|Home | About | Journals | Submit | Contact Us | Français|
Cardiosphere-derived cells (CDCs) isolated from human endomyocardial biopsies reduce infarct size and improve cardiac function in mice. Safety and efficacy testing in large animals is necessary for clinical translation.
Mesenchymal stem cells, which resemble CDCs in size and thrombogenecity, have been associated with infarction following intracoronary infusion. To maximize CDC engraftment while avoiding infarction, we optimized the infusion protocol in 19 healthy pigs. A modified cocktail of CDCs in calcium-free PBS, 100U/mL heparin and 250μg/mL nitroglycerin eliminated infusion-related infarction. Subsequent infusion experiments in 17 pigs with post-infarct left ventricular dysfunction showed CDC doses ≥107 but <2.5×107 result in new myocardial tissue formation without infarction. In a pivotal randomized study, 7 infarcted pigs received 300,000 CDCs/kg (~107 total) and 7 received placebo (vehicle alone). Cardiac MRI 8 weeks later showed CDC treatment decreased relative infarct size (19.2% to 14.2% of left ventricle infarcted, p=0.01), while placebo did not (17.7% to 15.3%, p=0.22). End-diastolic volume increased in placebo, but not in CDC-treated animals. Hemodynamically, dP/dt max and dP/dt min were significantly better with CDC infusion. There was no difference between groups in the ability to induce ventricular tachycardia, nor was there any tumor or ectopic tissue formation.
Intracoronary delivery of CDCs in a pre-clinical model of post-infarct LV dysfunction results in formation of new cardiac tissue, reduces relative infarct size, attenuates adverse remodeling, and improves hemodynamics. The evidence of efficacy without obvious safety concerns at 8 weeks’ follow-up motivates human studies in post-MI patients and in chronic ischemic cardiomyopathy.
Despite remarkable advances in the treatment of coronary disease, nearly 1/3 of patients who survive a myocardial infarction develop heart failure within 5 years.1 This burden of disease has driven investigation of progenitor cell therapy for the heart, including clinical trials of bone marrow-derived cells.2-7 More recently, the discovery of resident cardiac stem cells, which possess a pre-determined cardiac fate and can generate new heart tissue, has spurned interest in their clinical application.8-12 While present, cardiac stem cells exist in small numbers in vivo and their isolation and expansion ex vivo presents a challenge to their clinical use. Prior work in our lab developed a practical method to isolate and expand resident cardiac stem cells from endomyocardial biopsy specimens.13 These “cardiosphere-derived cells” (CDCs) express surface markers typical of stem cells (c-kit, CD105), and delivered to the heart after coronary ligation can generate new myocardium and improve cardiac function in mice. Successful clinical application could provide a novel therapy for post-infarct LV dysfunction and cardiomyopathy in general.
Given the results in mice, we hypothesized that CDC isolation and delivery following infarction in a large animal model could be performed safely and would result in formation of new cardiac tissue, which in turn would limit negative remodeling and improve cardiac function. We sought to test this hypothesis in a relevant clinical model, making every effort to determine whether the safety concerns that have plagued certain modalities of stem cell therapy would apply to CDC therapy.14,15
Studies were performed according to the “Position of the American Heart Association on Research Animal Use” with protocols approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine.
Under anesthesia, 3 types of progenitor cells were isolated from pigs. Mesenchymal stem cells (MSCs) and bone marrow mononuclear cells (BMMCs) were isolated and collected as reported previously.16,17 CDCs were isolated and cultured from endomyocardial biopsies. Briefly, 5F endomyocardial biopsy forceps (Argon Medical, Athens, TX) were advanced to the right ventricle through a jugular venous sheath under fluoroscopic guidance. Biopsies from the septum were stored in ice-cold cardioplegia and processed within 3 hours to yield CDCs as previously described13 (also see Supplemental Methods). CDCs used for intracoronary infusion were frozen in Pentaspan (B. Braun Medical, Bethlehem, PA) containing 5% DMSO with 2% albumin and stored in liquid nitrogen until the day of infusion. Freezing did not affect CDC viability (data not shown).
CDCs used for engraftment studies were transduced with lentiviral luciferase or nuclear-localized LacZ. Cells at 50% confluence were exposed to lentivirus at a multiplicity of infection of 20 in media containing Polybrene. Following exposure, fresh media was added to the culture flask and when confluent, CDCs were passaged per routine. Luciferase activity was measured (Promega, Madison, WI), to establish a standard luciferase activity curve for each animal. LacZ expression was confirmed by X-gal stain (Applied Biosystems, Norwalk, CT).
Progenitor cells [105 or 106 cells in 1ml of Ca2+-free PBS (PBS-)] were mixed with 20ml of fresh blood to measure thrombus formation. The mixture was placed in a 50 mL conical tube and kept at room temperature, with gentle inversion every 5 minutes. One and 8 hours after mixing, thrombus was collected for analysis by centrifugation. Tissue factor ELISA of separate progenitor cell lysate samples was performed per manufacturer instructions (R&D Systems, Minneapolis, MN).
Following induction of anesthesia, intracoronary infusion of CDCs was performed in 19 healthy farm pigs and 33 pigs with infarcted hearts (9 farm and 24 minature). Infarcted animals had anterior MI created as previously described,18 followed immediately by endomyocardial biopsy. Four weeks later intracoronary infusion was performed via a 3.0 × 8 mm over-the-wire coronary balloon (Boston Scientific, Natick, MA) placed in the mid-LAD. CDCs or placebo (carrier solution alone) were infused in 3 cycles of intermittent balloon inflation. Heart rate, blood pressure, end-tidal carbon dioxide, pulse oximetry, and 12 lead ECG were monitored throughout (see Supplemental Methods).
Animals receiving luciferase+ CDCs were sacrificed 24 hours after infusion (see Supplemental Figure 1 for schematic of this experiment). One gram tissue samples were taken from sites in the LV and RV and from the lungs, liver, spleen and kidneys then flash-frozen and stored at -80°C. Within 24 hours tissue was homogenized with 10% fetal bovine serum, then centrifuged. Luciferase activity of the supernatant was measured and converted to CDCs/g tissue using the standard curve created for each cell line.
Animals receiving lacZ+ CDCs were sacrificed 8 weeks after infusion (see Supplemental Figure 2 for schematic of this experiment). Tissue from the peri-infarct zone and remote areas of the heart was fixed in 4% paraformaldehyde. Frozen sections were used for X-gal and immunohistochemical staining, the latter using antibodies for beta-galactosidase conjugated to FITC (Abcam, Cambridge, MA) and α-sarcomeric actin (Abcam) with Alexa 568 secondary antibody (Invitrogen, Carlsbad, CA).
To determine whether engrafted CDCs influence post-infarct remodeling, 16 minature pigs were randomized to CDC (300,000 CDCs/kg) or placebo infusion. Four weeks after infarction 3T cardiac MRI (Siemens, Munich, Germany) was performed to determine LV systolic function and chamber size using 8 mm contiguous short axis slices; infarct size was measured using 6 mm contiguous slices with delayed gadolinium enhancement.19 Analysis was performed using Cine Tool 3.4 software (GE Healthcare, Milwaukee, WI). Immediately before infusion, left ventricular hemodynamic measurements were performed (Millar Instruments, Houston, TX). Blood samples were taken at infusion and 24 hours later to measure troponin I (TnI). Eight weeks later animals had repeat MRI, hemodynamic measurement, coronary angiography, and provocative electrophysiologic testing (EPS) followed by comprehensive autopsy including examination of the heart, lungs, liver, spleen, kidneys and brain. During EPS up to 3 extra stimuli were performed at the RV apex and outflow tract to attempt to induce ventricular tachycardia (VT; see Supplemental Figure 3 for schematic of this expermient).
All data are presented as mean +/- standard error of the mean (SEM). Statistical analyses were performed with the use of statistical software SAS (version 9.2, Cary, NC), where p<0.05 was deemed significant. Due to the limited sample size, the non-parametric Wilcoxon signed rank test was used to perform paired and unpaired comparisons. For MRI and hemodynamic analyses, where measurements were repeatedly taken from the same pig over time, a stratified linear mixed model was used.20 The authors had full access to the data and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
CDC isolation from endomyocardial biopsy specimens was highly successful: 33/35 animals (94%) were successfully biopsied with a mean total biopsy weight of 52.6±6.7mg. From these, CDCs were isolated in 30/33 cases (91%) for a mean final yield of 24.5±2.9 million cells grown in 2.7±0.1 passages and 26.9±1.3 days.
MSCs have been shown to produce microvascular occlusion and infarction following intracoronary infusion unless certain precautions are taken.21-23 This is not surprising, as their average cell diameter measured by light microscopy is 21.0±3.3 μm, exceeding the typical capillary luminal diameter of 7-10 μm.24 At 20.6±3.9 μm, CDCs are similar in size to MSCs (p=0.52), and larger than BMMCs, which measured 8.6±1.8 μm (p<0.0001; Figure 1a). Like MSCs, a pilot study of intracoronary infusion of 5×107 allogeneic CDCs resuspended in PBS containing Ca2+ resulted in a large infarction (see Supplemental Data).
Two major factors that contribute to microvascular obstruction from cell infusion include clumping and thrombus formation. To determine whether CDCs would clump prior to infusion, we suspended CDCs in PBS with Ca2+ (PBS+) or without (PBS-) for 1 hour at room temperature. Microscopic examination showed no clumping, regardless of whether the solution contained Ca2+ (Figure 2a).
To assess CDC thrombogenicity, we mixed cells suspended in PBS- or PBS+ with allogeneic arterial blood, and measured thrombus formation at room temperature. Visible thrombus began to form after 90 minutes in PBS-, but after only 40 minutes in PBS+. Total thrombus in either solution after 8 hours was directly related to the number of CDCs in suspension (Figure 2b). To rule out an immunologic reaction, we performed the same experiment using autologous blood and found no difference in the rate or amount of thrombus formed compared to allogeneic samples (data not shown).
To compare the relative thrombogenicity of different progenitor cells, we measured thrombus formation using 1×105 BMMCs, MSCs, or CDCs at 1 hour (Figure 2c) or 8 hours (Figure 2d) after mixture with allogeneic arterial blood. In this condition, MSCs and CDCs are more thrombogeneic than BMMCs. The presence of Ca2+ in the cell infusion solution significantly increased the amount of thrombus formed at 1 hour in all cells types, although the difference was no longer significant at 8 hours. To begin to determine why CDCs and MSCs are more thrombogenic than BMMCs, we measured tissue factor. CDC and MSC lysates had similar tissue factor concentrations; both were significantly higher than BMMCs (p<0.001; Figure 1b).
To limit ex vivo thrombus formation, we added heparin (100U/ml, or 200U/ml) to the PBS- cell suspension solution; 100U/ml sufficed to eliminate cell-related thrombus formation in all 3 cell lines at both 1 and 8 hours (Figure 2c,d).
To achieve measurable engraftment, but no infarction after intracoronary CDC infusion, we delivered 105, 106, or 107 luciferase+ CDCs using multiple infusion solutions in healthy farm pigs (n=2 for each condition); luciferase activity was measured 24 hours later. We defined the target delivery area of the heart as the LV septum and anterior wall (Figure 3a, areas 1-4), non-target areas as LV posterolateral and RV free walls (areas 5, 6), and non-target organs as lung, liver, spleen and kidney. Luciferase activity was reliably detected in the target area only when we infused 107 CDCs. At lower doses, minimal activity was detected, even in the target area (Figure 3b). No luciferase activity was detected in non-target areas or organs at any dose. Distribution of luciferase activity within the target area was consistent in every animal: maximum engraftment in the septum with lower levels in the anterior wall (Figure 3b).
To assess myocardial infarction following CDC delivery we measured serum TnI 24 hours after infusion (Figure 3c,d). Delivery of 107 CDCs resuspended with PBS- caused a sizable infarction with TnI of 21.0±14.6 ng/ml. The addition of 100U/ml heparin to the PBS-infusate reduced cardiac TnI without changing luciferase activity in the target area (Figure 3c). We then added 250μg/ml nitroglycerin (NTG), which is known to increase adenoviral gene transfer efficiency,25 to PBS- with heparin 100U/mL. This reduced post-infusion TnI to 0.71ng/ml, near the upper limit of normal (≤0.50ng/ml). Further experiments were conducted using this optimized infusion solution of PBS- with 100U/mL heparin and 250μg/mL NTG.
Using the optimized delivery protocol in the infarcted heart, we infused 106, 107, 2.5×107, or 5×107 luciferase+ autologous CDCs in farm pigs 4 weeks post-MI, measuring short-term engraftment by luciferase activity, and TnI 24 hours later to exclude infarction. CDC engraftment was detectable at all CDC doses, with highest engraftment in the septum; engraftment was minimal in non-target areas and organs. Maximum and total luciferase activity increased with progressively increasing CDC dosage (Figure 3d). TnI remained within normal limits (<0.5ng/ml) with <2.5×107 CDCs. Above this threshold, TnI increased in a dose-dependent fashion (Figure 3d), indicating that infusion of 107 CDCs in the infarcted heart achieves measurable early engraftment of CDCs without thromboembolic complication; doses ≥2.5×107 achieve greater engraftment, but risk infarction.
To assess long term CDC engraftment and differentiation, intracoronary infusion was performed in 8 adult miniature pigs 4 weeks after infarction, 2 animals each receiving placebo (infusate alone), 105, 106 or 107 autologous lacZ+ CDCs. At sacrifice 8 weeks later, mature lacZ+ cardiomyocytes and vascular cells were identifiable in the peri-infarct zone of both animals receiving 107 CDCs. Figure 4a shows examples of cardiomyocytes with lacZ+ nuclei separated by connective tissue in the infarct border zone; 4b illustrates cells with lacZ+ nuclei lining an arteriole. By virtue of their location, size and appearance, the latter are likely endothelial cells.13 To confirm the cardiomyocyte identity of the large non-vascular cells, co-expression of beta-galactosidase with α-sarcomeric actin was confirmed by immunohistochemistry (Figure 4c). No lacZ+ cells were identified in animals receiving placebo or lower doses of CDCs.
Having shown that CDCs can engraft and survive following infusion at doses that do not cause infarction, we set out to determine whether CDC delivery affects post-MI remodeling. Animals in the short-term engraftment experiment had an average weight at infusion of 38.1 kg, thus those animals receiving 107 CDCs received ~260,000 CDCs/kg. For extrapolation to clinical trials we used a dose of 300,000 CDCs/kg, approximating the dose that resulted in engraftment and well below the 650,000 CDCs/kg (~2.5×107) dose that caused significant infarction. To this end, 16 adult miniature swine were randomized to receive 300,000 autologous CDCs/kg or placebo 4 weeks after infarction. Fourteen animals (7 CDC-treated, 7 placebo-treated) completed the protocol and were used for final analysis. Of the 2 animals that did not complete the protocol one animal randomized to CDCs died unexpectedly 10 days after infusion; an autopsy showed significant heart failure. The second animal was randomized to placebo, but had a non-cardiac injury forcing removal from the protocol prior to infusion.
At MRI prior to infusion, animals on average had large infarcts (18.5±1.8% of LV mass) and significant LV systolic dysfunction (LVEF 38.6±2.3%); there was no difference in infarct size, LVEF or other MRI parameters between animals assigned to either group (see Table 1). At the time of infusion, all animals had a patent LAD and hemodynamic function did not differ between the two groups. Twenty four hours after infusion, TnI was mildly elevated in both groups (0.74±0.37ng/ml in CDC vs 0.95±0.48ng/ml in placebo), but with no significant difference between them (p=0.73).
Eight weeks later, analysis of delayed gadolinium enhancement MRI data from 7 CDC-treated animals and 6 placebo (baseline gadolinium-enhanced images were not obtained in 1 placebo animal, precluding comparison) showed relative infarct size (% of LV infarcted as a fraction of LV mass) decreased significantly in the group that received CDCs (19.2±2.5% to 14.2±3.1%, p=0.01), but not placebo (17.7±2.9% to 15.3±3.0%, p=0.22; Figure 5). Total LV mass increased with CDC infusion (58.8±3.8g to 76.7±6.1g, p=0.02), but was of only borderline significance with placebo (60.0±2.8g to 72.1±5.0g, p=0.05). Absolute infarct mass decreased modestly, albeit insignificantly, with CDCs (11.0±1.3g to 10.6±2.6g, p=0.89), but remained unchanged in placebo (10.5±1.6g to 10.6±1.7g, p=0.92). Thus, the decrease in relative infarct size in the CDC-treated animals reflects an underlying increase in LV mass coupled with a modest decrease in infarct mass. There was, however, no significant difference in final infarct size or LV mass between CDC-treated animals and placebo in unpaired analysis (see Table 1).
Analysis of left ventricular chamber size and LVEF by MRI at the time of infusion and 8 weeks later (n=7 in both groups) suggests that CDC infusion slows adverse remodeling. While CDC-treated animals had non-significant increases in both end-diastolic volume (EDV): 40.7±3.4mL to 49.6±4.3mL (p=0.10), and end systolic volume (ESV): 25.8±3.3mL to 31.8±4.4mL (p=0.26), in placebo animals EDV was significantly increased: 45.3±1.8mL to 58.6±4.3mL (p<0.01), and ESV nearly so: 27.7±2.5mL to 37.5±4.3mL (p=0.06). LVEF was preserved with CDC infusion (37.8±3.4% to 37.6±4.3%, p=0.98), but tended to decrease modestly in placebo (39.5±3.5% to 37.0±3.5%, p=0.59).
Hemodynamic measurement 8 weeks after infarction revealed animals treated with CDCs (n=7) had a greater dP/dt maximum than placebo (n=6): 2003±178 vs 1538±104 mmHg/sec2 (p=0.04; Figure 6a; also see Table 1) and a lower dP/dt minimum: -3021±375 vs -2007±219 mmHg/sec2 (p=0.03; Figure 6b). One placebo-treated animal was excluded from the analysis due to technical problems.
Provocative EPS revealed a high rate of inducible VT in both groups: 4/6 CDC-treated and 4/6 placebo were inducible, rates consistent with previous observations in porcine models;18,26 one animal in each group did not undergo EPS. While there was no apparent benefit to CDC infusion, there was also no suggestion of a proarrhythmic effect, which is a major concern in cell therapy for the heart.14,15
Coronary angiography 8 weeks after infusion found all infarct arteries patent with normal blood flow. Autopsy with examination of the heart, lungs, liver, spleen, kidneys, and brain 8 weeks after intracoronary infusion revealed no evidence of neoplastic or non-target cardiac tissue growth in either group. There were no significant differences in standard clinical laboratory values between groups during the protocol (see Supplementary Table 1).
We find that, following the systematic optimization of intracoronary infusion, autologous CDCs may be safely delivered in a pre-clinical model of ischemic cardiomyopathy. Furthermore, delivered CDCs have the ability to engraft, form mature cardiac cells and exert positive effects on infarct size, left ventricular remodeling and hemodynamic function. The study was conducted using standard clinical equipment and techniques in a model designed to replicate a clinical scenario in which CDCs might be used, namely the treatment of post-infarct LV remodeling and dysfunction.
Intracoronary infusion is an attractive method for cell delivery to the heart because it is widely available clinically, can be peformed in a minimally-invasive manner and has been used in a number of clinical trials.2-7,27-8 The larger diameter and greater ex vivo thrombogenecity of MSCs and CDCs compared to BMMCs described here may explain why high doses of BMMCs have been delivered in clinical studies (as many as 3 × 108 cells) without report of thrombotic complications,27 while MSCs have been reported to cause infarction after intracoronary infusion unless special precautions are taken.21-23 Our results show that with optimization, CDCs, like MSCs, can be safely delivered via intracoronary infusion, and furthermore, that at this dose CDCs can engraft and positively affect LV remodeling after infarction. And while BMMCs have been delivered at much higher doses in clinical trials, long-term follow-up has suggested that the benefits of BMMC transplantation may be transient.28 Indeed, the smaller size and non-thrombogenic characteristics of BMMCs may paradoxically result in lower relative retention in the heart following infusion.29 Ultimately, clinical trials are necessary to compare the relative effectiveness and durability of different cell types. This study, however, clearly shows that optimization of intracoronary CDC infusion is critical to achieve a positive effect without complications.
The paradigm proposed for clinical CDC isolation tested in this study shows that endomyocardial biopsy after infarction can be safely performed and results in growth of large numbers of CDCs. Clinically, it is unlikely that patients would undergo endomyocardial biopsy immediately after infarction, and one would expect that biopsy performed days to weeks after infarction would only be safer than in the model described here. The data presented here also indicate that, despite an acute infarction, in the majority of cases (91% of animals in this study) large numbers of CDCs can be isolated and grown. In humans, CDC isolation has also been successful in the vast majority of cases, even with biopsies from patients with chronic heart failure.13 Furthermore, our results suggest that CDCs delivered to the heart following such isolation are effective, having a positive impact on infarct size, remodeling, and hemodynamic function.
While relative infarct size was reduced in CDC-treated animals, it should be noted that there was not a significant change in absolute infarct size in either group, or between groups. Instead, the change in relative infarct size in the CDC-treated animals was due primarily to a significant increase in LV mass, which was not seen with placebo. If cardiac regeneration were to occur after cell therapy, one would predict either an increase in viable myocardium, a decrease in infarcted myocardium, or both. The present data suggest the first possibility, and the improved hemodynamics indicate that the increase in LV mass had positive functional benefits.
The global effects on LV mass and adverse remodeling seen in this study also suggest that CDCs may exert effects on sites outside the targeted delivery zone. While our work here showed that CDCs have the ability to form both mature cardiac myocytes and vascular cells in the peri-infarct zone, no evidence of significant engraftment was found in remote, “off-target” myocardium. This suggests that any remote effects of intracoronary CDC delivery results from translation of mechanical effects in the infarct zone to remote areas, or from paracrine factors. There is evidence that CDCs generate growth factors such as fibroblast and hepatocyte growth factors both in vitro and in vivo.30 Further study is necessary to determine how important either of these effects might be, and whether the resulting increase in LV mass is primarily due to hypertrophy of existing cardiomyocytes, or to tissue regeneration.
The ability of several different types of stem cells (both cardiac and non-cardiac) to improve cardiac function is currently being evaluated. While the limitations of CDCs include the need to obtain autologous myocardial tissue13 and the length of time necessary to grow the cells, the method described here is minimally-invasive and highly successful (91% of biopsies resulted in CDCs). In addition, the time necessary for CDC growth is significantly shorter than that required for other methods to produce cardiac-derived stem cells.9,10,12 While large numbers of bone-marrow derived stem cells can be obtained in a few days, the clinical effects of these treatments remain modest2-7,27-8 and the ability of such cells to truly regenerate myocardium remains uncertain.31-35 While it can be argued that the effects of CDCs reported here were modest as well, these effects were achieved with what appears to be low-level engraftment. Recent work suggests that the use of bFGF can enhance CDC engraftment and synergistically increase the positive effect of CDCs following infarction.36 Synergistic improvement of the results reported here could provide an important new tool for the treatment of cardiomyopathy.
While we made every effort to replicate the clinical conditions in which CDCs might be employed, this study was nevertheless performed in an iatrogenic model of myocardial infarction in a surrogate species, and admittedly, in a small number of animals. In this model, we have demonstrated that tissue for CDC growth can be safely obtained following myocardial infarction, that the resulting cells can be safely delivered via intracoronary infusion, and that there is a potential benefit from doing so. Taken together, the results show the potential promise of CDCs in the treatment of post-MI ventricular dysfunction, however, only clinical trials will determine whether this promise can be translated to patients. To this end, the CADUCEUS study, based on the technology presented here, has recently been initiated (see clinicaltrials.gov for details).
Funding Sources: Major funding for this work was from the NIH Specialized Centers for Cell Therapy (grant number U54HL081028). Support was also provided by the Donald W. Reynolds Foundation.
Conflict of Interest Disclosures: Eduardo Marbán owns equity in Capricor Inc. No funding for the research described here was provided by Capricor Inc.