Comparative analysis of the cardiotoxicity induced by doxorubicin and epirubicin and exploratory investigations of clinical parameters indicating the cardiotoxicity in healthy dogs

Abstract

Cardiotoxicity of doxorubicin (DOX) and epirubicin (EPI) has not been thoroughly compared in dogs. Fractional shortening (FS) is considered insufficient for detecting early myocardial damage induced by these drugs. This study aimed to compare the cardiotoxicity between EPI and DOX and to investigate parameters associated with histopathological findings induced by these drugs. Eight healthy beagles were randomly divided into two groups and administered either DOX or EPI until development of anthracycline-induced cardiac dysfunction (AICD), defined as consecutive FS below 30%, or until reaching a cumulative dose of 300 mg/m². After the final evaluation, the dogs were euthanized, and cardiac tissues were histopathologically evaluated. One dog in the EPI group developed AICD after six administrations (cumulative dose of 168 mg/m²), whereas no other dogs did. The number of vacuolated myocardial cells was significantly higher in the DOX group than in the EPI group. In the DOX group, but not in the EPI group, longitudinal strain (A4C-LS) and tissue motion mitral annulus displacement (A4C-TMAD) from the left apical four-chamber view, global longitudinal strain, myocardial performance index and velocity-time integral from left ventricle, mitral annular plane systolic excursion, and high-sensitivity cardiac troponin I showed significant changes after treatment. Furthermore, A4C-LS and A4C-TMAD were correlated with the extent of myocardial collagen fibers. These results suggest that EPI induces less severe myocardial injury than DOX, although individual variability in drug sensitivity exists. Echocardiographic parameters such as A4C-LS and A4C-TMAD may serve as sensitive markers for early myocardial injury.

INTRODUCTION

Doxorubicin (DOX) is one of the anthracycline antibiotics, which are a class of chemotherapeutic agents used in the treatment of hematopoietic and solid tumors in humans and dogs. However, anthracyclines have a characteristic side effect, anthracycline-induced cardiotoxicity (AIC), which can worsen the prognosis of cancer patients despite their effectiveness. Since AIC occurs in a dose-dependent and cumulative manner, it is recommended that the lifetime cumulative dose of DOX should be limited to 180–240 mg/m² (6–8 administrations) in dogs [23, 26].

Epirubicin (EPI) is a synthetic stereoisomer of DOX with comparable antitumor effectiveness, but the cardiotoxicity is milder than that of DOX in humans [34]. The safe lifetime cumulative dose of this drug is approximately double compared to that of DOX (1,000 vs. 550 mg/m²) in humans [34]. In dogs, EPI also presented less myocardial damage compared to DOX in a toxicological study [2], although the administration regimen used in this study was not a clinically used protocol with a shorter administration interval. At present, EPI is used in the same regimen as DOX in canine patients: it is administered once every three weeks at 30 mg/m² (or 1 mg/kg for dogs weighing less than 10 kg) until the lifetime cumulative dose reaches the limit similar to that of DOX, 180–240 mg/m² [21]. Although it might be possible that EPI could be safely administered within a larger lifetime cumulative dose in dogs, the cardiotoxicity and the safe lifetime cumulative dose of EPI have not been determined at present.

The pathological features of AIC are similar to those of dilated cardiomyopathy and include myocardial damage, cardiac dysfunction, and arrhythmia, ultimately leading to heart failure. Particularly, anthracycline-induced cardiac dysfunction (AICD) is of clinical concern and is detected by insufficient left ventricular (LV) systolic function evaluated by ejection fraction (EF) in human medicine and by fractional shortening (FS) in dogs. However, it is well known that myocardial damage identified by endomyocardial biopsy has already advanced by the time a decline in these parameters is detected [12, 15, 32]. Therefore, several early detection methods, such as global longitudinal strain measured using speckle-tracking echocardiography (STE) and high-sensitivity cardiac troponin I (hs-cTnI), are employed in addition to EF in human medicine [24]. In contrast, in veterinary medicine, the relationship between pathological findings and non-invasive examinations has not been thoroughly investigated, and no clinical assessment methods for AIC other than FS have been established in dogs.

The purposes of this study were as follows: 1) to compare the cardiotoxicity between EPI and DOX in healthy beagles primarily based on myocardial histopathological findings, and 2) to explore sensitive indices for myocardial damages by examining the associations of histopathological findings with echocardiographic parameters, including those clinically applied in human medicine and by assessing blood concentrations of hs-cTnI longitudinally.

MATERIALS AND METHODS

Animals

Eight beagles weighing between 9.1 to 11.5 kg (10.5 ± 0.8 kg, mean ± SD), aged 6.4 to 9.2 years (7.7 ± 1.1 years), including three males and five females, were enrolled in this study. All dogs were confirmed to be clinically healthy prior to enrollment, based on the results of physical examination, complete blood count (CBC), serum biochemistry, systemic blood pressure measurement using the oscillometric method, electrocardiography (ECG), thoracic radiography, and echocardiography. The dogs were housed individually in cages and fed a commercial dry food diet with free access to water. The Ethical Committee of The University of Tokyo approved all animal experimental procedures used in this study (Approval Number P22-107).

Experimental design

An outline of the present study is shown in Fig. 1. Eight dogs were randomly assigned to either the DOX or EPI group. Blood collection, ECG, and echocardiography were performed under sedation using butorphanol tartrate (0.2 mg/kg, iv). DOX or EPI was administered immediately after evaluation at a dose of 30 mg/m² via constant-rate infusion over 30 min. The administered drugs were blinded to the veterinarian who conducted the cardiologic evaluation (S.S.) until the data analysis was completed. This procedure was repeated every three weeks until AICD developed, defined as a consecutive reduction in FS to <30% based on the lower limit of the reference interval in dogs [5,6,7], or until the cumulative dose reached 300 mg/m². The general physical conditions of the dogs were checked every day, and CBC and serum biochemistry were performed one week after each administration. The dose was reduced by 10% if dogs showed grade 4 or more severe blood/bone marrow or grade 3 or more severe non-blood/bone marrow adverse events, as defined by VCOG-CTCAE v2.0 [22]. The final evaluation was conducted three weeks after the final administration. Necropsy was performed within approximately one week after the final evaluation, and the cardiac tissue was collected.

Fig. 1.

The outline of the present study. Blood collection and echocardiography were performed as cardiovascular evaluations at the time points indicated by circles. Doxorubicin (DOX) or epirubicin (EPI) was administered immediately after these evaluations at 30 mg/m². This procedure was repeated every three weeks until anthracycline-induced cardiac dysfunction (AICD), defined as a consecutive reduction in fractional shortening (FS) to <30%, developed or until the cumulative dose reached 300 mg/m². The final evaluation was conducted three weeks after the final administration. Necropsy was performed within approximately one week after the final evaluation, and the cardiac tissues were collected.

Echocardiography

Echocardiography was performed by a single veterinarian (S.S.) using commercially available ultrasound equipment (ARIETTA 750; FUJIFILM Healthcare, Tokyo, Japan) with a 2–9 MHz sector transducer. The dogs were manually restrained under sedation in both left and right lateral recumbency, with simultaneous ECG monitoring (lead II).

Details of the methods for measuring each echocardiographic parameter are provided in the Supplementary Materials and Methods. In the present study, we measured FS, EF from the left apical four-chamber view (A4C-EF), biplane LV ejection fraction (BP-EF), longitudinal strain from the left apical four-chamber view (A4C-LS), global longitudinal strain (GLS), tissue motion mitral annulus displacement from the left apical four-chamber view (A4C-TMAD), global tissue motion mitral annulus displacement (GTMAD), and mitral annular plane systolic excursion (MAPSE) as LV systolic functional parameters. LV myocardial performance index (MPI) and LV outflow tract velocity-time integral (VTI) were measured as overall LV functional parameters. Early diastolic transmitral flow velocity (E), early to late diastolic transmitral flow velocity ratio (E/A), and early diastolic mitral annular motion velocity (e’) were measured as diastolic functional parameters. End-diastolic LV internal diameter (LVIDd), end-diastolic interventricular septum (IVSd), end-diastolic LV posterior wall thickness (LVPWd) and left atrium to aortic root ratio (LA/Ao) were measured as left heart size parameters. Tricuspid annular plane systolic excursion (TAPSE) and right ventricle fractional area change (RVFAC) were measured as right heart functional parameters. Each parameter was calculated as the average of three consecutive cardiac cycles.

As an investigation of measurement variability in novel STE parameters, intraobserver variabilities of A4C-LS and A4C-TMAD at baseline and after completion of repeated drug administrations were evaluated using videos of two dogs each from the DOX and EPI groups. For each dog, measurements were obtained from three individual videos recorded on the same day. Mean values and standard deviations (SD) were calculated from three measurements, and the coefficient of variation (CV) was calculated as follows: CV (%)=(SD/mean) × 100. A CV of less than 20% was considered acceptable variability.

Histopathological examination

The heart, lung, liver, kidney, spleen, gastrointestinal tract, endocrine organs, and brain were collected at necropsy and fixed in 10% neutral buffered formalin for histopathological evaluation. Regarding cardiac tissues, samples were collected from the ventricles, right atrium (RA), sinoatrial node (SA), and atrioventricular node (AV) for histopathological examination. The ventricles were further divided into four regions: the left ventricular free wall (LVFW), apex (AP), septal wall (SW), and right ventricular free wall (RVFW). Tissues from the LVFW, SW, and RVFW were collected from just below the base of each region, while tissues from the AP were collected from the apical area of the LV. Tissues were embedded in paraffin wax, sectioned at a thickness of 4 μm, and stained with hematoxylin and eosin (HE). In addition, serial sections were subjected to Masson’s trichrome staining and immunohistochemistry (IHC).

Vacuolation and eosinophilic degeneration of cardiomyocytes were counted in five microscopic fields (×200) in the most severely affected region on each HE-stained slide of the LVFW, SW, AP, RVFW, and RA. Evaluation was performed based on the International Harmonization of Nomenclature and Diagnostic Criteria [43]. IHC detected cleaved caspase-3, and positive cells were counted in five microscopic fields (×200), with the highest count recorded. Fibrosis was expressed as the percentage of collagen fiber area within the total cardiac tissue using images of Masson’s trichrome-stained slides. Evaluation was performed by board-certified veterinary pathologists (JKC, KU), and the administered drugs were blinded to the pathologists.

Details of the methods for histopathological examination are provided in Supplementary Materials and Methods.

ECG recording

ECG was performed by holding the animal in the right lateral recumbency using an ECG monitor (Cardisuny D700, Fukuda M-E Kogyo Co., Ltd., Tokyo, Japan). All ECG parameters were recorded using limb leads under measurement settings of 50 mm/sec and 5 to 10 mm/mV. Heart rate was obtained, and arrhythmias were diagnosed based on the evaluation of heart rate, cardiac rhythm, P waves, PR intervals, QRS complexes, and ST segment abnormalities. Normal ST amplitude was defined as within 0.2 mV [42].

Measurement of hs-cTnI

Serum was separated from the collected blood samples and stored at −80°C until analysis. Serum hs-cTnI concentrations were determined using a three-site sandwich immunoassay (FUJIFILM VET Systems, Tokyo, Japan). The detection limit and upper reference limit in serum were set at 0.001 and 0.129 ng/mL, respectively, according to the commercial laboratory.

Data analysis

Student’s t-test was used to analyze differences in baseline characteristics and the total cumulative dose between the DOX and EPI groups. This test was also used to compare the histopathological findings including vacuolated cell count, eosinophilic degeneration cell count, and the collagen fiber area ratios in LVFW, AP, SW, and RVFW between the two groups. Paired t-tests were used to compare heart rate, hs-cTnI levels, and echocardiographic parameters between baseline and after completion of repeated administration of DOX or EPI within each group. The cumulative dose at the onset of ST-segment depression, which was the most frequent ECG abnormality in this study, was compared between the two groups using the Cox proportional hazards model. The total administered dose (mg/m²) was used as the time variable, and the occurrence of ST-segment depression (1 =occurred, 0 =not occurred at the cumulative dose reached 300 mg/m²) was set as the event variable. The treatment group (DOX or EPI) was included as an independent variable in the model.

Then, the datasets of all eight dogs were combined to investigate the associations of histopathological findings with the results of echocardiographic examinations. The correlations of ventricular histopathological findings in LVFW, AP, and SW with echocardiographic parameters were analyzed using Pearson’s correlation coefficient. Echocardiographic parameters that showed significant changes between baseline and after completion of repeated DOX or EPI administration were selected for correlation analysis. In addition, the correlations between these histopathological findings and the cumulative dose at which ST-segment depression was first detected were compared using Pearson’s correlation coefficient. Adjusted P-values to account for multiple comparisons were calculated using the Benjamini-Hochberg method. Statistical significance was set at P<0.05.

All statistical analyses were performed using R software (version 4.2.2, R Foundation for Statistical Computing, Vienna, Austria) and EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan).

RESULTS

Comparison of the cardiotoxicity between EPI and DOX

Baseline characteristics of dogs and administration of anthracyclines: The DOX group consisted of one neutered male and three intact females, while the EPI group included two neutered males and two intact females. The mean ± SD of age, body weight, and heart rate at baseline were 7.4 ± 1.1 years, 10.0 ± 0.8 kg, and 79.3 ± 15.7 bpm in the DOX group, whereas 8.1 ± 1.2 years, 10.9 ± 0.8 kg, and 91.3 ± 8.6 bpm in the EPI group. Table 1 shows heart rate, echocardiographic parameters, and the concentrations of hs-cTnI at baseline in each group. Regarding echocardiographic parameters, the mean ± SD of FS, LVIDd, and LA/Ao were 40.0 ± 3.6%, 31.8 ± 1.5, and 1.13 ± 0.11 in the DOX group, whereas they were 38.5 ± 4.0%, 31.5 ± 2.3, and 1.14 ± 0.12 in the EPI group. The mean ± SD of hs-cTnI was 0.014 ± 0.013 in the DOX group and 0.012 ± 0.010 in the EPI group. There was no significant difference in any of these parameters between the two groups at baseline.

Table 1. Comparisons of heart rate, echocardiographic parameters, and high-sensitivity cardiac troponin I between baseline and after the final drug administrations.

		DOX			EPI
		Baseline	Post	P-value	Baseline	Post	P-value
Heart rate		79.3 ± 15.7	88.0 ± 20.4	0.194	91.3 ± 8.6	78.3 ± 28.6	0.345
LV systolic function	FS (%)	40.0 ± 3.6	39.9 ± 1.5	0.936	38.5 ± 4.0	33.6 ± 6.7	0.118
	A4C-EF (%)	75.1 ± 4.4	70.5 ± 8.3	0.206	71.4 ± 9.2	62.3 ± 14.1	0.248
	BP-EF (%)	76.3 ± 3.7	70.2 ± 6.6	0.095	73.8 ± 6.2	65.0 ± 11.9	0.130
	A4C-LS (%)	19.5 ± 2.0	16.7 ± 1.8	0.023*	16.8 ± 1.4	17.0 ± 4.6	0.947
	GLS (%)	19.7 ± 2.1	16.3 ± 1.6	0.044*	16.8 ± 0.9	16.9 ± 2.3	0.914
	A4C-TMAD (%)	15.1 ± 1.6	11.8 ± 0.6	0.011*	14.1 ± 1.8	13.5 ± 2.2	0.698
	GTMAD (%)	15.4 ± 2.3	13.7 ± 1.4	0.144	14.9 ± 1.8	13.5 ± 1.5	0.282
	MAPSE (mm)	9.1 ± 1.0	7.8 ± 0.5	0.049*	9.2 ± 2.1	6.9 ± 1.3	0.064
Overall LV function	MPI	0.33 ± 0.07	0.50 ± 0.12	0.032*	0.32 ± 0.06	0.38 ± 0.09	0.102
	VTI (mm)	18.7 ± 2.3	14.3 ± 1.9	0.019*	18.6 ± 3.4	15.0 ± 1.4	0.054
LV diastolic function	E (m/sec)	0.87 ± 0.06	0.79 ± 0.12	0.117	0.82 ± 0.10	0.70 ± 0.18	0.262
	E/A	1.4 ± 0.3	1.3 ± 0.2	0.380	1.31 ± 0.17	1.28 ± 0.36	0.888
	e’ (cm/sec)	10.2 ± 0.9	8.8 ± 2.2	0.229	11.1 ± 1.7	10.4 ± 2.3	0.634
Left heart size	LVIDd (mm)	31.8 ± 1.5	32.0 ± 2.8	0.832	31.5 ± 2.3	32.9 ± 4.7	0.439
	IVSd (mm)	9.3 ± 1.5	9.4 ± 1.8	0.914	8.7 ± 0.4	8.2 ± 1.4	0.376
	LVPWd (mm)	8.7 ± 1.3	8.8 ± 2.0	0.868	8.7 ± 0.7	8.5 ± 0.6	0.679
	LA/Ao	1.13 ± 0.11	1.18 ± 0.03	0.382	1.14 ± 0.12	1.13 ± 0.07	0.841
RV systolic function	TAPSE (mm)	13.2 ± 3.2	13.9 ± 5.1	0.640	14.6 ± 1.6	13.6 ± 1.4	0.516
	RVFAC (%)	46.4 ± 3.1	49.9 ± 4.8	0.161	50.9 ± 4.2	45.9 ± 16.4	0.567
hs-cTnI (ng/mL)		0.014 ± 0.013	1.636 ± 0.405	0.004*	0.012 ± 0.010	0.663 ± 0.815	0.205

*P<0.05. DOX: doxorubicin, EPI: epirubicin, LV: left ventricle, RV: right ventricle, FS: fractional shortening, A4C-EF: ejection fraction from the left apical four-chamber view, BP-EF: biplane LV ejection fraction, A4C-LS: longitudinal strain from the left apical four-chamber view, GLS: global longitudinal strain, A4C-TMAD: tissue motion mitral annulus displacement from the left apical four-chamber view, GTMAD: global tissue motion mitral annulus displacement, MAPSE: mitral annular plane systolic excursion, MPI: LV myocardial performance index, VTI: LV outflow tract velocity-time integral, E: early diastolic transmitral flow velocity, E/A: early to late diastolic transmitral flow velocity ratio, e’: early diastolic mitral annular motion velocity, LVIDd: end-diastolic LV internal diameter, IVSd: end-diastolic interventricular septum, LVPWd: end-diastolic LV posterior wall thickness, LA/Ao: left atrium to aortic root ratio, TAPSE: tricuspid annular plane systolic excursion, RVFAC: right ventricle fractional area change, hs-cTnI: high-sensitivity cardiac troponin I.

Following anthracycline administration, Dog 6 in the EPI group was diagnosed with AICD, as it showed consecutive FS values below 30% after six administrations with a cumulative dose of 168 mg/m², resulting in the completion of drug administration (Fig. 2). The others did not present consecutive FS reduction, and they received doses at 300 mg/m² or higher. No dogs presented heart failure symptoms. In the DOX group, dose reduction was performed twice in one of the four dogs due to grade 4 neutropenia, but it was not required in the others. In the EPI group, dose reductions were required in all dogs: once in two dogs, twice in one dog, and three times in the other. All reductions were due to grade 4 neutropenia. Other adverse events included gastrointestinal signs, with grade 2 anorexia being the most severe.

Fig. 2.

The dose-dependent changes in fractional shortening (FS) in the doxorubicin (DOX) (A) and epirubicin (EPI) (B) groups. In Dog 6 of the EPI group, administration was terminated due to development of anthracycline-induced cardiac dysfunction (AICD) after six administrations with a cumulative dose of 168 mg/m². The remaining dogs received doses of 300 mg/m² or higher, and there were no consecutive FS values below 30% in these dogs.

The number of administrations was 12 in one dog and 10 in the others in the DOX group. In the EPI group, two dogs received 12 administrations, one dog received 11 administrations, and the other (Dog 6) received six administrations. The cumulative dose was 300.0 mg/m² in all dogs in the DOX group, while the mean ± SD of the cumulative dose was 276.3 ± 72.6 mg/m² in the EPI group. The cumulative dose was not significantly different between the groups.

Histopathological findings: Vacuolation and eosinophilic degeneration of cardiomyocytes were observed in the ventricular myocardium of dogs in both the DOX and EPI groups (Fig. 3). No evidence of necrosis or inflammation was observed. In Masson’s trichrome-stained slides, fibrosis was often observed along the degenerated cardiomyocytes (Supplementary Fig. 1).

Fig. 3.

Microscopic images of hematoxylin and eosin-stained left ventricular free wall (LVFW), apex (AP), and septal wall (SW) tissues of dogs administered doxorubicin (DOX) (Dog 3, A–C) or epirubicin (EPI) (Dog 8, D–F). Marked vacuolation and eosinophilic degeneration were observed in both the DOX and EPI groups. In addition, images from Dog 6, which developed anthracycline-induced cardiac dysfunction, are presented (G–I). The degree of myocardial change in Dog 6 did not appear to be markedly more severe than that in other dogs. Bar=50 μm.

The IHC for cleaved caspase-3 revealed that most cardiomyocytes were negative for immunostaining, with a maximum of only two positive cells per field of view at 20× objective (Supplementary Fig. 2).

No remarkable differences in myocardial changes were observed using any staining methods between Dog 6 and the other dogs without AICD (Fig. 3).

No obvious lesions suggestive of anthracycline-induced toxicity were observed in tissues other than the heart.

Comparisons of cardiac histopathological findings between DOX and EPI groups: At first, the vacuolated cell count, eosinophilic degeneration cell count, and the collagen fiber area ratio in the ventricle were compared between the DOX and EPI groups (Fig. 4). The vacuolated cell count was significantly higher in the DOX group (13.2 ± 8.3, mean ± SD) than in the EPI group (6.0 ± 5.7, mean ± SD) (P=0.01). Although not statistically significant, the eosinophilic degeneration cell count also tended to be higher in the DOX group (8.9 ± 5.8, mean ± SD) than in the EPI group (5.6 ± 3.4, mean ± SD) (P=0.06). The collagen fiber area ratio did not differ significantly between the two groups (0.9 ± 0.5 in the DOX group vs. 0.8 ± 0.5 in the EPI group, P=0.84).

Fig. 4.

Comparison of vacuolated cell count (A), eosinophilic degeneration cell count (B), and the collagen fiber area ratio (C) in the ventricular region between the doxorubicin (DOX) and epirubicin (EPI) groups. Each plot presents the mean values of five fields of view, with four ventricular areas plotted per dog: left ventricular free wall (LVFW), apex (AP), septal wall (SW), and right ventricular free wall (RVFW).

The vacuolated cell count, eosinophilic degeneration cell count, and the collagen fiber area ratio in the RA (Supplementary Fig. 3A, 3B, and 3C, respectively) were not significantly different between DOX and EPI groups. The collagen fiber area ratios in the SA and AV were also not significantly different between the groups (Supplementary Fig. 3D and 3E).

Comparisons of ECG findings between DOX and EPI groups: The arrhythmias observed in each dog are presented in Supplementary Table 1. ST-segment depression was the most common among the dogs and was observed in three dogs in the DOX group and four dogs in the EPI group, respectively. The cumulative dose at which ST-segment depression was observed was 275.7, 180, and 120 mg/m² in Dogs 1, 3, and 4 in the DOX group, and 184.2, 141, 30, and 120 mg/m² in Dogs 5–8 in the EPI group. The Cox proportional hazards model using the cumulative dose at the time when ST-segment depression was first detected showed that the hazard ratio for EPI relative to DOX to induce ST-segment depression was 3.26 (95% confidence interval: 0.5831–18.23) with P=0.178, indicating the difference was not statistically significant.

Although a J wave was observed in both DOX and EPI groups, it was already present at baseline in Dog 7. In the DOX group, one dog each exhibited left bundle branch block, right bundle branch block, and first-degree atrioventricular block.

Exploration of sensitive indices reflecting myocardial damages

Alterations in echocardiographic parameters: At baseline, the CV values of A4C-LS and A4C-TMAD, calculated from four dogs in the DOX and EPI groups, were 2.3 ± 1.5% and 3.0 ± 2.1%, respectively. After the completion of repeated administration, the CV values in the DOX group were 7.2% and 7.4% for A4C-LS, and 11.2% and 18.2% for A4C-TMAD. In the EPI group, the corresponding values were 7.1% and 6.0% for A4C-LS, and 12.1% and 13.2% for A4C-TMAD. Overall, the variability in A4C-LS and A4C-TMAD was within an acceptable range.

The comparisons of echocardiographic parameters between baseline and after the completion of repeated drug administration in each group are shown in Table 1. The following indices of cardiac function, mainly related to systolic function, significantly changed in the DOX group: A4C-LS, GLS, A4C-TMAD, MAPSE, MPI, and VTI, although no parameters significantly changed in the EPI group. Standard echocardiographic parameters, including FS, A4C-EF, and BP-EF, did not significantly change in either the DOX or EPI group. LVIDd also showed no significant change in either DOX or EPI group. The dose-dependent changes in these parameters for each dog are shown in Supplementary Fig. 4.

Correlations of histopathological findings with echocardiographic parameters and ECG findings: The six echocardiographic parameters that significantly changed after the completion of repeated DOX administrations, A4C-LS, GLS, A4C-TMAD, MAPSE, MPI, and VTI, were selected to examine their correlations with histopathological findings. No significant correlation was observed in any cardiac tissue location between either of these echocardiographic parameters and the vacuolated or eosinophilic degeneration cell count (Table 2). On the other hand, a significantly negative and strong correlation was observed between the collagen fiber area ratio in the LVFW and the A4C-LS (r= −0.82) and A4C-TMAD (r= −0.83).

Table 2. Correlations between histopathological findings in cardiac tissue and echocardiographic parameters.

			LVFW				AP		SW
			r	95% CI	P	Adjusted P	r	P	r	P
Vacuolated cell count	vs.	A4C-LS	0.28	-	0.51	-	0.51	0.20	0.27	0.52
		GLS	0.33	-	0.43	-	0.28	0.50	0.28	0.51
		A4C-TMAD	−0.16	-	0.71	-	0.19	0.66	−0.19	0.65
		MAPSE	−0.17	-	0.70	-	0.11	0.79	0.27	0.51
		MPI	0.12	-	0.77	-	0.41	0.32	0.34	0.41
		VTI	−0.17	-	0.69	-	−0.38	0.36	−0.11	0.80
Eosinophilic degeneration cell count	vs.	A4C-LS	−0.14	-	0.75	-	0.00	1.00	0.20	0.64
		GLS	−0.14	-	0.74	-	0.02	0.96	0.04	0.94
		A4C-TMAD	−0.35	-	0.40	-	−0.06	0.88	0.03	0.95
		MAPSE	0.05	-	0.90	-	0.29	0.48	0.14	0.74
		MPI	0.60	-	0.12	-	0.21	0.62	−0.11	0.80
		VTI	−0.38	-	0.36	-	−0.02	0.97	−0.09	0.84
Collagen fiber area ratio	vs.	A4C-LS	−0.82	−0.97− −0.26	0.01	0.03*	−0.23	0.58	0.17	0.69
		GLS	−0.74	−0.95− −0.08	0.03	0.06	−0.37	0.37	0.11	0.79
		A4C-TMAD	−0.83	−0.97− −0.31	0.01	0.03*	−0.05	0.91	−0.02	0.95
		MAPSE	−0.29	−0.83−0.52	0.49	0.49	−0.47	0.25	0.14	0.74
		MPI	0.37	−0.45−0.85	0.37	0.48	0.24	0.57	0.43	0.29
		VTI	−0.35	−0.85−0.47	0.40	0.48	−0.51	0.20	−0.15	0.72

*P<0.05. A4C-LS: longitudinal strain from the left apical four-chamber view, GLS: global longitudinal strain, A4C-TMAD: tissue motion mitral annulus displacement from the left apical four-chamber view, MAPSE: mitral annular plane systolic excursion, MPI: left ventricular myocardial performance index, VTI: left ventricular outflow tract velocity-time integral, LVFW: left ventricular free wall, AP: apex, SW: septal wall, r: correlation coefficient, 95% CI: 95% confidence interval.

Regarding ECG findings, the cumulative dose at the first detection of ST-segment depression was not significantly correlated with vacuolated cell counts in the LVFW (r=0.29, P=0.48), AP (r=0.45, P=0.26), or SW (r=0.26, P=0.53); with eosinophilic degeneration cell counts in the LVFW (r=−0.15, P=0.73), AP (r=−0.48, P=0.23), or SW (r=−0.15, P=0.73); or with the collagen fiber area ratio in the LVFW (r=−0.04, P=0.92), AP (r=0.08, P=0.85), or SW (r=0.28, P=0.50).

Changes in hs-cTnI concentrations: Figure 5 shows the changes in serum hs-cTnI concentrations throughout the administration periods (Fig. 5A and 5B) and up to a cumulative dose of 180 mg/m² (Fig. 5C and 5D). All dogs in the DOX group showed an increase in hs-cTnI concentrations above the upper reference limit within 3 weeks after receiving the cumulative dose of 180 mg/m², which is regarded as the clinical upper limit. In contrast, only two of the four dogs in the EPI group exhibited such an increase within the same period. Furthermore, the DOX group showed a significant increase in hs-cTnI concentrations after the final drug administration compared with baseline, whereas no such significant change was observed in the EPI group (Table 1).

Fig. 5.

Changes in hs-cTnI concentrations throughout the administration periods in the doxorubicin (DOX, A) and epirubicin (EPI, B) groups. Changes up to a cumulative dose of 180 mg/m² (indicated as dashed boxes in A and B) in the DOX (C) and EPI (D) groups are shown in an enlarged view. The horizontal dashed lines indicate the upper reference limit.

DISCUSSION

Comparison of the cardiotoxicity between EPI and DOX

In the histopathological examination, EPI induced significantly milder vacuolation in the ventricles compared with DOX in this study. Although not statistically significant, a similar trend was observed for eosinophilic degeneration. Therefore, EPI-induced myocardial damage was milder than that caused by DOX, based on histopathological findings. The present results were consistent with a previous toxicological study showing lower histopathological damage of myocardial tissues in dogs administered with EPI when compared to DOX-administered dogs [2]. In this previous study, a preclinical regimen was employed administering 0.4 mg/kg of DOX and EPI three times weekly for 13 weeks, resulting in higher weekly doses compared to the clinical regimen. In the present study, we employed regimens commonly used in clinical practice and confirmed the milder myocardial toxicity of EPI.

However, it should be noted that one of the dogs in the EPI group developed AICD, which was defined as consecutive FS reduction, when the cumulative dose reached 168 mg/m². Susceptibility to DOX-induced AIC has been reported to vary among individuals in both humans and dogs, resulting in different severities of myocardial damage despite receiving similar doses [17, 26, 35]. Therefore, the susceptibility to EPI-induced AIC may also vary among individuals. In three previous studies involving a total of 295 dogs that received the EPI-based protocol at a maximum cumulative dose of 210 mg/m², no decline in FS was reported [11, 21, 25]. However, echocardiographic evaluation was performed in only a portion of the dogs in those previous studies, suggesting that asymptomatic cardiac dysfunction may have been overlooked. Therefore, our findings indicate that EPI can cause AICD in dogs, even at cumulative doses below 180 mg/m².

Exploration of sensitive indices reflecting myocardial damage

In Dog 6, FS fell below 30% after a cumulative dose of 87 mg/m², but it improved temporarily before consecutive reductions. Dog 4 in the DOX group also showed FS below 30% once, but the reduction was not sustained. FS is influenced not only by systolic function but also by other factors including preload and afterload [13, 31]. Therefore, FS varies with hydration status and with neurohormonal tone regulating loading conditions, including the renin–angiotensin–aldosterone system, vasopressin, and the sympathetic nervous system. In addition, longitudinal systolic function declines in the early phase of myocardial damage, whereas circumferential and radial function increase compensatorily [31], resulting in FS being preserved until the late stage or even potentially appearing increased. Thus, the inconsistent changes in FS may reflect a combination of factors, including systolic function, loading conditions, and compensatory responses. Considering these variabilities and the insensitivity of FS for detecting AIC, further studies are essential to identify more stable and sensitive clinical markers, as demonstrated in the present study.

In the DOX group, six parameters related to systolic and overall LV function (i.e., A4C-LS, GLS, A4C-TMAD, MAPSE, MPI, and VTI) changed significantly after repeated DOX administration compared with baseline, whereas EF and FS did not. Both A4C-TMAD and A4C-LS showed CVs below 20% at both baseline and after repeated administration, indicating that intra-observer variability was within acceptable limits. These parameters are considered to have the greatest measurement variability among those evaluated in this study because their accuracy is influenced by operator proficiency [9]. A4C-LS, GLS, A4C-TMAD, and MAPSE reflect longitudinal LV mechanics, which are considered the most vulnerable and sensitive components to myocardial injury [28]. MPI reflects both systolic and diastolic LV function [1], whereas VTI reflects stroke volume and cardiac output [12]; both are regarded as indicators of overall LV function. In particular, GLS is considered the most sensitive marker for detecting AICD in humans, and its measurement during chemotherapy is strongly recommended in human clinical guidelines [24]. Although all parameters except MPI, as with FS and EF, are influenced by loading conditions, the more consistent changes observed in these six parameters suggest that they may detect AICD in dogs more sensitively than FS or EF. In contrast, no significant reductions in diastolic function parameters such as E, E/A, and e’ were observed in the present study. LV diastolic dysfunction has been reported to occur as part of AICD and to be associated with systolic dysfunction [30], and a significant reduction in E/A and e’ was observed before a decline in EF in human patients treated with EPI [27]. However, this change occurred after a significant reduction in longitudinal strain. Therefore, a decline in conventional diastolic function indices may occur following systolic dysfunction detected by longitudinal systolic parameters. In humans, GLS is considered more appropriate than diastolic function parameters for the assessment of AICD. In dogs, a significant reduction in E/A without a decline in FS has also been reported in a clinical study of DOX-treated cases [16]. However, longitudinal systolic function parameters were not assessed in that study, and it cannot be excluded that these parameters also changed, as did E/A. Based on these findings, parameters related to systolic and overall LV function may be more suitable for assessing AICD in dogs than conventional diastolic function parameters. However, further study is needed to examine the changes in diastolic function after treatments with DOX or EPI in detail by measurements of advanced echocardiographic parameters including strain rate, which was not measured in the present study. Although anthracyclines have also been reported to impair right ventricular (RV) systolic function, no significant changes in RV systolic function evaluated by RVFAC and TAPSE were observed in the present study. RVFAC is the most commonly used two-dimensional parameter for assessing RV function. However, it reflects only the two-dimensional motion of the RV, and may change only in the late phase of RV dysfunction [4]. In fact, a decline in RVFAC has been reported in only 10% of human patients [8]. Similarly, no significant reduction in TAPSE was observed in a meta-analysis of human patients treated with anthracyclines [39]. Our findings are consistent with these reports, suggesting that the clinical utility of assessing longitudinal systolic dysfunction in the RV for AICD detection may be limited, compared with that in the LV.

Assuming that these six parameters are promising clinical markers for assessing AICD, their relationships with myocardial vacuolation and eosinophilic degeneration were evaluated in dogs from both DOX and EPI groups. Although neither vacuolation nor eosinophilic degeneration of cardiomyocytes directly indicates myocardial cell death [10], they suggest sublethal injury. Vacuolation, in particular, may reflect mitochondrial swelling and subsequent dysfunction [36]. Therefore, these changes may be associated with impaired myocardial cell function, ultimately leading to AICD. However, in the present study, no correlation was observed between myocardial degeneration and the echocardiographic parameters that were significantly decreased in the DOX group. Even in a previous human study of 158 patients, which evaluated organelles such as mitochondria by electron microscopy, a poor correlation was observed between EF and histopathological changes in endomyocardial biopsy samples [12], consistent with the findings of the present study. In contrast, in the present study, A4C-LS and A4C-TMAD values measured after the final administration were significantly correlated with the extent of myocardial collagen fibers observed at autopsy. Two mechanisms are considered to contribute to myocardial fibrosis. The first is replacement fibrosis, which occurs secondary to myocardial cell loss. However, in the present study, myocardial damage was unlikely to have progressed to cell death based on the absence of apoptotic findings in IHC. The second mechanism is reactive fibrosis, which arises from chronic stimuli or inflammation without cell loss. This type of fibrosis has been observed in mice administered low-dose DOX and was attributed to inflammation following mitochondrial damage [37]. It is possible that a similar process also occurred in the cardiac tissues in the present study. Fibrosis itself can impair cardiac function by reducing cardiac compliance, even in the absence of myocardial cell loss. In human medicine, STE is considered a highly sensitive technique for detecting subtle myocardial contractile abnormalities [14, 20]. In addition to being indices of longitudinal contractile function, both A4C-LS and A4C-TMAD are STE-derived parameters. Therefore, these indices may have reflected myocardial status more sensitively compared to the conventional ones such as MAPSE, MPI and VTI. In addition, clear visualization of the left apical two- and three-chamber views is often difficult in dogs [41], and it might be one of the causes of no significant correlation between GLS and histopathological findings. Therefore, A4C-LS and A4C-TMAD, which can be obtained solely from the apical four-chamber view, may be more suitable than GLS. Although the relationship between compliance and longitudinal systolic function remains unclear, A4C-LS and A4C-TMAD may reflect the extent of fibrosis rather than myocardial degeneration. Further studies with large clinical cohorts are necessary to validate our results and to investigate their usefulness as predictors of events such as AICD and the development of heart failure.

Notably, Dog 6, which developed AICD following EPI administration, did not show markedly severe myocardial damage, including vacuolation or eosinophilic degeneration. Although the reason for the discrepancy between pathological and functional findings is unclear, one possible explanation is rapid progression of myocardial injury. In Dog 6, the severity of vacuolation and eosinophilic degeneration was comparable to that of other dogs despite a lower cumulative dose, suggesting that rapid progression of myocardial damage led to cardiac dysfunction before histopathological changes became apparent.

cTnI is a component of the troponin complex specific to cardiac muscle and is used as a circulating biomarker of myocardial damage. In dogs treated with DOX, conventional assays of cTnI did not show a significant increase [33, 38], whereas hs-cTnI, which can detect cTnI concentrations 10–100 times lower, showed a significant increase after the fourth administration of DOX [29]. In the present study, hs-cTnI levels exceeding the reference range were observed at a cumulative dose of 180 mg/m² in six of eight dogs, consistent with a previous report [29]. In addition, a marked increase followed by a decrease in hs-cTnI, forming a spike pattern, was observed in Dogs 1, 2, 4, and 6. Similar spike patterns of hs-cTnI and cardiac troponin T (cTnT) have also been observed in rats administered DOX [3]. Furthermore, another study demonstrated decreased IHC staining for cTnT in cardiomyocytes corresponding to increased circulating cTnT concentrations [19]. These findings suggest the possibility that the spike patterns occurred through the prominent and persistent myocardial damage, followed by a reduction in the number of cardiomyocytes releasing cTnI. Therefore, it is necessary to investigate the clinical significance and predictive value of hs-cTnI increases in canine cases treated with anthracyclines.

This study has several limitations. First, only healthy dogs were included. A recent study reported that tumor-bearing mice exhibited a more severe decrease in EF and FS than non-tumor-bearing mice [40], suggesting that the presence of tumors may affect cardiac responses to anthracyclines. Therefore, the response to anthracyclines observed in this study may differ from those in tumor-bearing dogs, and it remains unclear whether our results can be directly applied to clinical cases. Second, a previous study used an FS threshold of 28% to define DOX-induced decreases in FS [18], whereas the cutoff value was set at 30% in the present study. However, a more stringent requirement of two consecutive reductions was adopted, and the FS value at the time Dog 6 was diagnosed with AICD was 25.5%. In addition, the rate of reduction from baseline was 24.0%, which is sufficient to raise concern for cardiotoxicity [18]. Accordingly, this does not appear to undermine the diagnosis of AICD in this case. Third, the dose leading to heart failure could not be evaluated, as none of the dogs in this study developed an AICD severe enough to progress to heart failure. Further studies in clinical cohorts, including dogs with heart failure caused by DOX or EPI, are necessary to determine the lifetime cumulative dose limit of EPI. Fourth, a segmental functional assessment could not be performed in this study. Although GLS theoretically allows regional strain analysis using a polar map, it requires left apical two- and three-chamber views of sufficiently high quality, comparable to the four-chamber view. However, due to anatomical constraints, it is often technically challenging to obtain two- and three-chamber views with fully and clearly visualized endocardial borders in dogs. Therefore, only global functional indices were evaluated in the present study, which may have limited the anatomical specificity of the investigations of correlations between echocardiographic parameters and histopathological findings. Finally, the sample size was small, and the individual differences in myocardial injury and echocardiography parameters observed in the present study warrant investigation in larger cohorts.

In conclusion, pathological findings, echocardiographic results, and hs-cTnI suggested that EPI generally causes milder cardiotoxicity than DOX in healthy dogs. However, one dog in the EPI group developed cardiac dysfunction after receiving a cumulative dose of 168 mg/m², highlighting variability in individual sensitivity to EPI. The present study also suggested that the echocardiographic parameters A4C-LS, GLS, A4C-TMAD, MAPSE, MPI, and VTI, as well as biomarker hs-cTnI, are potential markers for detecting AIC and may be more sensitive than FS or EF. Further investigation with a larger clinical cohort is necessary to validate the usefulness of these parameters.

Conflicts of Interest

No potential conflicts of interest were disclosed.