Baseline characteristics among the EnD-CAD patients (Table 1)
Among 68 subjects receiving CD34+ cell therapy, three patients in the phase II trial were excluded, including one case died of brain stem hemorrhage at one month, another one was expired of traumatic cervical spine injury with hypoxia at one month, and the other case refused to continue follow-up one week after cell delivery. A total of 65 study patients followed up for at least six months were selected for the analyses. All EnD-CAD patients expressed high-risk baseline profiles such as 100% for one of atherosclerotic risk factors, >76% for diabetes mellitus, >90% for hypertension, >87% for dyslipidemia, 70% for PCI history, 100% for multi-vessel CAD, and >50% for chronic total occlusion (CTO) at left anterior descending (LAD) artery. The age, gender and rates of body mass index, old stroke, old myocardial infarction (MI) and CABG did not differ between 30 responders and 35 non-responders.
The laboratory finding showed that the white blood cell count was significantly lower in responders than in non-responders. However, the platelet count, hemoglobin, estimated glomerulus filtration rate (eGFR) and serum creatinine, total cholesterol, high-density lipoprotein, low-density lipoprotein and triglyceride did not differ between the two groups.
In addition, responders had significantly higher prescription of beta blocker and lower diuretic use than the non-responders. A hundred percent of patients took antithrombotic agents and a majority of them received guideline-directed medical therapy.
Laboratory parameters after G-CSF injection, clinical presentations and results of examinations during follow-up, and one-year outcomes (Table 2)
The post-G-CSF laboratory findings demonstrated that the circulating levels of leukocyte count, hematopoietic stem cells, young cells, neutrophils, CD34+ cells, CD45+ cells and stem cell percentage measured by flow cytometric analysis, as well as troponin-I did not differ between the two groups.
The responders had significant lower angina severity and insignificantly less dyspnea compared with the non-responders prior to CD34+ cell therapy. Within the follow-up period, the responders had better clinical symptomatic improvement in both angina and heart failure as compared with the non-responders at the time points of every 3 months, indicating the improvement of cardiac systolic function (i.e., LVEF improvement ≥7.0%) was correlated with the relief of clinical symptoms.
Regarding objective evaluations for functional capacity, angiogenesis, chamber sizes and cardiac/valvular functions, there were no significant differences between groups at baseline. However, the responders had higher coronary angiogenesis score and less echocardiographic grade 2 or 3 diastolic dysfunction than the non-responders. Notably, difference of LVEF on 3D echocardiography between follow-up period and baseline began to be significant at 3 months after stem cell therapy, implicating good clinical and subclinical responses could be observed as early as 3 months since delivery of CD34+ cells. After one-year follow-up, composite endpoints occurred in nearly one half of EnD-CAD patients but did not differ between the two groups. Around 1 in 5 patients needed hospitalization for acute decompensated heart failure, and nearly 1 in 4 patients received salvage myocardial revascularization strategy for relief of refractory angina in both groups.
Identification of “predictors of good responder” to CD34+ cell therapy from baseline characteristics or presentations (Table 3, Fig. 1 and 2)
To understand which baseline variable could be predictive of good responder prior to CD34+ cell therapy in patients with EnD-CAD, logistic regression analysis was performed. In univariate analysis, male gender, higher baseline leukocyte count, CCS angina score ≥3 and grade of diastolic dysfunction ≥2 were identified as potentially poor responders to the cell therapy. On the contrary, former smoker and higher baseline angiogenesis score could be used to predict good response to the cell therapy. After multivariate adjustment for the above potential variables, “presence” of former smoker and “absence” of male gender, Canadian Cardiovascular Society (CCS) angina score ≥3, and grade of diastolic dysfunction ≥2 on the initial survey were identified as independent predictors of good responder after stem cell therapy.
H-L test shown in Fig. 1A demonstrated sensitivity 86.7%, specificity 70.6% and accuracy 78.1% after considering the above four predictors (p=0.777, which was higher than cutoff value of 0.5). To facilitate an efficient evaluation in clinical practice, nomogram in Fig. 2 was utilized to calculate estimated good response rate of CD34+ cell therapy for EnD-CAD. Summation of individual points available from gender, former smoker, CCS angina score and grade of diastolic dysfunction helps researchers to assess the probability of good responder when EnD-CAD patients are enrolled as candidates for cell therapy.
Identification of “early predictors of good responder” after receiving G-CSF injection or CD34+ cell therapy (Table 4, Fig. 1 and Supplemental Table 1)
Owing to a lot of useful information available after administering G-CSF or transfusing CD34+ cells, those variables viable to be recognized as “early predictors” of good responder, e.g., vasculogenic activity of stem cells, biomarkers before and after G-CSF injection and detailed imaging for angiogenesis (refer to Supplemental Table 1), were collected and added to the logistic regression analysis. Table 4 shows in addition to the four aforementioned baseline characteristics, elevation of post G-CSF-treated neutrophil count was found negatively associated with good response to the cell therapy after multivariate adjustment. After adding this newly-identified independent predictor to the H-L test, a better predictor power was obtained with sensitivity 83.3%, specificity 85.3% and accuracy 84.4% (p=0.881, which was much higher than cutoff value of 0.5) (Fig. 1B). Regrettably, we did not further identify other potential predictors of good responder from remaining variables after G-CSF, prior to, or after stem cell therapy.
As shown in Supplemental Table 1, the responders had significantly higher soluble angiogenesis levels of vascular endothelial growth factor (VEGF) and hepatocytes growth factor (HGF) in circulation as compared with the non-responders. On the other hand, the circulatory level of stromal cell-derived factor (SDF)-1α, another angiogenic/proinflammatory factor, was consistently lower in the responders than in the non-responders prior to and after G-CSF injections as well as after CD34+ cell therapy. Additionally, SDF-1α levels in coronary sinus checked at different time points were also significantly lower in the former group than in the latter one. However, coronary angiogenesis on Wimasis analysis, angiogenic capacity on Matrigel assay, and levels of angiopoietin (ANP)-1, epidermal growth factor (EGF) and transforming growth factor (TGF)-β1 (i.e., three soluble angiogenesis factors) on enzyme-linked immunosorbent assay (ELISA) did not differ between the two groups, suggesting prediction of good responder to cell therapy was mainly dependent on baseline characteristics rather than on those variables collected after G-CSF or CD34+ cell therapy. These findings were very useful and practical on the screening for potential good responders to the cell therapy in the early stage of trial.