HiPSC-cardiomyocyte monolayers and spheroids show different cardiotoxic responses to anti-cancer treatment

Background

Cardiotoxic effects of anti-cancer therapies (ACT) is an increasing concern, particularly in the context of dramatic improvements in cancer survival. Preclinical in vitro models are essential in early phase cardiotoxicity screening and can provide early mechanistic insights relevant to patients(1). Use of human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) allows assessment of ACTs in a high throughput system with fewer ethical concerns than in vivo testing, an intact human genome and promise of patient specific testing. Innovations such as 3D hiPSC-CM spheroids are intended to more closely recapitulate in vivo conditions than recognised 2D models by increasing maturation, paracrine signalling and mechanistic translation into clinic(2). However, few comparisons of cardiotoxic responses have established pros and cons of monolayers (2D) vs spheroids (3D) to ACT treatment.

Aims

We compare cardiotoxic responses between hiPSC-CM 2D and 3D models during exposure to clinically relevant doses of doxorubicin (Dox) to inform the role of 2D/3D in in vitro cardiotoxicity screens.

Methods

Commercial hiPSC-CMs from a healthy male donor were used. Cells were plated in 2D (50k cells/well; 96-well plate) or 3D (5k cells/well; U-bottom ultra-low attachment plates) and Dox (50,100,300,600,1000nM) or 0.1% DMSO vehicle treated for 6-days with media changes every 48hrs. Baseline electrophysiology (fluorescence) and contraction were recorded at baseline and after 24,48,72 and 144hrs Dox treatment. Lactate dehydrogenase (LDH) release in media was measured daily to estimate cytotoxicity.

Results

Dox caused dose-dependent shortening of action potential duration (APD) in 2D (48hr %ΔΔAPD50; 300nM Dox -33.9±2.4 p<0.05, Fig 1). In the 3D equivalent Dox treatment caused an initial phase of APD prolongation before a decrease in APD50 over chronic treatment (72hr %ΔΔAPD50; 300nM Dox 32.1±1.8 p<0.05, 144hr ΔAPD50; -5.6±5.5 p<0.05). Contractile dysfunction was significant >300nM Dox in 2D and 3D whereas cytotoxicity was evident ≥1000nM Dox. We developed a model (Fig 2) from three empirical functions: APD decrease, APD increase and an excitability limit defining CM non-functionality to estimate QT-interval change and myocardial non-excitability. Applying the model to a Dox patient plasma concentration profile (75mg/m2 dose) predicts no significant changes to QT, and a 5% loss of functional myocardium (3).

DISCUSSION: Structural differences may confound divergent APD responses to Dox in 2D vs 3D. Recently glycolytic activation (GA) was shown to be a druggable target for Dox cardiotoxicity(4). Our data suggests APD increase in 3D may be induced by extracellular acidosis from lactate efflux into limited extracellular space, which is not replicated in 2D as the extracellular space is significantly larger(5). Our data suggest 3D better informs complex metabolic-structural-functional cardiotoxic axes, potentiating better mechanistic translation into clinic.Figure 1  Figure 2