PROJECT SUMMARY
Tyrosine kinase inhibitors (TKIs) are specified small-molecule inhibitors of the activity of tyrosine kinases and
have shown therapeutic impact on treating cancers. Despite being a prevalent class of drugs in the
pharmaceutical industry, their development and clinical application are frequently hindered by a wide range of
cardiovascular complications: QT prolongation/arrhythmia, left ventricular dysfunction, congestive heart failure,
ischemia, myocardial infarction, and hypertension. Cardiotoxicity's repercussions impede the advancement of
novel TKIs for cancer therapy and furthermore cause failures in preclinical drug discovery and clinical
development. Therefore, there is an urgent need to assess cardiotoxicity induced by TKIs used in cancer
patients' treatment. However, there is still no targeted and effective preclinical trial due to a) only focusing on
anti-tumor effects without systematic examination of coexisting cardiovascular effects, b) limited efficiency and
accuracy of cardiotoxicity prediction, and c) inter-species differences between animals and humans on
cardiotoxic responses. To address this unmet need, a new in vitro human model for comprehensively assessing
TKIs-indued cardiotoxicity is proposed here. Recently, human pluripotent stem cell (hPSC)-derived vascularized
cardiac organoids (VCOs) developed in PI’s and other labs have shown great promise in emulating the human
heart in both cardiovascular structure and function, which makes it an ideal in vitro drug toxicity evaluation system
targeting cardiovascular cells by TKIs. However, three significant challenges remain in applying hPSC-derived
VCOs to cardiotoxicity evaluation: 1) insufficient and uncoupled assessment of organoid structural and functional
properties; 2) continuous generation of large datasets from multifaceted characterizations that requires more
integrated analysis without human bias; 3) low efficiency with more individual hPSC lines. It is hypothesized that
artificial intelligence (AI)-driven biomedical data featuring and predicting address the challenges for accurately
assessing TKI-induced cardiotoxicity by phenotyping the cardiovascular structure and function of VCOs. The
central goal is to establish an organoid-AI system to assess TKI-induced cardiotoxicity efficiently and accurately
with three specific aims: Aim 1. Predict TKI-induced toxicity on cardiovascular structure in VCOs by Generative
AI algorithm; Aim 2. Phenotype TKI-induced cardiovascular dysfunction in VCOs; Aim 3. Establish a TKI-induced
cardiotoxicity assessment system with an organoid-AI system. Upon completion of the proposed project, a
comprehensive human organoid-AI system for assessing TKI-induced cardiotoxicity will be established to
increase the efficiency and accuracy of preclinical drug safety evaluation of TKIs on human cardiotoxicity on both
cardiovascular structure and function.