Early Detection and Serial Monitoring of Anthracycline-Induced Cardiotoxicity Using T1-mapping Cardiac Magnetic Resonance Imaging: An Animal Study

Abstract

A reliable, non-invasive diagnostic method is needed for early detection and serial monitoring of cardiotoxicity, a well-known side effect of chemotherapy. This study aimed to assess the feasibility of T1-mapping cardiac magnetic resonance imaging (CMR) for evaluating subclinical myocardial changes in a doxorubicin-induced cardiotoxicity rabbit model. Adult male New Zealand White rabbits were injected twice-weekly with doxorubicin and subjected to CMR on a clinical 3T MR system before and every 2-4 weeks post-drug administration. Native T1 and extracellular volume (ECV) values were measured at six mid-left ventricle (LV) and specific LV lesions. Histological assessments evaluated myocardial injury and fibrosis. Three pre-model and 11 post-model animals were included. Myocardial injury was observed from 3 weeks. Mean LV myocardium ECV values increased significantly from week 3 before LV ejection fraction decreases (week 6), and ECVs of the RV upper/lower insertion sites and papillary muscle exceeded those of the LV. The mean native T1 value in the mid-LV increased significantly increased from week 6, and LV myocardium ECV correlated strongly with the degree of fibrosis (r = 0.979, p < 0.001). Myocardial T1 mapping, particularly ECV values, reliably and non-invasively detected early cardiotoxicity, allowing serial monitoring of chemotherapy-induced cardiotoxicity.

Figure 1

Experimental timeline and numbers of animals evaluated at each time point. All rabbits underwent cardiac magnetic resonance imaging (CMR) before and every 2–4 weeks after drug administration. Circles indicate scanning CMR; black squares indicate sacrifice after CMR scanning. Crosses indicate unexpected deaths.

Figure 2

Serial changes in the extracellular volume (ECV) and native T1 values. (a) Serial changes in the ECVs of the left ventricle (LV) myocardium and other specific LV lesions according to modelling time. The mean ECV of the mid-LV increased significantly beginning at 3 weeks (p = 0.002). Compared with the LV myocardium, the ECVs of the right ventricle (RV) upper and lower insertion sites and the papillary muscle were higher. (b) Serial changes in the native T1 values of the LV myocardium according to modelling time. Mean native T1 values at the mid-LV increased significantly beginning at week 6 (p < 0.001).

Figure 3

Histological findings of myocardial injury. (a) Normal myocardium from a control subject. (b) Severe myocardial injury with intracytoplasmic vacuolization (arrowheads) and myofibril loss (arrows) are visible in the left ventricular (LV) septal wall of a 3-week model (myocardial injury score = 3). (c) Diffuse interstitial oedema is visible in the LV lateral wall of a 3-week model. Haematoxylin and eosin staining; magnification, ×400 (d) Myocardial injury score according to the modelling time.

Figure 4

Histological images of extracellular collagen deposition in the left ventricle (LV) myocardium. (a) Minimal collagen fibres are visible in the interstitium of a baseline subject (measured collagen volume fraction [CVF] = 3.04%). (b) No significant increase in collagen fibres was observed in a 3-week model (CVF = 4.28%). (c) Abundant collagen deposition is visible in a 16-week model (CVF = 25.00%). Picrosirius red staining; original magnification, ×200 (d) Myocardial fibrosis according to the modelling time.

Figure 5

Receiver operating characteristic (ROC) curves for diagnosis. (a) Areas under the ROC curves (AUCs) for the extracellular volume (ECV), native T1, and left ventricle ejection fraction (LVEF) in post-modelling subjects were 0.974 (95% confidence interval, 0.955–0.989), 0.893 (0.848–0.930), and 0.916 (0.874–0.951), respectively. (b) In 12- and 16-week modelling subjects, the AUCs for the ECV, native T1, and LVEF were 0.981 (0.969–0.990), 0.882 (0.855–0.910), and 0.857 (0.801–0.906), respectively.

Figure 6

Measurement of T1 values in the myocardium and specific lesions of the left ventricle (LV) at the mid-ventricle on the short-axis plane. Endocardial (red line) and epicardial borders (green line) of the LV wall were delineated semi-automatically; six segments were delineated automatically (white lines). A round <5 mm² region of interest (ROI; orange circle) was drawn in the LV cavity on pre-contrast (a) and post-contrast T1 mapping images (b). To measure T1 values of specific LV lesions in the short-axis plane, ROIs covering specific lesions (blue circle 1: right ventricle (RV) upper insertion site, 2: RV lower insertion site, 3: papillary muscle) were drawn on pre-contrast (c) and post-contrast T1-mapping images (d).