Intravenous Autologous Bone Marrow Mononuclear Stem Cell Therapy for Ischemic Stroke
A Multicentric, Randomized Trial
Background and Purpose—Pilot studies have suggested benefit from intravenous administration of bone marrow mononuclear stem cells (BMSCs) in stroke. We explored the efficacy and safety of autologous BMSCs in subacute ischemic stroke.
Methods—This was a phase II, multicenter, parallel group, randomized trial with blinded outcome assessment that included 120 patients. Patients with subacute ischemic stroke were randomly assigned to the arm that received intravenous infusion of autologous BMSCs or to control arm. Coprimary clinical efficacy outcomes were Barthel Index score and modified Rankin scale at day 180. Secondary outcomes were change in infarct volume, National Institute of Health Stroke Scale (NIHSS) at day 90 and 180. Main safety outcomes were adverse events, any new area of 18fluorodeoxyglucose positron emission tomography uptake in any body part over 365 days.
Results—Fifty-eight patients received a mean of 280.75 million BMSCs at median of 18.5 days after stroke onset. There was no significant difference between BMSCs arm and control arm in the Barthel Index score (63.1 versus 63.6; P=0.92), modified Rankin scale shift analysis (P=0.53) or score >3 (47.5% versus 49.2%; P=0.85), NIHSS score (6.3 versus 7.0; P=0.53), change in infarct volume (−11.1 versus −7.36; P=0.63) at day 180. Adverse events were also similar in the 2 arms, and no patient showed any new area of 18fluorodeoxyglucose uptake.
Conclusions—With the methods used, results of this hitherto first randomized controlled trial indicate that intravenous infusion of BMSCs is safe, but there is no beneficial effect of treatment on stroke outcome.
- adult stem cells
- bone marrow cells
- cell- and tissue-based therapy
- cerebral infarction
- randomized controlled trial
- stem cell transplantation
WHO estimates that globally ≈15 million people experience stroke annually, of which 6 million die and 4 million are left with significant disability.1 Despite advances in acute care and secondary preventive strategies, stroke remains a major burden on healthcare system worldwide. Intravenous thrombolysis is the only approved therapy for acute ischemic stroke.2 However, few stroke patients receive this therapy because of its narrow time window.3 There is a clear need for a treatment that improves neurological outcome after stroke. In experimental model of stroke, intravenous, intrastriatal, and intraarterial infusion of mononuclear stem cells from bone marrow have improved neurological outcome through reduced apoptosis, decreased peri-infarct inflammation, and angiogenesis.4–6 The transdifferentiation of transplanted BMSCs has also been reported.7 Phase I studies have shown feasibility and preliminary safety of intravenous infusion of autologous BMSCs in acute and subacute ischemic stroke,8,9 but no randomized study has been reported. Yet, there is immense interest about this therapy among health professionals, media, and public. Many private for-profit healthcare facilities are offering this therapy at high cost to unsuspecting and desperate patients weeks to months after stroke without any randomized evidence.10
We conducted a randomized, multicenter study to investigate the effects of intravenous infusion of autologous BMSCs in subacute ischemic stroke (Intravenous Autologous Bone Marrow Mononuclear Cell Therapy for Ischemic Stroke [InveST]). We tested the hypothesis that in patients with subacute ischemic stroke, intravenous infusion of autologous BMSCs between 7 to 30 days after onset results in reduction of infarct volume and improvement in neurological function at day 180 of follow-up compared with those without the infusion.
Subjects and Methods
Study Design and Setting
This study was a phase II, randomized, multicenter, open-label, parallel group trial with blinded end point assessment (Figure 1). We included patients with subacute ischemic stroke from 5 postgraduate teaching hospitals fully funded by Government of India.
Standard Protocol Approvals, Registration, and Patients Consents
The Department of Biotechnology, Government of India, Ethics Committee and all the center ethics committees approved the protocol. The trial was registered at the Clinical Trial Registry-India (CTRI-PROVCTRI/2008/091/00046) and clinicaltrial.gov (NCT01501773). All patients or their legally authorized representatives gave written informed consent. All the procedures were followed in accordance with institutional guidelines.
Patients were randomly assigned using permuted block randomization in a 1:1 ratio to receive either intravenous infusion of autologous BMSCs (BMSC arm) or to a control treatment where neither BM aspiration nor sham infusion were performed. The central data management office used a computer to generate the randomization sequence stratified by center with a varying block size of 4, 6, and 8. To ensure allocation concealment, centers had to contact the central office using e-mail, fax, or telephone. The office allotted a unique identification number to each patient and intimated allocation within 12 hours of the contact. Care providers and patients were not masked; however, the outcome (modified Rankin scale [mRS], Barthel Index [BI], and imaging) assessors were masked to treatment allocation.
Inclusion and Exclusion Criteria
Inclusion criteria were an age of 18 to 75 years, computed tomography, or MRI scan of the head showing relevant infarct within the middle cerebral artery or anterior cerebral artery territory and excluding a hematoma, onset of stroke between 7 and <30 days, Glasgow Coma Scale score >8, BI score of ≤50, National Institute of Health stroke scale (NIHSS) score of ≥7 and inability to walk unaided or raise upper limb by 90° and clinically stable condition for ≥48 hours. Definition of stable patients and exclusion criteria are given in Methods Ia and Ib in the online-only Data Supplement, respectively. Study oversight and quality control measures are given in Methods II in the online-only Data Supplement.
At each center, bone marrow was aspirated aseptically from the posterior iliac crest under local anesthesia (1% lidocaine). BMSCs were separated by ficoll density separation method. Quality control criteria for the release of cell product included bacterial sterility, >90% viability by trypan blue (Himedia, India) exclusion, and cell morphology by Giemsa’s staining. Immunophenotyping to count CD34+ cells was done using flow cytometry (see Methods III in the online-only Data Supplement for details). The cells suspended in phosphate-buffered saline were infused into antecubital vein of the patients in BMSC arm.
Monitoring for Infusion-Related Toxicity
Besides hematologic indices, pulmonary, renal, hepatic, and renal systems were monitored. Neurological assessment was performed daily until day 7 or hospital discharge.
Methods and Measurement
Baseline assessment included demographic and clinical details and MRI of brain, electroencephalography, whole body 18fluorodeoxyglucose positron emission tomography (PET) scan (in 4 centers), NIHSS, and BI. All patients received conventional treatment according to current guidelines,11 but the treatment arm, in addition, received BMSCs intravenously within 48 hours of random allocation. No medical therapy was delayed or denied because of patient’s participation into the study.
In-Hospital Assessment and Follow-Up Schedule
Follow-up assessment and investigations were done on day 7 (including mRS), day 90, day 180, and day 365 to evaluate safety and efficacy. Follow-up MRI scans were performed at day 90 and day 180. Electroencephalographies were done at day 180 and day 365, and 18fluorodeoxyglucose-PET scan was required at 4 centers having this facility at baseline, day 180, and day 365. mRS, BI, and NIHSS were administrated on day 90, day 180, and day 365 at each site. A central telephonic follow-up for all patients were done from coordinating center at day 90, day 180, and day 365 by a trained and blinded assessor (unaware of patient arm) who determined the vital status and administered the scales to surviving patients.
Magnetic Resonance Imaging
Standardized sequences were obtained including T1, T2 series, and a fluid-attenuated inversion recovery sequences. Volumetric analysis was performed on fluid-attenuated inversion recovery images at the central core laboratory using Analyze software (version 8.1; AnalyzeDirect, Inc, KS) to determine the volume of infracts and volume of both lateral ventricles. All images were saved in CDs and mailed to the coordinating center where they were stored, anonymized, and analyzed.
The coprimary clinical efficacy outcomes were measured by BI score and mRS score at day 180 assessed centrally by the assessor by telephone. The secondary clinical efficacy outcome was NIHSS score at day 90, day 180, and day 365. Outcome assessors were blinded to intervention and baseline scores. (A brief description of BI, NIHSS, and mRS can be found in Methods IV in the online-only Data Supplement.) The worst value on the 3 scales (mRS-6, BI-0, and NIHSS-42) was assigned to patients who died. Secondary imaging efficacy outcome was change in infarct volume between baseline and day 90 and day 180. All imaging analyses were performed centrally on deidentified data. Analysts were unaware of the treatment arms and clinical information. The safety outcomes included death, adverse events (serious and nonserious), epileptiform discharges in electroencephalography, and evidence of any new growth on PET scan at day 365.
Sample size was calculated for superiority hypothesis on the BI and mRS. Calculation based on the data in the only published controlled trial of stem cell (mesenchymal)12 yielded a sample size of 16 (standardized effect size=2), but our study used BMSCs at lower doses, and hence, we hypothesized only one third of the effect size (standardized effect size=0.68; with 90% power and α level of 5%). Calculation with standard formula13 yielded a sample size of 45 per group. Adjusting for 10% losses to follow-up and 1 interim analysis, we estimated a sample size of 120 in total. This confers 90% power to detect standard effect size of 0.6 on mRS scale score at 5% significance level.13 However, interim analysis was abandoned as recruitment had completed by the time criterion (6-month follow-up in 60 patients) for the analysis was fulfilled.
Statistical analysis was based on the scores obtained by central telephonic assessment and followed intention-to-treat principle. Descriptive statistics included mean for numeric data with normal distributions, median otherwise, and proportions for categorical data. Primary efficacy analysis was shift analysis of modified Rankin scale scores using the Cochran-Mantel-Haenszel test with SAS version 9.2, unadjusted and adjusted BI score and mRS score. Secondary outcomes were analyzed using unadjusted Student t test for means and χ2 test for proportions. The analyses were repeated after adjustment for baseline scale scores, NIHSS, and infarct volume using regression methods. All data analyses (other than shift analysis) were done using SPSS v 17.0. Stratified analyses were done using RevMan v 5.1. All analyses were 2-sided, and P values <0.05 were considered statistically significant.
Between November 2008 and June 2010, 5 centers screened 423 patients and included 120 patients meeting the eligibility criteria for the study. Of these, 60 each were assigned to the BMSC or control arm. One patient in each arm withdrew consent soon after randomization, but the one in control arm agreed for telephonic follow-up.
Baseline characteristics (Table 1) were well balanced between the 2 arms, except infarct volume, which was 24.84 cm3 higher in control arm than in BMSC arm. None of the patients had received intravenous tissue-type plasminogen activator or endovascular therapy.
All patients received the conventional treatment. Bone marrow aspiration was successfully completed without any adverse event in 58 patients in the BMSC arm (1 withdrew and 1 missed because of logistical difficulty). The aspiration yielded 108.9±33.9 mL of aspirate. The mean number of mononuclear cells infused was 280.75 million (SD, 162.9) containing CD34+ cells of 2.9 million (SD, 2.8). BMSCs viability was 93.2% (SD, 5.75). Median time from onset to cell infusion was 18.5 days (interquartile range [IQR], 9.2) and from randomization 1 day (IQR, 1). Median time from aspiration to infusion was 3.20 (IQR 0.4) hours. All patients were monitored in hospital for a minimum of 48 hours. No patient had any hemodynamic, pulmonary, or allergic complication during BMSCs infusion.
Central telephonic follow-up was complete for 116 (96.7%) at day 90, 118 (98.3%) at day 180, 117 (97.5%) at day 365, and 118 (98.3%) at the end of the study. Five (8.4%) of 59 patients in BMSC arm and 5 (8.3%) of 60 in control arm died before day 180. Three more patients died at day 195, day 206, and day 221 in BMSC arm. Median follow-up time was 640 (IQR, 221) days in control arm and 634 (IQR, 229) days in BMSC arm.
Ninety-four (84% of survivors) patients had MRI at day 90 and 87 (81% of survivors) at day 180. Of these, 84 (89%) were analyzable at day 90 and 71 (82%) at day 180. Sixty-four (59%) patients had whole body-18fluorodeoxyglucose-PET scan at day 180 and 49 (47%) at day 365. Seventy-nine (73%) patients had electroencephalography at day 180 and 78 (74%) at day 365.
The BI score at day 180 showed no difference between the BMSCs and control arm (Table 2). Analysis adjusted for infarct volume, baseline NIHSS, and baseline BI did not change the results. BI score at day 90 and day 180 of the 2 arms was also similar (Figure 2). Scores of mRS in control arm versus BMSC arm at day 180 showed no difference between the BMSC and control (Table 2). Cochran-Mantel-Haenszel shift analysis of the scores did not reveal any statistically significant difference (P value=0.56 [unadjusted]; 0.53 [adjusted for infarct volume and day 7 mRS]; Figure 3).
No significant difference in NIHSS score and change in infarct volume at day 90 and day 180 were observed between the BMSCs and control arm (Table 2). No relationship was observed between cell dose and outcomes (Results I and Table I in the online-only Data Supplement). Subgroup analysis did not show statistically significant interaction between cell dose and NIHSS and side of hemiplegia (Results II in the online-only Data Supplement).
Kaplan–Meier survival curve was comparable between the 2 arms (Figure I in the online-only Data Supplement). The adverse events and serious adverse events were also comparable between the 2 arms (Table II in the online-only Data Supplement). Whole body 18fluorodeoxyglucose-PET did not reveal any new growth over 180 days in 64 patients and over 365 days in 56 patients (Figure II in the online-only Data Supplement). Electroencephalographies revealed epileptiform discharges in 3 patients, all in BMSCs arm but 2 before infusion and only 1 patient at both days 180 and 365 postrandomization.
Our study is hitherto the first and the largest randomized controlled trial comparing intravenous infusion of autologous BMSCs and control in patients with subacute ischemic stroke. The findings of this open-label trial with blinded outcome assessment indicate that intravenous infusion of BMSCs at a mean of 18.5 (IQR, 9.25) days after onset of ischemic stroke is safe but does not improve neurological outcome compared with the control arm. However, this is the first study to support safety of BMSCs in a comparative study.
Our results do not confirm the reported benefit in the preclinical studies4–6 and pilot clinical studies.8,9 Several possible reasons may pertain to eligibility criteria or timing, dose, and route of cell administration. We address these one by one. Our eligibility criteria was guided by our pilot studies and aimed to select patients who had moderate, not mild or severe stroke. Although mild strokes have uniformly good outcome, severe stroke have generally poor outcome unlikely to respond to the intervention. Therefore, we selected moderately severe stroke. As regards timing, preclinical evidence would favor administering the intervention within the first few days, although marrow stromal cell therapy administered ≤30 days after stroke has been shown effective in rodent models,14 and our pilot study supported this result.9 However, it may be noted that cell type used in the present study was different from marrow stromal cell. We decided to administer the cells from second to fourth week for 2 reasons: First, we were aware from our pilot study that many patients with NIHSS >7 tend to deteriorate in the first week and require hemicraniectomy. Clearly the effect of hemicraniectomy, even if balanced between the groups, would have masked the effect, if any, of the cell therapy. Second, we also considered that patients are being stabilized in the first week and cytokines released from the intervention may destabilize the patients, increase serious adverse events, and prompt the Data Safety Monitoring Board to stop the trial, a threatening proposition for an emerging therapy in 2007 when the study was planned. In one study8 of similar patients, 2 of 10 patients required hemicraniectomy, prompting the investigators to change the eligibility criteria to avoid such patients while we designed our study to avoid the high-risk period of first week. However, late administration of cells remains a possible explanation for lack of benefit observed in this study. Still, the findings are relevant because many practitioners are offering cell therapy to patients with stroke after weeks and months of onset.
Was the dose of cells sufficient? One of our study objectives was to examine any relationship between cell dose and effect. Our patients received a median dose of 268 million cells, the estimated effective dose from our pilot study, although highest quartile had >425 million cells. We did not find any dose-response relationship, raising doubt about cause and effect relationship between BMSCs infusion and outcome, and suggesting that inadequate dose is not a likely explanation for failure to find benefit with the BMSCs. In myocardial infarction also, there has been no significant correlation between BMSCs cell dose and effect.15
The issue of appropriate route of therapy is not yet settled, but >19 preclinical studies showing benefit with cell therapy have used intravenous route. Moreover, comparative studies in rodent models have shown greater or similar benefit associated with intravenous route than through intra-arterial or intracerebral route.16,17 The route of delivery does not seem to explain the lack of significant benefit in our study.
Several features of this study increase the confidence in the internal validity of findings. Experimental design, central and independent group for sequence generation, and concealed allocation from a remote site, low attrition, and blinded central outcome assessment by a trained assessor limit the risk of bias in our study, thus supporting its internal validity. However, generalizability of the findings only to conditions of the study, not for stroke within 1 week or after 1 month of onset, is one of its limitations. Future studies should focus on stroke within 1 week of stroke onset.
Limitations of this study include lack of blinding of patients or physicians, but as required by ethics committee, BM aspiration or sham infusion was not performed in our control arm. Potential bias related to patient or physician behavior and infarct volume imbalance might be expected to favor the BMSC arm, yet this arm did not perform better than the control. The study has limited power but reasonably meets the requirement for a phase II study. Although we excluded patients with Glasgow Coma Scale score ≤8, yet it is possible that our study population had too severe stroke (mean infarct volume 99.3) to benefit from cell therapy. Future studies need to take this into account.
In conclusion, under the conditions of InveST trial, BMSC is safe but ineffective in the treatment of moderately severe subacute ischemic stroke. Until ongoing or further randomized trials show efficacy, this treatment should not be used in clinical practice, and patients should not accept such therapy without question.
We thank all the participating patients, nurses, and trial staffs at each of the study centers. A special thanks to Dr Philip Michael Bath for his help with the manuscript revision.
Sources of Funding
The multicentric InveST trial was supported by Department of Biotechnology, Ministry of Science and Technology, Government of India.
*Complete lists of investigators and collaborators are listed in Material I in the online-only Data Supplement.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.007028/-/DC1.
- Received August 7, 2014.
- Revision received October 2, 2014.
- Accepted October 3, 2014.
- © 2014 American Heart Association, Inc.
- Mackay J,
- Mensah G
- Whiteley WN,
- Thompson D,
- Murray G,
- Cohen G,
- Lindley RI,
- Wardlaw J,
- et al
- Lloyd-Jones D,
- Adams RJ,
- Brown TM,
- Carnethon M,
- Dai S,
- De Simone G,
- et al
- 10.↵Tour2India4health: Medi Tours and Health Tourism Services Web site. http://www.tour2india4health.com/Stem-Cell-Therapy-in-India.html. Accessed January 15, 2014.
- Hulley SB,
- Cummings SR,
- Browner WS,
- Grady DG,
- Newman TB
- Browner WS,
- Newman TB,
- Hully SB
- Yang B,
- Migliati E,
- Parsha K,
- Schaar K,
- Xi X,
- Aronowski J,
- et al