Artificial intelligence guided prediction of cardiac repair

Cardiac Function Improvement and Bone Marrow Response Outcome Analysis of the Randomized Perfect Phase III Clinical Trial of Intramyocardial CD133+ Application After Myocardial Infarction

Gustav Steinhoff, Julia Nesteruk, Markus Wolfien, Guenther Kundt, Jochen Boergermann, Robert David, Jens Garbade, Jana grosse, Axel Haverich, Holger Hennig, Alexander Kaminski, Joachim Lotz, Friedrich W Mohr, Paula Mueller, Robert Oostendorp, Ulrike Ruch, Samir Sarikouch, Anna Skorska, Christof Stamm, Gudrun Tiedemann, Florian M Wagner, Olaf Wolkenhauer


The phase III clinical trial PERFECT was designed to assess clinical safety and efficacy of intramyocardial CD133+ bone marrow stem cell treatment combined with CABG for induction of cardiac repair.


Multicentre, double-blinded, randomised placebo controlled trial.


The study was conducted across six centres in Germany October 2009 through March 2016 and stopped due slow recruitment after positive interim analysis in March 2015.


Post-infarction patients with chronic ischemia and reduced LVEF (25–50%). Interventions: Eighty-two patients were randomised to two groups receiving intramyocardial application of 5 mL placebo or a suspension of 0.5–5 × 106 CD133+.


Primary endpoint was delta (∆) LVEF at 180 days (d) compared to baseline measured in MRI.

Findings (prespecified)

Safety (n = 77): 180 d survival was 100%, MACE n = 2, SAE n = 49, without difference between placebo and CD133+. Efficacy (n = 58): The LVEF improved from baseline LVEF 33.5% by +9.6% at 180 d, p = 0.001 (n = 58). Treatment groups were not different in ∆LVEF (ANCOVA: Placebo +8.8% vs. CD133+ +10.4%, ∆CD133+vs placebo +2.6%, p = 0.4).

Findings (post hoc)

Responders (R) classified by ∆LVEF ≥ 5% after 180 d were 60% of the patients (35/58) in both treatment groups. ∆LVEF in ANCOVA was +17.1% in (R) vs. non-responders (NR) (∆LVEF 0%, n = 23). NR were characterized by a preoperative response signature in peripheral blood with reduced CD133+ EPC (RvsNR: p = 0.005) and thrombocytes (p = 0.004) in contrast to increased Erythropoeitin (p = 0.02), and SH2B3 mRNA expression (p = 0.073). Actuarial computed mean survival time was 76.9 ± 3.32 months (R) vs. +72.3 ± 5.0 months (NR), HR 0.3 [Cl 0.07–1.2]; p = 0.067.Using a machine learning 20 biomarker response parameters were identified allowing preoperative discrimination with an accuracy of 80% (R) and 84% (NR) after 10-fold cross-validation.


The PERFECT trial analysis demonstrates that the regulation of induced cardiac repair is linked to the circulating pool of CD133+ EPC and thrombocytes, associated with SH2B3 gene expression. Based on these findings, responders to cardiac functional improvement may be identified by a peripheral blood biomarker signature.