Decoding anthracycline‐ and trastuzumab‐related cardiac dysfunction prediction: HFA‐ICOS scores versus strain imaging

Abstract

Aims

Baseline Heart Failure Association–International Cardio‐Oncology Society (HFA‐ICOS) scores and serial left ventricular global longitudinal strain (LV‐GLS) measurements have been found to be useful in predicting cancer therapy‐related cardiac dysfunction (CTRCD). However, their integration for the purpose of improving prognostic accuracy remains unclear; and we aimed to develop a predictive model for CTRCD using baseline HFA‐ICOS scores and the relative decline of LV‐GLS in patients on anthracycline or trastuzumab.

Methods

We prospectively enrolled 443 chemotherapy‐naïve women with breast cancer and cardiovascular risk factors, scheduled to receive anthracycline (n = 333) or trastuzumab (n = 110). Participants were stratified by the HFA‐ICOS risk score. The left ventricular ejection fraction (LVEF) and LV‐GLS were evaluated using echocardiography at baseline, before each treatment cycle, and every 3 months in the first year post‐chemotherapy. CTRCD was a new LVEF reduction ≥10 percentage points to an LVEF < 50%, irrespective of symptoms.

Results

In terms of HFA‐ICOS stratification, 258 patients (58.2%) were low risk, 180 (40.6%) were moderate risk and 5 (1.2%) were high risk. The proportions of low‐ and moderate‐risk patients were similar in the anthracycline and trastuzumab groups. Twenty‐four (7.2%) and seven (6.4%) patients treated with anthracycline and trastuzumab, respectively, displayed asymptomatic CTRCD. The addition of the baseline HFA‐ICOS risk score did not improve the performance of the significant relative decline of LV‐GLS > 15% in predicting both anthracycline [area under the receiver‐operating characteristic curve (AUC) 0.93, 95% confidence interval (CI) 0.89–0.96, sensitivity 87.5%, specificity 93.2%] and trastuzumab (AUC 0.97, 95% CI 0.88–0.99, sensitivity 85.7%, specificity 93.2%)‐related cardiac dysfunction.

Conclusions

Contemporary anthracycline and trastuzumab‐based regimens resulted in similarly low incidences of CTRCD. In this context, LV‐GLS evolution was the best predictor of CTRCD.

Left ventricular global longitudinal strain (LV‐GLS) is the best tool for predicting CTRCD due to anthracycline and trastuzumab.

Introduction

Anthracycline and HER‐2 inhibitors, such as trastuzumab, are cornerstones in chemotherapy for breast cancer. Cancer therapy‐related cardiac dysfunction (CTRCD) caused by these two regimens has been widely recognized in cardio‐oncology, for many years, as type I and type II cardiotoxicity, respectively. ¹ , ² Therefore, a conjunctive or subsequent regimen of HER‐2 inhibitors following anthracycline has been less commonly used in contemporary breast cancer chemotherapy protocols due to a significantly increased risk of CTRCD without a substantial improvement in treatment efficacy. The risk of trastuzumab‐related cardiac dysfunction has been reduced to 0.4%–3.2% in recent clinical trials. ³ , ⁴ The mean left ventricular ejection fraction (LVEF) decline post‐anthracycline chemotherapy completion has been estimated to be mild, at 5.4% (range 3.5%–7.3%), which may not appear as symptomatic heart failure in modern anthracycline dosing regimens. ⁵ Although efforts in recent years have significantly reduced the incidence of cardiac toxicity, predicting the risk of CTRCD in patients treated with anthracycline and trastuzumab remains a priority to optimize cardio‐oncology outcomes. ⁶ , ⁷

Left ventricular global longitudinal strain (LV‐GLS) on echocardiography is the most reliable method of detecting CTRCD early, on imaging during and after anthracycline or trastuzumab chemotherapy. ⁶ , ⁸ Compared with cardiac serum biomarkers, such as high‐sensitivity cardiac troponin (hs‐cTn) and B‐type natriuretic peptide (BNP/NT‐proBNP), LV‐GLS has shown superior prognostic value in the current literature. ⁹ , ¹⁰ , ¹¹ The relative reduction cut‐off of LV‐GLS > 15% had high sensitivity and specificity, ranging between 45% and 86% and 71% and 86%, respectively, in predicting anthracycline and trastuzumab‐related cardiac dysfunction. ⁹ The considerable variability in the accuracy of LV‐GLS can be attributed to differences in echocardiography surveillance frequency, CTRCD definitions and chemotherapy protocols among studies. Evidence supporting the role of LV‐GLS in patients undergoing trastuzumab treatment, however, has been less robust than that in those receiving anthracycline. ¹⁰ Additionally, prior to the commencement of anticancer therapy, the Heart Failure Association–International Cardio‐Oncology Society (HFA‐ICOS) score is the only recommended clinical risk stratification tool in recent European Society of Cardiology (ESC) guidelines, although this score has not been conclusively validated in prospective studies of various anticancer agents. ⁶ Several questions persist regarding whether LV‐GLS and the HFA‐ICOS score have similar roles in patients treated with anthracycline or trastuzumab, and whether the combination of these two well‐known tools can enhance prognostic accuracy in predicting CTRCD. The aim of this study was to develop an optimal predictive model for CTRCD prediction utilizing the baseline HFA‐ICOS score and the relative decline of LV‐GLS > 15% in a chemotherapy‐naïve breast cancer cohort with cardiovascular risk factors.

Methods

Design

This multicentre prospective cohort study recruited 443 consecutive female patients newly diagnosed with breast cancer (Stages I–IV), who received anthracycline or trastuzumab as adjuvant or neoadjuvant chemotherapy at Nhan Dan Gia Dinh Hospital and Oncology Hospital, Ho Chi Minh City, Vietnam, from 1 September 2020 to 31 December 2022. Patients received a four‐cycle anthracycline regimen (60 mg/m² per cycle of doxorubicin or equivalent) with a 21 day intercycle interval or an 18‐cycle trastuzumab regimen with a 21 day intercycle interval in the anthracycline and trastuzumab groups, respectively (Figure 1). Patients were eligible if they had at least one of the following cardiovascular risk factors: ≥65 years of age, hypertension, diabetes mellitus, dyslipidaemia, atrial fibrillation, obesity or chronic kidney disease. The exclusion criteria included anti‐HER2 and anthracycline combination chemotherapy, severe valvular heart disease and poor image quality on echocardiogram (defined as ≥2 inadequate visualized myocardial segments in an 18‐segment model). Patients were evaluated by standard demographic and clinical data, the HFA‐ICOS risk stratification score, ⁶ hs‐cTn I, comprehensive echocardiography and medical therapy at baseline. All patients underwent clinical examination and standard echocardiography, including LVEF and LV‐GLS, before every anthracycline or trastuzumab cycle, 3 weeks after the final dose and every 3 months in the first year after the final chemotherapy dose, irrespective of the HFA‐ICOS risk category. The primary endpoint of this study was the occurrence of CTRCD during the study period. This was defined as a new LVEF reduction by ≥10 percentage points to an LVEF < 50%. The persistent decline in the LVEF was confirmed by two subsequent echocardiography reports. All study participants provided informed consent, and the study protocol was approved by the Medical Ethical Committee of University of Medicine and Pharmacy at Ho Chi Minh (230/HDDD‐DHYD). The trial was performed in accordance with the principles of the Declaration of Helsinki.

Figure 1.

Study flowchart. Upon completion of the study, 333 patients receiving anthracycline and 110 patients receiving trastuzumab adhered to the echocardiography surveillance frequency as outlined in the study protocol.

Two‐dimensional echocardiography

Echocardiography was conducted at rest by a single examiner utilizing a Philips Affiniti ultrasound system (Philips Healthcare, Andover, MA, USA), following the guidelines of the American Society of Echocardiography and the European Association of Cardiovascular Imaging. ¹² , ¹³ Data were collected from four, three and two chambers across three consecutive cardiac cycles with a frame rate exceeding 50 fps, and stored in raw DICOM format for subsequent offline analysis. Images were captured at baseline (prior to the first cycle of chemotherapy), before each subsequent cycle, 3 weeks after the completion of chemotherapy and every 3 months in the first year thereafter. In total, at least 9 and 23 echocardiographic examinations were performed per patient in the anthracycline and trastuzumab groups, respectively, throughout the study period. All images were transferred to a core laboratory for computation by two cardiologists (H. H. N. and D. T. V.), who were blinded to the clinical data of the patients. Speckle‐tracking analyses of LV‐GLS and semiautomatic LVEF calculations, using the biplane Simpson rule, were executed using Philips aCMQ (QLAB 15.0, Philips Healthcare, Andover, MA, USA). The strain indices for each echocardiography report were compared with baseline values, with variations defined as relative change [relative change at each cycle = (current strain value − baseline strain value)/baseline strain value]. The baseline strain index was identical to the strain value prior to cycle 1. After the development of CTRCD, patients underwent echocardiography every 3 months until the end of the study. A total of 5417 echocardiography reports were analysed at the core lab, and the interobserver mean errors were 3.9% and 8.6% for LV‐GLS and LVEF, respectively.

Statistical analysis

Data are provided as means ± SDs when normally distributed, medians and IQRs for skewed distributions, and frequencies and percentages for categorical variables. A χ ² or unpaired t‐test was used to compare the demographic data, cardiovascular risk factors, HFA‐ICOS risk levels, and medical therapy between the anthracycline and trastuzumab groups, or between the CTRCD and no‐CTRCD groups. Due to the small number of high‐risk patients in this study, all participants were stratified into two risk levels: moderate‐high and low risk. Paired t‐tests were employed to demonstrate significant changes in ∆LV‐GLS and ∆LVEF between the independent groups at each anthracycline or trastuzumab cycle. ∆LV‐GLS and ∆LVEF were reported as relative and absolute reductions, respectively, from the baseline value. Receiver‐operating characteristic curves were created to define the sensitivity and specificity of the relative change in LV‐GLS > 15%, HFA‐ICOS ≥ 2 or the integrated model to predict CTRCD. Statistical analyses were performed using IBM SPSS Statistics V.27 (IBM Corp) and R V.4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). Two‐sided P values were used, and P < 0.05 was considered statistically significant.

Results

Patient characteristics

Of the 443 female patients with newly diagnosed breast cancer and cardiovascular risk factors, hypertension was the predominant cardiovascular risk factor in both the anthracycline and trastuzumab groups, affecting 287 (64.8%) participants. This was followed by dyslipidaemia in 175 (39.5%), diabetes mellitus in 115 (26%), chronic kidney disease in 37 (8.4%) and obesity in 29 (6.5%). At baseline, there were no significant differences in pre‐specified cardiovascular risk factors between the anthracycline (n = 333) and trastuzumab (n = 110) groups, other than age. Patients receiving trastuzumab chemotherapy were older than those treated with anthracycline (60.2 ± 8.3 and 57.8 ± 7.5 years, respectively). The median doxorubicin cumulative dose was 227 ± 11.8 mg/m², with no patient receiving a cumulative dose ≥250 mg/m² of doxorubicin or its equivalent. Due to patient selection by oncologists at the study sites, none of the participants presented with an LVEF < 50%; increased hs‐cTn I; prior CTRCD, heart failure or cardiomyopathy; or a history of coronary artery disease prior to anticancer treatment. According to the HFA‐ICOS score, 258 (58.2%) were at low risk, 180 (40.6%) at moderate risk and 5 (1.2%) at high risk. In general, the anthracycline and trastuzumab groups had similar HFA‐ICOS risk stratification, mean LVEF, mean LV‐GLS and cardioprotective therapies at baseline. Regarding neurohormonal therapies, angiotensin‐converting enzyme‐inhibitor (ACE‐i)/angiotensin receptor blocker (ARB) and beta‐blockers were initiated in nearly a quarter (23.9%) and 13.8% of patients, respectively (Table 1).

Table 1.

Baseline patient characteristics of the anthracycline and trastuzumab groups.

Characteristics	Anthracycline group (n = 333)	Trastuzumab group (n = 110)	P
Age (years)	57.8 ± 7.5	60.2 ± 8.3	0.005
BMI (kg/m2)	24.3 ± 3.3	24.2 ± 2.8	0.775
Type of chemotherapy
Doxorubicin, n (%)	265 (79.6)
Epirubicin, n (%)	68 (20.4)
Cumulative dose of doxorubicin (mg/m2)	227.7 ± 11.8	N.A.
Breast cancer stage, n (%)
Stage 0–I	11 (3.3)	9 (8.2)	0.168
Stage II	179 (53.8)	57 (51.8)	0.787
Stage III	120 (36)	35 (31.8)	0.758
Stage IV	23 (6.9)	9 (8.2)	0.765
Baseline cardiovascular risk factors, n (%)
Age ≥ 80 (years)	0	1 (0.9)	0.116
Age 65–79 (years)	136 (40.8)	63 (57.3)
Hypertension	218 (65.5)	69 (62.7)	0.717
Diabetes mellitus	90 (27)	25 (22.7)	0.453
Atrial fibrillation	4 (1.2)	1 (0.9)	0.346
Dyslipidaemia	132 (39.6)	43 (39.1)	0.532
Obesity	24 (7.2)	5 (4.5)	0.427
Chronic kidney disease	27 (8.1)	10 (9.1)	0.646
HFA‐ICOS category
Low risk	198 (59.5)	60 (54.5)	0.331
Moderate risk	131 (39.3)	49 (44,5)	0.454
High risk	4 (1.2)	1 (0.9)
Baseline neurohormonal therapy
ACE‐i/ARB	79 (23.7)	27 (24.5)	0.9
Beta‐blocker	48 (14.4)	13 (11.8)	0.897
Baseline cardiac biomarker
hs‐cTn I (ng/L)	9.3 ± 1.9	10.1 ± 2.3	0.398
Baseline left ventricular function
LVEF (%)	64 ± 5	63,6 ± 5.3	0.474
>55% (n %)	319 (95.8)	104 (94.5)	0.878
50–54% (n %)	14 (4.2)	6 (5.5)	0.678
LV‐GLS (%)	−18.8 ± 1.3	−18.9 ± 1.1	0.469
< −18% (n %)	253 (76)	85 (77.3)	0.565
−16% to −18% (n %)	72 (21.6)	25 (22.7)	0.766
> −16% (n %)	8 (2.4)	0	0.258

Note: Values are mean ± SD or n (%).

Abbreviations: ACE, angiotensin‐converting enzyme; ARB, angiotensin receptor blocker; BMI, body mass index; HFA‐ICOS, Heart Failure Association–International Cardio‐Oncology Society; LV‐GLS, Left ventricular global longitudinal strain; LVEF, left ventricular ejection fraction.

Development of CTRCD

According to similar baseline clinical and echocardiographic characteristics, 24 (7.2%) and 7 (6.4%) patients receiving anthracycline and trastuzumab, respectively, developed asymptomatic CTRCD during active chemotherapy and the 12 months thereafter. Anthracycline and trastuzumab exhibited significantly different patterns in CTRCD development. While 79.2% of anthracycline‐related cardiac dysfunction cases were detected after chemotherapy completion, especially in the first 3 months, 100% of CTRCD due to trastuzumab was recorded during the 18‐cycles of active treatment (Figure 2). CTRCD occurred more frequently in patients stratified as moderate‐high baseline risk (HFA‐ICOS score ≥ 2) than in those stratified as low risk in both the anthracycline (12.9% vs. 3.8%, P = 0.01) and trastuzumab (12% vs. 1.9%, P = 0.007) groups. The HFA‐ICOS ≥ 2 cut‐off point at baseline effectively stratified the risk of CTRCD during the study period (Figure 3).

Figure 2.

Differences in the development patterns of anthracycline and trastuzumab‐related cardiac dysfunction. The data presented provide a comprehensive analysis of the occurrences and probabilities of CTRCD in patients undergoing treatments with anthracycline (A) and trastuzumab (B).

Figure 3.

Baseline HFA‐ICOS risk score in predicting CTRCD. Kaplan–Meier estimation for CTRCD due to anthracycline (A) and trastuzumab (B) according to baseline HFA‐ICOS risk score ≥ 2. CTRCD occurred more frequently in patients stratified as moderate‐high baseline risk (HFA‐ICOS score ≥ 2) than in those stratified as low risk in both the anthracycline (12.9% vs. 3.8%, P = 0.01) and trastuzumab (12% vs. 1.9%, P = 0.007) groups.

Evolution of LVEF and LV‐GLS

Although the reduction of LV‐GLS consistently preceded the decline of LVEF, these echocardiographic evolutions exhibited different behaviours following exposure to anthracycline and trastuzumab. Within the cohort, mean LVEF did not deteriorate by the end of the study, irrespective of the chemotherapy regimens used. However, patients who developed anthracycline‐related cardiac dysfunction demonstrated a significant reduction in mean LVEF at 1 year follow‐up, compared with the baseline measurement (53.6 ± 5.2% vs. 63.5 ± 4.6%, P < 0.00001). This decline was not observed in those with trastuzumab‐related cardiac dysfunction (59.7 ± 3.6% vs. 59.7 ± 6.5%, P = 0.531). Regarding the evolution of LV‐GLS, the mean LV‐GLS in all patients receiving anthracycline showed a significant decrease after 12 months of the completion of chemotherapy (−18.3 ± 1.4% vs. −18.8 ± 1.3%, P < 0.0001). Conversely, in the trastuzumab group, the mean LV‐GLS only decreased by the end of study, and only in patients who developed CTRCD (−17.2 ± 1.2 vs. −18.9 ± 1.2, P < 0.00001) (Figure 4).

Figure 4.

Evolution of LVEF and LV‐GLS in breast cancer patients on anthracycline and trastuzumab. We present data for the CTRCD group (red line) and the no‐CTRCD group (black line). At the end of the study period, breast cancer patients who developed anthracycline‐related cardiac dysfunction exhibited a significant decrease in LVEF (A) (53.6% ± 5.2% vs. 63.5% ± 4.6%, P < 0.00001) and LV‐GLS (C) (−15.5% ± 1.5% vs. −18.8% ± 1.7%, P < 0.00001) compared with baseline values. In patients receiving trastuzumab, mean LVEF (B) remained stable, irrespective of CTRCD development. However, mean LV‐GLS at 12 month follow‐up (D) showed a significant reduction in the CTRCD group (−17.2% ± 1.2% vs. −18.9% ± 1.2%, P < 0.00001).

Integrated approach in predicting CTRCD

The utilization of a relative reduction of LV‐GLS > 15% was demonstrated to exhibit good accuracy in predicting both anthracycline (sensitivity 87.5%, specificity 93.2%, positive predictive value 50% and negative predictive value 99%) and trastuzumab‐related cardiac dysfunction (sensitivity 85.7%, specificity 93.2%, positive predictive value 46% and negative predictive value 99%). Performance tests indicated that the HFA‐ICOS score ≥ 2 cut‐off had a sensitivity of 70.8% and 85.7%, and a specificity of 61.8% and 55.3%, when predicting the development of asymptomatic CTRCD due to anthracycline and trastuzumab, respectively.

In general, the baseline HFA‐ICOS stratification did not enhance the predictive value of the relative reduction of LV‐GLS during the study period. The area under the receiver‐operating characteristic curve (AUC) for the baseline HFA‐ICOS core ≥ 2, the relative decline of LV‐GLS > 15% and the integrated model of these two established tools in patients receiving anthracycline was 0.66 [95% confidence interval (CI) 0.61–0.71], 0.93 (95% CI 0.89–0.96), and 0.91 (95% CI 0.84–0.97), respectively. Regarding prediction of CTRCD due to trastuzumab, LV‐GLS also demonstrated the best performance with an AUC of 0.97 (95% CI 0.88–0.99), compared with the HFA‐ICOS score (AUC 0.66, 95% CI 0.48–0.93) and the integrated model (AUC 0.96, 95% CI 0.86–0.99) (Figure 5).

Figure 5.

Predictive models for CTRCD due to anthracycline and trastuzumab. receiver operator characteristic curves to predict CTRCD due to anthracycline (A) and trastuzumab (B) using baseline HFA‐ICOS risk score ≥ 2 (blue line), relative reduction of LV‐GLS > 15% (green line) and the integrated model (red line).

Discussion

The results of this study revealed similarly low incidences of anthracycline and trastuzumab‐related cardiac dysfunction in patients with comparable baseline cardiovascular risk factors and left ventricular systolic function. The low likelihood of developing significant left ventricular dysfunction poses a challenge in formulating an effective strategy to identify populations at increased risk of CTRCD. In this context, the relative reduction of LV‐GLS demonstrated exceptional performance in predicting CTRCD for both anthracycline and trastuzumab. Baseline HFA‐ICOS not only exhibited lower accuracy but also did not significantly enhance the predictive performance of LV‐GLS (Graphical abstract).

Optimizing echocardiography strategies for detecting CTRCD due to anthracycline and trastuzumab

To the best of our knowledge, this is the first prospective study to identify the incidence and compare differences in the development patterns of type I and type 2 chemotherapy‐induced cardiotoxicity within the same real‐world cohort of patients with breast cancer and comparable cardiovascular risk factors. Importantly, the high frequency of echocardiography surveillance may accurately reflect the true incidence of CTRCD. It also ensured that all adverse events were detected early and addressed at an asymptomatic stage. Our study further showed that anthracycline and trastuzumab had comparable effects on LVEF evolution as mean baseline LVEF values were similar prior to anthracycline or trastuzumab initiation (64.5 ± 5% and 63.6 ± 5.3%, P = 0.474) and at the 1 year follow‐up (63.6 ± 5.7% and 64.5 ± 4.9%, respectively, P = 0.459). The low incidences of anthracycline‐ and trastuzumab‐related cardiac dysfunction (7.2% and 6.4%, respectively) reflect the safety of contemporary chemotherapy protocols. Moreover, as all patients had no exposure to cancer treatments before enrolling in this study and received only anthracycline or trastuzumab‐based regimens, these results substantiated the limited data regarding the independent impact of these two anticancer treatments on cardiotoxicity.

The incidence of anthracycline‐induced cardiac dysfunction has decreased notably in recent years, approaching the incidence of CTRCD due to trastuzumab. Previously, the incidence of anthracycline‐related cardiac dysfunction ranged from 9.3% to 43.8%. ⁹ In the SUCCOUR (Strain Surveillance of Chemotherapy for Improving Cardiovascular Outcomes) trial, only 11% and 5% of patients developed LVEF‐ and LV‐GLS‐guided anthracycline‐related cardiac dysfunction at 1 year follow‐up, respectively. Notably, 21% of participants were stratified as high/very high HFA‐ICOS risk at baseline, and most of them received chemotherapy protocols containing both anthracycline and trastuzumab. ¹⁴ Contrary to the results of previous studies indicating significant long‐term reductions in mean LVEF following anthracycline exposure, regardless of CTRCD occurrence, the data from our cohort, SUCCOUR ⁴ and the recent meta‐analysis by Jeyaprakash et al. ⁵ demonstrated that mean LVEF generally remained stable during the study periods with modern anthracycline dosing. Importantly, in patients who developed CTRCD, most adverse events occurred in the early stage after the completion of anthracycline chemotherapy. The median elapsed time between the final cycle of anthracycline and cardiotoxicity development was reported as 3.5 months by Cardinale et al. ¹⁵ Our study found that 41.7% of anthracycline‐related cardiac dysfunction was detected by echocardiography surveillance within 3 months after regimen completion, and 40% of these CTRCD events occurred in low‐risk patients. According to current ESC guidelines, additional echocardiography is not recommended within 3 months after treatment completion for low or moderate HFA‐ICOS risk patients. ⁶ This recommendation was challenged by our findings, as most patients with breast cancer in our cohort were classified as low or moderate risk. We suggest that the protocol of repeating echocardiography within the first 3 months post‐anthracycline treatment should be applied not only to high‐/very high‐risk patients but also to the low‐ and moderate‐risk ones, to detect CTRCD early and initiate cardioprotective therapies promptly.

Real‐world data indicated that patients treated with trastuzumab were less likely to receive guideline‐adherent echocardiography monitoring compared with those on other chemotherapy regimens, such as anthracycline or taxane. ¹⁶ This discrepancy may be attributed to the long treatment duration and the perceived low incidence of trastuzumab‐related cardiac dysfunction reported in contemporary clinical trials; this was approximately 3%–4%. ¹⁷ , ¹⁸ Data regarding the incidence and development characteristics of CTRCD due to trastuzumab are limited in the current literature, and recommendations are predominantly based on expert opinion. ⁶ , ¹⁹ Consequently, unlike the established primary prevention strategies for anthracycline, a single algorithm for LVEF and LV‐GLS echocardiography surveillance is recommended for all cancer patients receiving trastuzumab, regardless of baseline HFA‐ICOS risk. ⁶ In our cohort, more than half (54.5%) of patients receiving trastuzumab were stratified as low risk, and no instances of CTRCD were recorded 3 months after initiating trastuzumab in these participants. This suggests that a one‐size‐fits‐all echocardiography surveillance protocol every 3 months during the active trastuzumab treatment period may not be necessary for patients at low risk. In other words, echocardiography could be repeated every 6 months after the first surveillance at 3 months in this risk category. Notably, when echocardiography was repeated before every cycle of trastuzumab, the incidence of trastuzumab‐related cardiac dysfunction reported in our study more accurately reflected the comparable impact of trastuzumab on cardiotoxicity relative to anthracycline. However, CTRCD due to trastuzumab was transient as mean LVEF did not significantly decrease at 1 year follow‐up compared with the baseline value in patients that developed trastuzumab‐related cardiac dysfunction. The median time to recovery of LVEF in trastuzumab‐related cardiac dysfunction was 5 to 6 weeks, with most CTRCD cases recovering fully, even without the administration of heart failure medications in many instances. ²⁰ , ²¹ When echocardiography surveillance was conducted every 3 months, as in many previous trials, the incidence of CTRCD due to trastuzumab may have been underestimated because CTRCD can develop and recover spontaneously between consecutive echocardiographic assessments.

Prediction of CTRCD in patients on anthracycline and trastuzumab

According to the ESC guidelines, at baseline, the HFA‐ICOS score, which incorporates cardiovascular history/risk factors, cardiac imaging, biomarkers and previous exposure to potentially cardiotoxic cancer therapies, is recommended for stratifying the risk of cardiovascular adverse events. This score aids in organizing personalized preventive strategies and cardiovascular monitoring. It is important to note that the HFA‐ICOS score was developed for risk estimation of all cardiovascular complications, not limited to left ventricular dysfunction, and prospective validation is necessary for its application to specific anticancer treatments. ⁶ Additionally, following the initiation of anthracycline or trastuzumab, cardiac imaging is recommended, preferably via echocardiography with LV‐GLS, along with measurement of cardiac biomarkers such as hs‐cTn and BNP/NT‐proBNP, to detect subclinical myocardial injury. ⁶ Although either new relative decline in LV‐GLS > 15% or new increases in cardiac biomarkers are considered important criteria for identifying the early stages of CTRCD, LV‐GLS monitoring is supported by more robust evidence. ² In the CARDIOTOX (CARDIOvascular TOXicity induced by cancer‐related therapies) registry, neither baseline hs‐cTn nor NT‐proBNP demonstrated prognostic value for the development of CTRCD with LVEF < 40% or clinical heart failure, given that 84.5% and 20.5% of participants were treated with anthracycline and trastuzumab, respectively. ²² The release of cardiac biomarkers varies for different cancer treatments, and more importantly, the cut‐off values for new increases in hs‐cTn or BNP/NT‐proBNP for defining CTRCD have not been established. ²² On the contrary, a relative reduction in LV‐GLS of >15% compared with baseline has been suggested as the ideal tool to predict future significant LVEF decrease and guide cardioprotective therapy. ¹⁰ , ¹⁴ , ²³ However, the prognostic value of LV‐GLS in predicting LVEF deterioration and subsequent development of heart failure remains less established in patients undergoing non‐anthracycline chemotherapy, such as trastuzumab.

In addition to the high negative predictive value of a new relative reduction of LV‐GLS > 15% in patients receiving anthracycline, our study notably showed that the absence of this significant decline in LV‐GLS can also safely rule out the development of CTRCD events during follow‐up in patients treated with trastuzumab who had similar baseline cardiovascular risk factors. To the best of our knowledge, this is the first prospective study employing a strict echocardiography protocol to confirm the role of LV‐GLS in predicting cardiac dysfunction due to non‐anthracycline containing chemotherapy. Previous small studies have demonstrated AUCs for new relative reduction of LV‐GLS in predicting CTRCD up to 0.84 and 0.85 in patients receiving anthracycline and trastuzumab, respectively. ⁹ Although the cut‐off for relative reduction of LV‐GLS varied among studies, the better performance of LV‐GLS monitoring in our study can best be explained by the higher frequency of echocardiography surveillance during active treatment and follow‐up. Recognizing that the more frequently echocardiography surveillance is performed, the greater the likelihood of detecting CTRCD, in our study LV‐GLS was systematically assessed at each chemotherapy cycle, irrespective of anthracycline or trastuzumab use, and subsequently at 3 month intervals. While LV‐GLS has been consistently established as an ideal tool for early prediction of subsequent decrease in LVEF, the evidence supporting LV‐GLS‐guided cardioprotective management to prevent the development of heart failure remains to be debated. ⁴ , ²⁴ The relatively low incidence of CTRCD with contemporary chemotherapy protocols has made it challenging to demonstrate the benefit of early initiation of heart failure therapies based on significant reduction of LV‐GLS in trials to date.

The external validation of HFA‐ICOS is also complicated by the minimal overall change in LVEF following anthracycline or trastuzumab exposure. Few studies involving patients undergoing anthracycline‐based regimens have indicated acceptable performance of the HFA‐ICOS score in relation to its prognostic value in predicting CTRCD. In a retrospective validation using the CARDIOTOX registry, HFA‐ICOS exhibited good discriminatory power, with an AUC of 0.78 for predicting CTRCD at 12 months. Additionally, patients classified as moderate risk, defined by an HFA‐ICOS ≥ 2, had more than three times the risk of developing CTRCD compared with those at low risk. ²² However, this finding was not replicated in a recent prospective validation cohort of 172 patients with breast cancer treated with anthracycline and trastuzumab, ²⁵ as well as in the SUCCOUR trial. ⁴ While stratifying patients at baseline as moderate risk (HFA‐ICOS ≥ 2) proved to be a useful strategy to identify those at higher risk of subsequent CTRCD, our study indicated that this prognostic approach was less accurate than monitoring LV‐GLS. Notably, the predictive performance of the LV‐GLS model was not enhanced by incorporating HFA‐ICOS stratification. Conversely, data from both Milk et al. ²⁶ and Negishi et al. ²⁷ revealed that a combined model, which included LV‐GLS and clinical variables, exhibited superior receiver‐operating characteristics than did individual models. However, these small studies had variations in their clinical predictive model components, and the prognostic value of relative reduction of LV‐GLS was limited by the low frequency of echocardiography monitoring during active treatment and follow‐up periods. Perhaps it is still an unmet need to adjust the scores of several more potent components within the HFA‐ICOS risk stratification tool to improve its discrimination power. Accordingly, monitoring LV‐GLS in an effective manner to detect a new relative reduction in LV‐GLS > 15% remains the most reliable strategy to detect myocardial injury due to both anthracycline and trastuzumab early, as supported by the current literature.

Study limitations

Similar to most real‐world data in cancer patients receiving anthracycline and trastuzumab, the majority of patients enrolled in this study were classified as low to moderate risk due to the candidate selection protocol for potential cardiotoxicity chemotherapy at the study sites. Consequently, the findings may not be representative of the prognostic value of LV‐GLS and HFA‐ICOS in the high‐risk HFA‐ICOS category. Although cardiac biomarkers are not recommended to be re‐assessed frequently in non‐high‐risk patients, another significant limitation of this study was the exclusion of biomarkers during follow‐up. This choice was primarily influenced by the absence of standardized recommendations for repeating biomarker assessments at the time the study protocol was developed in September 2020. In addition, prospective validation in a relatively small cohort of patients treated with trastuzumab may reduce confidence in the results, necessitating further evaluation in larger cancer patient populations.

Conclusions

Contemporary anthracycline and trastuzumab‐based regimens result in similarly low incidences of CTRCD within an Asian population. The new relative reduction of LV‐GLS > 15% demonstrated superior predictive performance compared with the baseline increased HFA‐ICOS score in predicting CTRCD due to both anthracycline and trastuzumab.

Conflict of interest statement

The authors have no relationships relevant to the content of this paper.

Nguyen, H. H. , Giang, N. M. , Vo, D. T. , Ho, T. H. Q. , and Ngoc‐Hoa, C. (2025) Decoding anthracycline‐ and trastuzumab‐related cardiac dysfunction prediction: HFA‐ICOS scores versus strain imaging. ESC Heart Failure, 12: 3667–3677. 10.1002/ehf2.15399.