Total Ischemic Time as an Independent Predictor of Response to Stem Cell Therapy in Patients with ST Segment Elevation Myocardial Infarction

1 Department of Internal Medicine, University of Texas Health Science Center at Houston, Houston, TX 77030, USA 2Division of Cardiovascular Medicine, University of Texas Health Science Center at Houston, Houston, TX 77030, USA 3 Cardiology Service, Ankara Mevki Military Hospital, 06110 Ankara, Turkey 4Center for Clinical and Translational Sciences, University of Texas Health Science Center at Houston, Houston, TX 77030, USA 5Utah Cardiology PC, Layton, UT 84041, USA


Introduction
Autologous bone marrow stem cell (BMSC) injection in patients with ST segment elevation myocardial infarction (STEMI) has been investigated as a new treatment strategy for the past decade.After the first encouraging pilot study [1] in 2002, many trials have since been published.Although several studies failed to show any favorable outcomes from BMSC [2][3][4][5], there have been many trials which demonstrated the beneficial effects of BMSC [6][7][8][9][10][11][12].Several meta-analyses have confirmed that BMSC injections improved left ventricular systolic function, on average, by 3% [13][14][15].Despite the promising results in these studies, there were many inconsistencies in the selection criteria; incongruent cell types, varying administration time, and variable injection routes are a few examples.Selected patients were also heterogeneous in terms of baseline left ventricular ejection fraction (LVEF) and time from symptom onset to reperfusion.Unfortunately, it is still difficult to identify the best model to demonstrate the effectiveness of stem cell therapy.
In this study, we attempted to define an independent parameter which can increase the efficacy of BMSC: we focused on whether the total ischemic time (TIT), defined by the time of symptom onset to the time of percutaneous coronary intervention (PCI), was associated with LVEF improvement in response to BMSC administration in STEMI patients.

Study Selection Criteria.
The study selection criterion was adapted from Zhang et al. [13], and it aimed to investigate the effect of BMSC administration on LVEF improvement as a function of total ischemic time and baseline LVEF.
2.1.1.Inclusion Criteria.Studies that included STEMI patients who had PCI performed within 12 hours of symptom onset and BMSC treatment.Studies which were included were either randomized or placebo controlled trials, having at least 10 patients in each respective trial and a reported clinical endpoint of LVEF improvement.Only English language literature was searched.

Exclusion Criteria.
Greater than 12 hours of symptom onset to PCI, reperfusion other than PCI, BMSC delivery other than intracoronary cell delivery (e.g., intramyocardial, peripheral blood, etc.), use of stem cells other than BMSC, and the use of granulocyte colony-stimulating factor to induce BMSC release.Nonrandomized or noncontrolled trials, and trials with less than 10 patients in each respective study were also excluded.

Data Search.
We performed a systematic literature search for randomized controlled clinical trials evaluating BMSC effects on LVEF and left ventricular end systolic volume (LVESV) in STEMI patients who underwent PCI.PubMed, Ovid's MEDLINE, and Cochrane Evidence-Based Medicine (EBM) Reviews databases from January 1990 through August 2010 were searched to identify all eligible trials by using the following medical subject heading (MeSH) terms and text words: bone marrow, stem cell, progenitor cells, cell transplantation, cell therapy, myocardial infarction, and cardiac repair.Additionally, the reference lists of reviews were searched for all relevant studies.

Study Enrollment and Data
Extraction.Data for LVEF, LVESV, and major adverse cardiovascular events (MACE) were extracted and checked by three investigators according to our inclusion criteria and classified as short term (3 to 6 months) and long term (>6 months) according to follow-up periods.Left ventricular performance parameters were extracted from the magnetic resonance imaging (MRI) results, if they existed.Otherwise, echocardiograph (Echo), angiographic (Angio), and/or single photon emission computed tomography (PET) results were taken, respectively.

Data Synthesis and Statistical
Analyses.For statistical analysis, STATA 11 software (StataCorp, College Station, TX) and R (Development Core Team, Version 3.0.0)were used. values < 0.05 were considered statistically significant.
Meta-analysis [24][25][26] was conducted to evaluate the overall BMSC effect by comparing the mean difference between the BMSC and placebo group.Fixed and random effects model were performed according to the overall heterogeneity, which was examined with  2 and Cochran's chisquare test.The inverse variance method was applied to weigh the mean difference of each included study.
For analysis on LVEF, mean difference between the BMSC and placebo group, LVEF change , and their 95% confidence interval (CI) were calculated.In particular, LVEF change = LVEF BMSCchange − LVEF Placebochange , where LVEF BMSCchange and LVEF Placebochange are the mean change of the LVEF from the baseline in the BMSC and placebo group, respectively.Most of the studies reported the mean and standard deviation (SD) for LVEF BMSCchange and LVEF Placebochange .For those studies in which the mean and SD were missing, an estimation was made of the mean and SD based on the reported mean and SD of LVEF at baseline, during followup, and the estimated correlation coefficient between the baseline and follow-up LVEF within each group [27].The reported values in Schächinger et al. [8] and Yousef et al. [22] were utilized to estimate the correlation coefficients, within each group, for short and long term followup, respectively.Similarly, the mean difference of LVESV, LVESV change , and its 95% CI were evaluated.The weighted mean difference (WMD) was reported for each study and the statistical significance of the overall effect of all studies was performed by  test.
If the significance of heterogeneity was found, a metaregression analysis was performed to assess the relationship between LVEF change , TIT (<6 hours versus ≥6 hours), and baseline LVEF prior to PCI (<45% versus ≥45%).
For the MACE data set, the number of patients () who had cardiovascular deaths, rehospitalization in the form of heart failure (HF) or myocardial infarction (MI), or ventricular arrhythmias was modeled by the meta-analysis based on the Mantel-Haenszel method.The risk ratio (RR) and the 95% CI were calculated.

Selected Studies and Characteristics.
Five hundred fortyseven studies were identified from the electronic databases.After exclusion of duplicated or nonrelated articles, 43 papers were saved as references.Twenty-seven of these were excluded based on the exclusion criteria.
For LVEF data, a total of 20 articles reporting the results of 14 trials (13 trials with 855 patients for short term, and 7 trials with 466 patients for long term) were included in this study.Six articles in the long term group had the results of the same studies of those in the short term category [5,16,18,20,21,23].Eighteen trials contained LVESV data; four of them were excluded because the results were reported as LVESV index.For MACE analysis, we used 10 trials for the short term and 6 trials in the long term analysis.Additionally, in order to remove the effects of an outlier, we excluded one study [28] (since its mean PCI time was 20 hours) because this severely prolonged ischemic time left little hope for beneficial effects from PCI.
The characteristics of the studies, including patient number, follow-up months, TIT, and the method of LVEF   measures are summarized in Table 1.The baseline and post-PCI measurements of left ventricular parameters, as well as MACE, are shown in Table 2.The mean values for the baseline LVEF and LVESV between the BMSC and placebo group are grossly similar, with a range of 0-4.4% and 0-8 mL/m 2 , respectively.In Table 3, the studies are separated into short or long term followup and categorized based on their subgroup characteristics, as defined by TIT (<6 hours or ≥6 hours) and baseline LVEF prior to PCI (<45% or ≥45%).

Effect on LVEF.
Cochran's chi-square test for heterogeneity suggested significance on LVEF change for both short term ( < 0.0001) and long term followup ( < 0.0001) (Figure 1).Therefore, a meta-analysis based on a random effects model was performed to obtain the overall LVEF improvement within the BMSC group compared with placebo.The overall LVEF improvement demonstrated significant difference between the BMSC and placebo group, favoring the BMSC category with a mean difference of 3            Based on the evidence of heterogeneity on LVEF, a random effects meta-regression model was built to evaluate the relationship between the BMSC effect and two binary covariates (TIT, <6 hours or ≥6 hours, and baseline LVEF, <45% or ≥45%).
For both short and long term followup, the statistical significant regression coefficients suggest that the binary indicator of TIT (<6 hours or ≥6 hours) is an independent predictor affecting the improvement of LVEF when comparing the BMSC to the placebo groups (4.96%, 95% CI, 0.72 to 9.19,  = 0.02 and 6.24%, 95% CI, 0.46 to 12.02,  = 0.03, resp.).Although there was no statistical significance, there appeared to be a trend towards improved LVEF for the BMSC category within the LVEF < 45% subgroup, at short term and long term followup (0.49%,  = 0.82 and 1.55%,  = 0.67, resp.).

Effect on LVESV.
Cochran's chi-square test for heterogeneity suggested significance on LVESV change for long term ( < 0.0001) but not short term followup ( = 0.238) (Figure 3).Therefore, a meta-analysis based on fixed effects model and random effects model for short term was performed to obtain the overall LVESV benefits comparing the BMSC to the placebo group.Compared to placebo, the treatment effect of BMSC on LVESV improvement demonstrated a mean difference of −6.45 mL/m 2 (95% CI, −9.19 to −3.72,  < 0.0001) in the short term and −10.83 mL/m 2 (95% CI, −21.81 to 0.16,  = 0.05) in the long term followup.Because of nonsignificance of heterogeneity and the limited number of trials, a meta-regression analysis was not conducted on LVESV for both short term and long term trials.

Discussion
Overall, from the systematic review and meta-regression analysis, the total ischemic time (TIT), defined by the time of symptom onset to the time of percutaneous coronary intervention (PCI), is shown to be a significant factor for LVEF improvement when comparing the BMSC to the placebo group if intervention is performed after 6 hours for both short term and long term followup (Figure 2).Although there was also a trend towards improvement of LVEF with decreased TIT (<6 hours), the data was not statistically significant.
The reason for this effect can be multifactorial; it is our belief that with TIT < 6 hours, the rapid restoration of blood flow to the infarcted artery will reduce the damage to the myocardium, thus limiting the potential for benefit with the administration of BMSC.Accordingly, Denktas et al. [29] demonstrated that for patients presenting with STEMI, a TIT of less than 120 minutes allows for improved clinical outcomes because of reduced infarct size.Therefore, it is possible that the beneficial effects of BMSC will be masked for the subgroup of people who present with reduced TIT.Another potential reason that there was no statistical significance for LVEF improvement with decreased TIT can be that the patients included in our selected studies who presented with their first STEMI may have already had impaired LV function at baseline, which was not identified prior to their presentation.Thus, the improvement potential of LVEF during followup will be limited in that group of patients despite having a reduced TIT.

Figure 1 :
Figure 1: Forest plot of WMD with 95% CI on LVEF demonstrated favorable LVEF outcomes within the BMSC group compared with the placebo group.The size of data markers represents the statistical weight of each included study before it was pooled to the overall effect.(a) Trials in short term followup: heterogeneity:  2 = 53.63,df = 12,  < 0.0001, and  2 (variation in WMD attributable to heterogeneity) = 77.6%.(b) Trials in long term followup: heterogeneity:  2 = 63.86,df = 6,  < 0.0001, and  2 (variation in WMD attributable to heterogeneity) = 90.6%.

Figure 3 :
Figure 3: Forest plot of WMD with 95% CI on LVESV demonstrated favorable outcomes within the BMSC group compared with the placebo.The size of data markers represents the statistical weight of each included study before it was pooled to the overall effect.(a) Trials in short term followup: heterogeneity:  2 = 9.21, df = 7,  = 0.238, and  2 (variation in WMD attributable to heterogeneity) = 24%.(b) Trials in long term followup: heterogeneity:  2 = 19.9,df = 4,  = 0.0005, and  2 (variation in WMD attributable to heterogeneity) = 79.9%.
All values are listed as "mean (SD)" unless otherwise denoted.* Median (range).

6 ISRN Stem Cells Table 2 :
(a) Short term studies: baseline and post-PCI measurements of LV parameters and MACE.(b) Long term studies: baseline and post-PCI measurements of LV parameters and MACE.

Figure 4 :
Figure 4: MACE at short term and long term followup.
The difference in LVEF between the BMSC and placebo group after PCI.

Table 1 :
Characteristics of studies.