Testing the Translational Power of the Zebrafish: An Interspecies Analysis of Responses to Cardiovascular Drugs

Abstract

The zebrafish is rapidly emerging as a promising alternative in vivo model for the detection of drug-induced cardiovascular effects. Despite its increasing popularity, the ability of this model to inform the drug development process is often limited by the uncertainties around the quantitative relevance of zebrafish responses compared with nonclinical mammalian species and ultimately humans. In this test of concept study, we provide a comparative quantitative analysis of the in vivo cardiovascular responses of zebrafish, rat, dog, and human to three model compounds (propranolol, losartan, and captopril), which act as modulators of two key systems (beta-adrenergic and renin-angiotensin systems) involved in the regulation of cardiovascular functions. We used in vivo imaging techniques to generate novel experimental data of drug-mediated cardiovascular effects in zebrafish larvae. These data were combined with a database of interspecies mammalian responses (i.e., heart rate, blood flow, vessel diameter, and stroke volume) extracted from the literature to perform a meta-analysis of effect size and direction across multiple species. In spite of the high heterogeneity of study design parameters, our analysis highlighted that zebrafish and human responses were largely comparable in >80% of drug/endpoint combinations. However, it also revealed a high intraspecies variability, which, in some cases, prevented a conclusive interpretation of the drug-induced effect. Despite the shortcomings of our study, the meta-analysis approach, combined with a suitable data visualization strategy, enabled us to observe patterns of response that would likely remain undetected with more traditional methods of qualitative comparative analysis. We propose that expanding this approach to larger datasets encompassing multiple drugs and modes of action would enable a rigorous and systematic assessment of the applicability domain of the zebrafish from both a mechanistic and phenotypic standpoint. This will increase the confidence in its application for the early detection of adverse drug reactions in any major organ system.

Keywords: beta-adrenergic receptor; cardiovascular effects; comparative pharmacology; drug safety; meta-analysis; preclinical species; renin–angiotensin system; zebrafish.

Figure 1

Summary of the methodological approach employed in the study. The experimental quantification of zebrafish (7 dpf) cardiovascular responses to propranolol, losartan, and captopril (left) was combined with the subsequent meta-analysis of mammalian preclinical and human data extracted from the literature (right) to generate quantitative understanding of interspecies similarities in effect size and direction.

Figure 2

Integrated representation of the effects induced by propranolol (125 µM), losartan (5 mM), and captopril (50 mM) on six cardiovascular endpoints measured in zebrafish larvae (7 dpf), after 1 and 48 h exposure. Each graph represents the effect size observed for each endpoint in treated fish (red) versus control fish (blue), expressed as ratio between mean treated value and mean control value. For example, a treated value of 1.1 indicates a 10% increase versus the control value.

Figure 3

Dose–response of five cardiovascular parameters measured in zebrafish larvae (7 dpf) following exposure to propranolol, losartan, and captopril after two different exposure times (1 and 48 h). Data are presented as mean ± SEM (n = 4–6). Statistically significant differences from the control group are displayed as *(p < 0.05).

Figure 4

Overview of the effects of propranolol, losartan, and captopril on the heart rate of human, dog, rat, and zebrafish. Data are expressed as standardized mean difference (treated vs. control) ± 95% confidence interval. The data related to human, dog, and rat were retrieved from the literature, whereas the zebrafish data were generated in the present study. Each data point represents a different treatment group. The same dataset was used to perform a quantitative meta-analysis. A detailed description of the results is provided in the Supplementary Material Data Sheet 1 .

Figure 5

Overview of the effects of propranolol, losartan, and captopril on the blood flow of human, dog, rat, and zebrafish. Data are expressed as the standardized mean difference (treated vs. control) ± 95% confidence interval. The data from human, dog, and rat were retrieved from the literature, whereas the zebrafish data were generated in the present study. Each data point represents a different treatment group. The same dataset was used to perform a quantitative meta-analysis. A detailed description of the results is provided in the Supplementary Material Data Sheet 1 .

Figure 6

Meta-analysis of the effects of propranolol on blood flow in zebrafish, rat, dog, and humans. Effect size reported as standardized mean difference ± 95% confidence interval.

Figure 7

Meta-analysis of the effects of losartan on blood flow in zebrafish, rat, dog, and humans. Effect size reported as standardized mean difference ± 95% confidence interval.

Figure 8

Meta-analysis of the effects of captopril on blood flow in zebrafish, rat, dog, and humans. Effect size reported as standardized mean difference ± 95% confidence interval.

Figure 9

Overview of the effects of propranolol, losartan, and captopril on the blood vessel diameter of human, dog, rat, and zebrafish. Data are expressed as the standardized mean difference (treated vs. control) ± 95% confidence interval. The data related to human, dog, and rats were retrieved from the literature, whereas the zebrafish data were generated in the present study. Each data point represents a different treatment group. The same dataset was used to perform a quantitative meta-analysis. A detailed description of the results is provided in the Supplementary Material Data Sheet 1 .

Figure 10

Overview of the effects of propranolol, losartan, and captopril on the stroke volume of human, dog, rat, and zebrafish. Data are expressed as the standardized mean difference (treated vs. control) ± 95% confidence interval. The data related to human, dog, and rat were retrieved from the literature, whereas the zebrafish data were generated in the present study. Each data point represents a different treatment group. The same dataset was used to perform a quantitative meta-analysis. A detailed description of the results is provided in the Supplementary Material Data Sheet 1 .