Dyballa Sylvia, Miñana Rafael, Rubio-Brotons Maria, Cornet Carles, Pederzani Tiziana, Escaramis Georgia, Garcia-Serna Ricard, Mestres Jordi, Terriente Javier
ZeClinics SL, IGTP (German Trias and Pujol Institute), Badalona 08916, Spain.
CIBER Epidemiology and Public Health.
Toxicol Sci. 2019 Oct 1;171(2):283-295. doi: 10.1093/toxsci/kfz165.
Cardiovascular drug toxicity is responsible for 17% of drug withdrawals in clinical phases, half of post-marketed drug withdrawals and remains an important adverse effect of several marketed drugs. Early assessment of drug-induced cardiovascular toxicity is mandatory and typically done in cellular systems and mammals. Current in vitro screening methods allow high-throughput but are biologically reductionist. The use of mammal models, which allow a better translatability for predicting clinical outputs, is low-throughput, highly expensive, and ethically controversial. Given the analogies between the human and the zebrafish cardiovascular systems, we propose the use of zebrafish larvae during early drug discovery phases as a balanced model between biological translatability and screening throughput for addressing potential liabilities. To this end, we have developed a high-throughput screening platform that enables fully automatized in vivo image acquisition and analysis to extract a plethora of relevant cardiovascular parameters: heart rate, arrhythmia, AV blockage, ejection fraction, and blood flow, among others. We have used this platform to address the predictive power of zebrafish larvae for detecting potential cardiovascular liabilities in humans. We tested a chemical library of 92 compounds with known clinical cardiotoxicity profiles. The cross-comparison with clinical data and data acquired from human induced pluripotent stem cell cardiomyocytes calcium imaging showed that zebrafish larvae allow a more reliable prediction of cardiotoxicity than cellular systems. Interestingly, our analysis with zebrafish yields similar predictive performance as previous validation meta-studies performed with dogs, the standard regulatory preclinical model for predicting cardiotoxic liabilities prior to clinical phases.
心血管药物毒性导致了临床阶段17%的药物撤市、上市后药物撤市的一半，并且仍然是几种已上市药物的重要不良反应。药物诱导的心血管毒性的早期评估是必不可少的，通常在细胞系统和哺乳动物中进行。当前的体外筛选方法具有高通量，但在生物学上较为简化。哺乳动物模型虽然能更好地预测临床结果，但其通量低、成本高且存在伦理争议。鉴于人类和斑马鱼心血管系统之间的相似性，我们建议在药物早期发现阶段使用斑马鱼幼体，作为生物学可翻译性和筛选通量之间的平衡模型，以解决潜在的问题。为此，我们开发了一个高通量筛选平台，该平台能够实现体内图像的全自动采集和分析，以提取大量相关的心血管参数：心率、心律失常、房室传导阻滞、射血分数和血流等。我们使用这个平台来评估斑马鱼幼体检测人类潜在心血管问题的预测能力。我们测试了一个包含92种具有已知临床心脏毒性特征的化合物的化学文库。与临床数据以及从人类诱导多能干细胞心肌细胞钙成像获得的数据进行交叉比较后发现，与细胞系统相比，斑马鱼幼体对心脏毒性的预测更可靠。有趣的是，我们对斑马鱼的分析得出的预测性能与之前用狗进行的验证性荟萃研究相似，狗是临床前预测心脏毒性的标准监管模型。
