arteries, CSCs were shown to extravasate through the disrupted vasculature and home across cardiac vessel walls into the infarcted myocardium via an SDF-1/CXCR4–mediated mechanism (48), thereby transdifferentiating into endothelial cells, smooth muscle cells, and myocytes to form new vasculature and regenerate the infarcted region (45, 48). Studies demonstrated that CSC transplantation in patients with CD resulted in an improvement in LVEF (49), and autologous CDC administration increased viable heart mass and regional contractility (5052). Because of the low number of cells that may be harvested from cardiac biopsies, CSC/MSC combination therapies are also being evaluated as an alternative approach that may be more efficacious than using either cell type alone (47).

Because hiPSCs were demonstrated to differentiate into cardiomyocytes in vitro (53), the use of hiPSC-derived cardiomyocytes has recently emerged as a method for generating large amounts of patient-specific cardiomyocytes for cellular transplantation. In preclinical models, iPSC-derived cardiomyocytes were shown to promote angiogenesis and reduce infarct size, ventricular wall stress, and cardiomyocyte apoptosis through yet unknown molecular mechanisms (54). However, the lack of standardized protocols to generate fully mature hiPSC-derived cardiomyocytes, combined with the risk of tumorigenesis and cardiac arrhythmias, highlights crucial concerns that must be addressed before clinical use (55, 56).

Intracoronary and intramyocardial (epicardial and transendocardial) injections are the two most widely used methods of delivery of cellular therapies in CD (Fig. 1) (56, 57). Intracoronary injections deliver cells into one of the major coronary arteries (left anterior descending, left circumflex, or right coronary arteries). This route of administration is less invasive than intramyocardial injection, which typically involves surgical intervention or endocardial access (56). Compared to intracoronary injections, intramyocardial injections of stem cells generally resulted in three- to fivefold greater myocardial retention after delivery (50, 58); however, other studies have reported intracoronary injections to be equally effective as intramyocardial injections (59). A meta-analysis of MSC therapies in preclinical and clinical studies concluded that transendocardial delivery exhibited the greatest reduction in infarct size and improvements in LVEF, whereas intracoronary delivery demonstrated no improvement (14). Because this demonstrates that intramuscular injections may offer greater therapeutic benefit to patients, studies have also aimed to optimize the delivery approach to reduce the invasiveness of the procedure and increase cell retention after delivery (57, 60). Currently, 45 of the 51 active and recruiting clinical trials have specified a targeted approach to stem cell delivery for the treatment of CD (Table 2).


Among the types of stem cells, MSCs have shown the most promise for liver diseases in clinical trials (61). Upon MSC transplantation, the secretion of MSC paracrine factors (IL-10, TNF-α, HGF, and transforming growth factor–β3) induced by host fibrotic hepatic stellate cells (HSLCs) locally inhibits HSLC proliferation and attenuates collagen synthesis, whereas direct cell-cell contact between MSCs and HSLCs is known to promote HSLC apoptosis through Notch1 signaling (6266). In preclinical models of acute liver injury, MSCs also play an immunomodulatory role in adaptive and innate immunity by suppressing monocytes/macrophages, hepatic natural killer T cells, and dendritic cells, while depleting infiltrating CD4+ T cells and suppressing their inflammatory signaling [interferon-γ (IFN-γ), CCR5, and CXCL10)] (65, 67, 68). Clinical studies using MSCs in acute-on-chronic liver failure report improvements in measures of liver function such as the end-stage liver disease [Model for End-stage Liver Disease (MELD)] score and alanine aminotransferase concentration (8, 69), without significant adverse events or complications (70, 71).

Although intravenous administration was the most common route of administration used by several clinical studies (8), this approach was associated with a low cellular presence of MSCs in the injured liver tissue, with a significant proportion of cells reported to accumulate in the spleen and lungs (72). Intraportal delivery and intrahepatic arterial delivery are reportedly superior to intravenous injection because they allow a greater number of injected cells to seed into the injured liver (Fig. 1) (8, 12, 73). In a meta-analysis of 23 controlled trials, Zhao et al. (8) concluded that intrahepatic injection exhibited earlier clinical improvements in the MELD score and concentrations of albumin and total bilirubin when compared to intravenous injection. Similarly, in a porcine model of acute liver failure, intraportal injection was superior to other routes, although other studies have questioned the effect of portal metabolites on the function of MSCs given by this route of administration (7). Hence, further clinical studies will be required to systematically compare the optimal local delivery approach of MSC therapies on the basis of cell retention at the target site and therapeutic efficacy. Currently, 4 of the 18 active and recruiting clinical trials have specified a targeted approach of stem cell delivery for the treatment of liver disease (Table 2).


Studies on stem cell therapy to treat type 2 diabetes mellitus (T2DM) have almost entirely revolved around the use of MNCs and MSCs. Rather than differentiating MSCs into β cells, of which the yield has been reported to be exceedingly low (74), MSC infusions have instead been reported to promote host insulin sensitivity by up-regulating glucose transporter type 4 expression in target tissues, increasing tyrosine phosphorylation of insulin receptor substrate 1, increasing phosphorylation of protein kinase B (p-Akt), and inhibiting the secretion of the proinflammatory cytokine TNF-α (75, 76). In addition, through secreted exosomes, MSCs have been shown to protect endogenous pancreatic islet survival, promote insulin secretion, repair islet architecture, and stimulate endogenous β cell regeneration through the phosphatidylinositol 3-kinase (PI3K)/Akt pathway (7577). Studies using MNCs or MSCs as a stem cell source generally showed modest improvements in primary clinical outcomes: a decrease in glycosylated hemoglobin (HbA1c) percentages, improvement in C-peptide concentrations, and a reduction in daily insulin requirement (22, 7883). In a systematic review comparing the therapeutic effect of MNCs and MSCs, the efficacy of targeted BM-MNC administration was comparable to WJ-MSCs and UC-MSCs (84). Because MNCs and MSCs may exert their therapeutic effects through different cell-dependent mechanisms (83, 85), it has also been suggested that a combination of multiple sources may result in greater therapeutic effect for patients with T2DM (82).

In T2DM, intra-arterial injections have been performed using a minimally invasive endovascular approach to deliver cells into the gastroduodenal, superior pancreaticoduodenal, splenic, celiac, or superior mesenteric arteries or directly into the pancreas via the dorsal pancreatic artery (Fig. 1) (9, 22, 7882). Because the mobilization of stem cells in patients with diabetes appears to be impaired (86), defining the optimal route for targeted delivery of stem cells will be fundamental to maximizing their clinical benefit. To date, studies aimed at evaluating administration routes for T2DM have been scarce. Sood et al. (9) studied the migration and retention of BM-MNCs labeled with the positron emission tomography (PET) tracer fluorine 18-fluorodeoxyglucose to demonstrate that a targeted intra-arterial approach, via the superior pancreaticoduodenal or splenic arteries, resulted in greater homing of MSCs to the pancreas compared to conventional intravenous administration. The study also noted that reductions in HbA1c and insulin requirement were less pronounced after intravenous administration compared to intra-arterial delivery (9). Subsequently, in a randomized case-controlled trial, intra-arterial delivery offered no significant reduction in insulin dose requirement compared to intravenous delivery (87). However, both studies were limited by small sample sizes and reported no significant differences when intra- and intergroup comparisons were analyzed. Currently, one of the three active and recruiting T2DM clinical trials have specified a targeted approach for stem cell delivery (Table 2).


Stem cell therapies for acute kidney injury (AKI), characterized by a decrease in renal function resulting in an increase in oxidative stress, reduced glomerular filtration, and renal fibrosis (88, 89), have almost exclusively been conducted in animal models using MSCs as a cell source. Preclinical studies using MSCs have been reported to reduce reactive oxygen species through nuclear factor E2-related factor 2/antioxidant response element signaling and up-regulation of antioxidant enzymes (90, 91), decreased expression of proinflammatory cytokines and proteins (92), and reduced evidence of renal apoptosis (92). In addition, MSCs used to treat AKI have been shown to protect against renal fibrosis by down-regulating profibrotic signal transducer and activator of transcription 3 and matrix metalloproteinase–9, inhibiting renal tubular cell epithelial-mesenchymal transition into matrix-secreting fibroblasts, and preventing the release of downstream profibrotic factors (93, 94). Studies using MSC infusions in AKI models have thus reported improvements in kidney function, assessed by decrements in serum urea and creatinine (11, 91), albeit to varying degrees depending on the tissue source of MSCs (4, 95).

Animal studies evaluating the optimal administration route have suggested that intra-arterial delivery through the renal artery showed the greatest therapeutic effect, histological recovery, and cell localization to the kidneys when compared to parenchymal and systemic venous injections (10, 11, 96, 97), although one study reported no significant differences in delivery route (98). Of the three clinical trials of MSC therapies performed in patients with AKI since 2008, only one study (NCT00733876, phase 1) was completed, whereas the other two trials (NCT01275612, phase 1; NCT01602328, phase 2) were withdrawn and terminated, respectively (99). In the completed study (NCT00733876), BM-MSCs delivered into 15 patients intra-arterially through the suprarenal aorta to avoid pulmonary entrapment prevented postoperative and late deterioration of renal function while reducing the length of stay and hospital admission rates by 40% (99). The preliminary data appeared to match historical case controls at the same institution and data obtained from preclinical studies and likely will be further investigated in a subsequent phase 2 trial (99). In contrast, in the terminated multicenter ACT-AKI trial (NCT01602328) of 156 patients with postcardiac surgery AKI, Swaminathan et al. (100) reported that intra-aortic delivery of MSCs resulted in no significant difference in functional measures of renal function (time to renal functional recovery, 30-day all-cause mortality, need for dialysis) and thus terminated the trial because of futility (101). Currently, there are no active or recruiting clinical trials that have specified a targeted approach to stem cell delivery for the treatment of AKI.


Earlier studies on cellular therapies for stroke have most commonly involved the use of self-renewing multipotent neural progenitor cells or neural stem cells (NSCs) (102), which can be isolated from primary tissue sources or derived from embryonic stem cells or iPSCs. NSCs can support regeneration through neurogenesis via direct integration or stimulating endogenous proliferation and by promoting angiogenesis, axonal growth, and microglia activation/proliferation (103). NSCs also promote bystander immunomodulation by releasing neurotrophic factors such as glial cell line–derived neurotrophic factor and nerve growth factor (104); down-regulating adhesion molecules [intercellular adhesion molecule–1 (ICAM-1) and vascular cell adhesion molecule–1] (105, 106); and reducing expression of proinflammatory factors [TNF-α, IL-4, IL-6, nuclear factor κB (NF-κB), monocyte chemotactic protein 1, and macrophage inflammatory protein–1α] in the presence of damage-associated molecular patterns within cerebral tissue (105107). Furthermore, these cells can inhibit the activation and recruitment of neutrophils, microglia, macrophages, and T cells (105107). Alternatively, non-neuronal stem cell sources such as MNCs or MSCs have been used in most recent cellular therapies for stroke (108, 109). In contrast to NSCs, only a small percentage of transplanted MNCs and MSCs integrate into the target tissue as differentiated neuronal cells (108, 110); instead, most transplanted cells exert anti-inflammatory, immunomodulatory, neuroprotective, and angiogenic effects through paracrine-mediated mechanisms and secreted factors (111114). After stroke, up-regulation of trophic factors within the ischemic brain, presumably through MSC-secreted factors (bFGF, brain-derived neurotrophic factor, PDGF-AA, angiopoietin-2, and CXCL16) (115), provides neuroprotection and induces endogenous NSC proliferation and differentiation (114116). MSCs were also shown to provide immunomodulatory effects by reducing microglia, macrophage, and monocyte infiltration (116, 117); alleviating microglia neurotoxicity via CX3CL1 and IL-5 (117); and attenuating inflammation by decreasing the local expression of IFN-γ and TNF-α while increasing IL-4 (112).

MNCs promote neuroprotection by modulating the expression profiles of oligodendrocytes and neurons through the PI3K/Akt pathway, which is suggested to inhibit apoptosis and enhance antioxidant activity (24, 118). Surprisingly, MNCs were reported to induce white matter axonal remodeling and synaptic plasticity through the increased expression of neurofilament and synaptophysin (119), effects not seen with NSCs or MSCs. MNCs also mediated immunomodulatory effects in middle cerebral artery occlusion–mediated ischemic brain injury by reducing the number of CD45+/CD11b+ cells (microglia/macrophages), CD45+/B220+ B cells, and CD4+ helper T cells (111, 120). The reduction in immune cells was correlated with decreased expression of proinflammatory cytokines (TNF-α and IL-1β) and reduced DNA binding of NF-κB (which targets TNF-α and IL-1β) in the ischemic cerebral tissue (111, 120), suggesting that MNCs may also act by inhibiting immune cell migration into the brain after a stroke. Clinical trials using NSCs, BM-MSCs, UC-MSCs, and peripheral blood stem cells have reported promising improvements in clinical measures of stroke: neurologic deficit score (National Institutes of Health Stroke Scale), motor function (Fugl-Meyer Assessment), functional independence (Functional Independence Measure), and daily life ability (Barthel index) (13, 121).

In animal studies, administration routes of cellular therapies for stroke have commonly used both systemic (intravenous) and nonsystemic routes (intra-arterial, intranasal, intrathecal, and intracerebral). Intravenous delivery and intra-arterial delivery are considered less invasive approaches than intracerebral delivery, which requires a craniotomy; however, studies comparing both routes of administration have reported mixed results regarding the biodistribution of grafted cells in the brain. Yang et al. (122) and Vasconcelos-dos-Santos et al. (109) achieved similar brain homing and functional efficacy using intra-arterial versus intravenous routes of administration, whereas Kamiya et al. (123) reported greater brain retention and functional outcomes under intra-arterial administration. Greater clinical benefit was observed in 16 clinical studies using locoregional routes (intracerebral, intracarotid, and injection into the subarachnoid space) versus intravenous administration (Fig. 1) (13). Several preclinical studies have also reported promising findings with intranasal delivery, which is a noninvasive administration route that bypasses the blood-brain barrier (BBB) via cellular migration from the olfactory bulb (124, 125). Brain-targeted intranasal delivery of MSCs in rats was shown to reduce cell death in the infarct regions while promoting local cerebral blood flow, angiogenesis, and neurogenesis in the cortex (126). Because intranasal administration avoids some of the issues associated with intravascular delivery such as cell entrapment to peripheral organs and arterial occlusions, further studies are warranted to assess its clinical efficacy. Currently, 9 of the 13 active and recruiting clinical trials have specified a targeted approach to stem cell delivery for the treatment of stroke (Table 2).


Alzheimer’s disease (AD), a neurodegenerative disorder characterized by aggregates of β-amyloid (Aβ) peptide–containing senile plaques and hyperphosphorylated tau-containing neurofibrillary tangles, is highly dependent on the activation of microglia into the M1 subtype (associated with AD disease progression) or the M2 subtype (associated with neuroprotection through the secretion of neurotrophic and anti-inflammatory factors) (127). In the context of AD, MSCs have been shown in preclinical studies to not only promote M2 activation of microglia through MSC-secreted CCL5 (128130) but also enhance neuroprotection (131), reduce oxidative stress (128, 130), increase Aβ peptide clearance (131), reduce inflammation, and promote neurogenesis (130, 131), which improves spatial learning and alleviates AD-associated impairments in animal models (128, 129, 132). In addition, MSCs have played a key role in up-regulating Aβ-degrading enzymes (neprilysin and insulin-degrading enzyme) through secreted factors such as soluble ICAM-1 and by suppressing M1 microglial proinflammatory activity, which further promotes Aβ plaque breakdown (133). Despite promising preclinical data, clinical trials using MSCs for AD have been inconsistent in their approaches or designs, which have limited realization of the full clinical potential of MSCs for the treatment of AD. One clinical trial using MSCs for AD has been completed to date (NCT01297218 and NCT01696591), which did not show a significant improvement in cognitive decline or neuroprotective effects with UCB-MSC injections into the hippocampus and precuneus (5). Authors attribute the lack of significant findings to the low sensitivity of neuroimaging (in contrast to biochemical analyses used in mouse studies) and to differences in AD phenotypes between preclinical models and patients that may result in a differential response to UCB-MSC treatment (5).

In patients with AD, MSCs need to reach the cortex to exert their neuroprotective effects and prevent the destruction of neurons, as well as reach the subventricular zone and the subgranular zone in the dentate gyrus to stimulate regeneration of damaged cortical neurons (134, 135). Studies examining the effect of MSCs have used routes of administration that are ineffective, unfeasible, or impractical for repeated injections. Intraparenchymal injections can directly deliver MSCs to targeted structures in the brain, but this requires brain surgery under general anesthesia and cannot be easily repeated on a regular basis. In contrast, direct intra-arterial injection of MSCs into the brain mitigates the requirement for invasive surgeries, as necessary for intraparenchymal and intracerebroventricular injections, and allows for repeated access (Fig. 1). After MSCs are delivered into cerebral circulation via intra-arterial injection, they need to cross the BBB and leave the vasculature to enter the brain parenchyma. Under healthy, homeostatic circumstances, MSCs do not cross the BBB or enter the brain. However, in AD, the BBB is altered in structure and function, which is thought to facilitate the passage of MSCs into the brain (136). There are currently no published studies that have systematically compared MSC delivery routes in AD. Four of the nine active and recruiting clinical trials have specified a targeted approach to stem cell delivery for the treatment of AD (Table 2).


Over the next decade, IRM will play a central role in the clinical care of patients by enabling targeted delivery of stem cell therapies directly into damaged/injured organs using minimally invasive approaches. The growing trend toward using targeted routes of administration reflects a paradigm shift away from conventional systemic delivery to enable a higher concentration of cells to reach target organs while avoiding the loss of administered cells to off-target peripheral organs, such as the lungs, spleen, and liver. Beyond the disease states covered in this review, targeted administration of stem cell therapy is also currently being explored for many other conditions including epidermolysis bullosa (intradermal) (137), systemic sclerosis (intramuscular) (138), multiple system atrophy (intra-arterial; NCT03265444) (139), progressive supranuclear palsy (intra-arterial; NCT01824121), type 1 diabetes mellitus (intra-arterial; NCT01374854), Parkinson’s disease (intra-arterial and intracerebral; NCT03724136) (140), traumatic brain injury (intra-arterial) (141), and amyotrophic lateral sclerosis (intrathecal, intraspinal, and intramuscular) (142), among others.

To ensure that IRM reaches its full potential, research and development will need to focus on producing specialized delivery systems for cellular therapies and building the required clinical infrastructure to enable IRM to have a strong foundation to expand upon in the years to come (143, 144). Indeed, delivery strategies for stem cell therapies must consider technologies that reduce sheer stress applied to cells during infusions, guarantee accurate cell placement at desired sites, and increase cellular retention at these sites to optimize cell viability and therapeutic efficacy (145). For instance, the delivery of cellular therapies in CD has been facilitated by the use of a stainless steel or nitrol surface, which allows for multiple repeated transendocardial injections (MyoCath catheter, Bioheart Inc.); the adoption of a balloon-mounted microneedle improved cell extravasation into the perivascular space in transcoronary injections (Cricket microinfusion catheter, Mercator MedSystems Inc.); and the off-label use of existing delivery systems, such as the over-the-wire balloon, has been implemented to control the infusion pressure and rate of intracoronary injections (144). Regarding clinical infrastructure, good manufacturing practice (GMP) facilities with expertly trained staff, clinicians, and biologists will be required to store, purify, characterize, and expand therapeutic cells, as well as to counsel patients on the benefits, costs, and risks of cellular therapies compared to standard treatments (143). In California, GMP facilities including academic centers, clinical sites, and biotechnology companies that would allow for standardized cell expansion, delivery, and long-term follow-up on clinical outcomes have been established with support from the California Institute for Regenerative Medicine (143). Thus, driving IRM forward will require the tight integration of stem cell biology, translational biology, clinical expertise, and hospital infrastructure.

Interventional physicians are well positioned to act as a “hub” between other specialties and may facilitate transdisciplinary collaboration and optimize effective pathways for patients to receive stem cell therapies for different indications. In the coming decade, a subset of interventional physicians who can perform cellular transplantation to regenerate damaged and failing organs may emerge to complement their transplant surgical colleagues who perform solid organ transplantation. In these cases, reducing the procedural duration of cellular transplantation, improving tolerability, reducing cost, and minimizing invasiveness as compared to whole-organ transplantation will hopefully expand the number of patients who can benefit from cellular therapy as either a potential cure or an interim measure before organ transplant.

However, fully establishing IRM as a pillar of health care is not without barriers. Preclinical studies have aroused considerable interest in stem cell therapies for different diseases, but improving the clinical translation of these therapies will require increased stringency to address the disconnect between preclinical models and human diseases. Some clinical studies have reported only modest clinical benefit from targeted stem cell therapies; the discrepancy between preclinical and clinical results is likely due to lack of standardization in identifying specific patient cohorts according to defined comorbidities, disease severity, co-medications, dosing regimen, and other factors that may have differential therapeutic responses to certain stem cell therapies (6, 146). For instance, compared with clinical trials that use doses ranging from 1 million to 10 million cells/kg, MSCs in preclinical studies have been administered at far higher doses (50 million cells/kg), indicating that the optimal dosing regimen does not directly translate from animal models to patients (6). Preclinical models have also yet to fully account for the complicated etiology and clinical presentation of disease states in patients, and thus, the implementation of stringent inclusion criteria based on prognostic biomarkers may aid in identifying homogeneous patient cohorts in subsequent clinical trials (146, 147). Greater assessment of cohort phenotypic characteristics, and how these features differ from or alter the therapeutic response observed in established preclinical models, would allow for further resolution and discrimination between outcomes of therapies as they are translated from the benchtop to the bedside.

Unlike pharmacological therapies that are tightly controlled on the basis of medication structure and dosage, cellular therapies require standardization along multiple checkpoints (route of administration, stem cell source, transport time, and culture conditions, the latter including the temperature, cell passage number, cell dose, and dosing regimen) to achieve sustainable approaches that can be widely implemented with confidence (85, 146). One example of a discrepancy in cell therapy is the intracoronary BM-MSC treatment of acute myocardial infarction: One trial compared dosages of either 12.5 × 106 or 25 × 106 cells (single injection), whereas another opted to use 1.0 × 106 per kilogram (two injections) (NCT01781390 and NCT01652209; Table 2). In addition, only a minority of active and recruiting stem cell clinical trials have specified both the cell dose and dosing regimen, and few have systemically characterized the culture conditions of their stem cell source (6). Aside from the mechanistic and therapeutic evaluation of cellular therapies, future studies are strongly urged to also consider the scalability of the therapeutic cells, prioritizing the controlled proliferation of cells while navigating complex regulatory, economic, and production hurdles endemic to therapy development.

Improvements to stem cell therapies will also mandate the concurrent development of new imaging modalities (multimodal, photoacoustic, and magnetic particle imaging techniques) or contrast agents to label, facilitate, monitor, and track the engraftment of administered stem cells intra- and postoperatively (148150). Similar to traditional optical, ultrasound, magnetic resonance (MR), PET, computed tomography, and nuclear imaging techniques, new imaging modalities must balance high temporal/spatial resolution with high sensitivity/contrast (148). The use of new contrast agents will require further validation of long-term toxicity, biodegradation, photostability, and adverse immune reactions. In addition, these agents must not affect the proliferation, differentiation potential, pluripotency, migration, or viability of the therapeutic stem cells.

The current landscape of stem cell therapies for the repair and regeneration of tissues and organs of various disease states has been met with increasing clinical attention. The recent shift toward locoregional techniques of the administration of stem cell therapies in IRM is becoming an increasingly attractive approach. Thus, greater recognition for better standardization of therapeutic procedures and patient cohorts will be necessary, given the complexity of different cellular therapies and limited understanding of precise therapeutic mechanisms of action. The principle of locoregional delivery may be applied more broadly beyond stem cell therapy within regenerative medicine applications, including cell-based and cell-free interventions, such as the targeted administration of gene therapies, antibodies, adhesion molecules, extracellular vesicles, nucleic acids, liposomes, and nanoparticles. Together, with a multidisciplinary and concentrated effort between interventional physicians, biologists, engineers, and other health care professionals, IRM holds considerable promise to be a major asset for regenerative medicine.


Acknowledgments: We thank A. N. Thomas for the draft figure illustration. Funding: This work was supported by grants from the NIDDK (R01DK119293 and P30DK116074), the Akiko Yamzaki and Jerry Yang Faculty Scholar Fund in Pediatric Translational Medicine, and the Stanford Maternal and Child Health Research Institute and the SIR Foundation Ring Development Grant. Competing interests: The authors declare that they have no competing interests.

Stay Connected to Science Translational Medicine

Navigate This Article