Global Longitudinal Strain as Predictor of Chemotherapy-Induced Cardiotoxicity

Background: Chemotherapy-induced cardiotoxicity (ChC) is an important complication among patients receiving anthracyclines. Biomarkers and imaging parameters have been studied for their ability to identify patients at risk of developing ChC. Left ventricular global longitudinal strain (LV-GLS) is a sensitive parameter for detecting systolic dysfunction despite the presence of preserved left ventricular ejection fraction (LVEF). Objective: To evaluate the role of the LV-GLS as a predictor of ChC. Methods: This was a post-hoc analysis of the Carvedilol for Prevention of Chemotherapy-Related Cardiotoxicity trial, which evaluated the primary prevention of cardiotoxicity with carvedilol during doxorubicin chemotherapy in a population of patients with breast cancer. Cardiotoxicity was defined as a reduction ≥10% in LVEF. LV-GLS was determined before chemotherapy in patients with no prior cardiovascular disease or echocardiogram abnormalities. Results: Thirty-one patients for whom a complete echocardiography study including measurement of LV-GLS was performed before chemotherapy were included in this analysis. An absolute LV-GLS<16.9% before chemotherapy showed 100% sensitivity and 73% specificity for predicting cardiotoxicity (area under the curve [AUC], 0.85; 95% confidence interval [CI], 0.680–0.959; p<0.001). In this population, LVEF values before chemotherapy did not predict ChC (95% CI, 0.478 to -0.842; p=0.17). The association of low LV-GLS (<17%) and brain-type natriuretic peptide serum levels (>17 pg/mL) at 2 months after chemotherapy increased the accuracy for detecting early-onset ChC (100% sensitivity, 88% specificity; AUC, 0.94; 95% CI, 0.781–0.995; p<0.0001). Conclusions: Our data suggest that LV-GLS is a potential predictor of ChC. Larger studies are needed to confirm its clinical relevance in this clinical setting.


Introduction
[7] The detection of cardiotoxicity is routinely performed by left ventricular ejection fraction (LVEF).Although LVEF predicts the occurrence of HF, it has limited sensitivity. 8Failure to detect subtle changes in LV systolic function occurs for many reasons: the need for geometrical assumptions for calculations, possible inadequate visualization of the LV apex, impossibility of identifying marginal regional wall motion abnormalities, and intrinsic measurement variability. 9A decreased LVEF after chemotherapy is often a sign of already extensive myocardial damage and HF. 10 Due to increased morbidity and mortality among patients with chemotherapy-related HF, higher-sensitivity markers of subclinical cardiac dysfunction and myocardial injury have been investigated to detect chemotherapy-induced cardiotoxicity (ChC).For this purpose, the evaluation of two-dimensional speckle-tracking imaging has emerged.This technique allows for the study of global and regional myocardial deformation.Several studies have already emphasized the role of LV global longitudinal strain (LV-GLS) to detect subtle alterations in systolic function particularly related to ANT chemotherapy. 11The evaluation of GLS for the detection of subclinical LV dysfunction induced by chemotherapy is recommended by expert consensus. 12nsidering this new field assessing ANT-induced cardiotoxicity using LV-GLS, we conducted a post-hoc analysis of the randomized double-blind placebo-controlled Carvedilol Effect in Preventing Chemotherapy Induced CardiotoxicitY (CECCY) trial, which aimed to evaluate the LV-GLS before ANT chemotherapy as a predictor of cardiotoxicity.

Study design
This post-hoc analysis of the CECCY trial evaluated the primary prevention of cardiotoxicity with carvedilol during doxorubicin chemotherapy in women with breast cancer.
Cardiotoxicity was defined as a ≥10% reduction in LVEF.Patients were recruited and followed up at two different institutions, the Heart Institute and the Cancer Institute from the University of Sao Paulo, Sao Paulo, Brazil.The Ethics Committee of the Heart Institute and Cancer Institute from the University of Sao Paulo, Sao Paulo, Brazil, review board of both institutions approved the trial protocol.All methods were performed in accordance with the relevant guidelines and regulations.All participants were informed of the research objectives, research protocol, and treatment alternatives, and all participants provided written informed consent to participate.The trial was registered at ClinicalTrials.gov(NCT01724450) before study initiation.

Study patients
The CECCY trial included patients with HER2-negative breast cancer tumor status and therapy that included ANT, cyclophosphamide, and taxane from April 2013 to January 2017.The standard chemotherapy protocol comprised four cycles of cyclophosphamide 600 mg/m 2 and doxorubicin 60 mg/ m 2 every 21 days (with a total cumulative dose of 240 mg/m 2 ), followed by paclitaxel 80 mg/m 2 weekly for 8 weeks.The trial design and results were described elsewhere. 13

Study procedures
The present post-hoc analysis included only patients who underwent echocardiography studies and accomplished follow-up at the Heart Institute from the University of Sao Paulo, where the institutional protocol included speckle tracking echocardiography.Eligible patients underwent comprehensive transthoracic echocardiography before starting chemotherapy including proper imaging acquired to perform the strain analysis.Patients with an unsatisfactory acoustic window for the speckle tracking analysis due to artifacts caused by breast reconstruction techniques were excluded from the study.Echocardiographic studies were performed using a commercially available system (Vivid E9; General Electric, GE Vingmad Ultrasound AS, Norway) equipped with a 2-5-MHz transducer.All measurements were performed and reported according to American Society of Echocardiography recommendations (Lang 2015, Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging).LVEF was measured by Simpson's rule.

Original Article
with an adjusted device to record three cardiac cycles within a period of 100 msec before and after the cycle.The second harmonic image, in grayscale, with a frame rate of 40-80 frames/s was used.To measure LV-GLS, cardiac images were obtained through the apical windows (APLAX, A4C and A2C).After the acquisition, the studies were stored for offline analysis with the EchoPAC software (v30 12; GE Vingmad Ultrasound AS).The analysis was performed of the sixteen LV segments, and the quantitative peak systolic longitudinal strain was quantified for each segment as well as the whole LV during a cycle cardiac.All scans were read by experienced board certified echocardiographers who were blinded to the patients' clinical information.

Statistical analysis
The data are expressed as median and 95% confidence interval.We tested the normality of a variable distribution using the D'Agostino-Pearson test.For comparisons between two independent samples, an unpaired t-test was used for variables with Gaussian distribution, while the Mann-Whitney rank sum test was applied for variables with non-Gaussian distributions.For comparison between paired samples with Gaussian distribution, the paired t-test was used, while the Wilcoxon matched pairs ranked test was used for pairs with a non-Gaussian distribution.We used receiver operating characteristic (ROC) curve analysis to determine the accuracy and optimal cutoff.The best cutoff for each variable was chosen through the shortest distance from the upper left angle to the curve obtained in the graph of the ROC curve by the method of DeLong et al. 14 P values <0.05 were considered significant.

Results
In this post-hoc analysis of the randomized double-blind placebo-controlled CECCY trial, we evaluated 53 patients who underwent echocardiography with the speckle tracking technique at the Heart Institute.Of these, six (11%) had an unsatisfactory acoustic window and were excluded from the analysis.Another 16 patients were excluded for attending subsequent echocardiographic follow-up at another institution.The population included in the analysis was at low risk of cardiovascular.The baseline characteristics of the remaining 31 patients are described in Table 1.Our population had low prevalence of cardiovascular comorbidities and risk factors for cardiotoxicity.In this group, 3 (9.7%)developed cardiotoxicity (LVEF decrease ≥10% from baseline).Considering the alternative and most accepted definition of cardiotoxicity as a decrease of 10 percentage points to a value below the low normal value of 50%, 15 only 1 patient fulfilled the criteria for cardiotoxicity and had a decrease of LVEF to 35% and a GLS of 13%.The 3 patients with cardiotoxicity were aged 51-63 years, only one had a diagnosis of arterial hypertension, and none had other risk factors for cardiovascular disease such as diabetes mellitus, hypercholesterolemia under statin treatment, or current/past smoking status.
Table 2 shows the median and 95% CI values for serum LVEF, BNP, and troponin before and 2, 4, and 24 weeks after starting chemotherapy stratified according to baseline LV-GLS.Baseline LVEF was not significantly different in patients with an LV-GLS ≤16.9% versus >16.9%.Similarly, there was no intergroup difference in baseline BNP or troponin

Original Article
values.However, LVEF evaluated by echocardiography was significantly lower after 4 weeks of chemotherapy in the group with a baseline LV-GLS ≤16.9% (p=0.003).Furthermore, serum BNP measured after 4 weeks of chemotherapy treatment was higher in the group with an LV-GLS ≤16.9% (p=0.004).There was no significant intergroup difference for troponin values at any time point.
Figure 2 shows median and 95% CI values for LVEF and LV-GLS before and after 12 months after chemotherapy.We observed that LV-GLS decreased significantly from baseline values (p=0.005),whereas LVEF did not.In the follow-up period after chemotherapy, LV-GLS decreased more than 5%, 10%, and 15% from baseline in 77%, 66%, and 42% of patients, respectively, whereas LVEF decreased more than 10% in only 9.7% of patients.

Discussion
In this post-hoc analysis of the randomized double-blind placebo-controlled CECCY trial that evaluated the role of the speckle tracking echocardiography in ANT-induced cardiotoxicity, LV-GLS was a potential predictor of ChC in patients with a low prevalence of cardiovascular comorbidities and risk factors for cardiovascular disease.In this scenario, LV-GLS was a better predictor of cardiotoxicity than LVEF.In addition, the combination of LV-GLS and BNP during follow-up could be a predictor of cardiotoxicity.
There has been great interest in the early detection of cardiotoxicity to reverse and prevent associated cardiomyopathy. 16LVEF is a strong predictor of cardiac events, but it lacks sensitivity for the detection of subclinical changes in cardiac function. 17Strain is defined as change in the length of the myocardium divided by the original length of the myocardium and peak systolic deformation between systole and diastole. 18LV-GLS has emerged as the main measurement of subclinical myocardial dysfunction and demonstrated utility in predicting subsequent reductions in LVEF in patients after cancer treatment. 8,19,20Ali et al. 8 demonstrated that an absolute LV-GLS value of less than -17.5% was associated with an increase in HF among patients with hematologic cancer undergoing ANT chemotherapy.Charbonnel et al. 11 showed that an LV-GLS value of greater than -17.45% obtained after ANT 150 mg/m 2 is an independent predictor of future cardiotoxicity.
Our study findings are in concordance with other studies that showed that the strain measure before chemotherapy predicts the development of cardiotoxicity. 19e SOCCOUR trial recently compared cardioprotection guided by changes in LV-GLS versus LVEF among patients undergoing ANT chemotherapy.In the trial, 331 patients were randomized to receive angiotensin-converting enzyme inhibitors or angiotensin receptor blockers and betablockers guided by a ≥12% relative reduction in LV-GLS (GLS-guided arm) or 10% absolute reduction in LVEF (EF-guided arm).Patients were followed for LVEF and cancer

Original Article
therapy-related cardiac dysfunction (symptomatic drop of >5% or asymptomatic drop of >10% to <55%).At 1 year of follow-up, the LVEF did not change significantly in either group.However, in the GLS-guided arm, there was a greater use of cardioprotection and fewer patients met the cardiotoxicity criteria (5.8% vs. 13.7%;p=0.02).Patients who received cardioprotection in the EF-guided arm exhibited a greater reduction in LVEF at follow-up than those in the GLSguided arm (9.1±10.9% vs. 2.9±7.4%;p=0.03) supporting the use of GLS for detecting cardiotoxicity. 21konomou et al. 22 published a meta-analysis that evaluated the prognostic value of GLS for predicting cardiotoxicity and included 21 studies comprising 1782 patients with cancer treated with ANT with or without trastuzumab.The authors found an incidence of cardiotoxicity of 9.3-43.8%.Four studies evaluated the association of GLS before treatment initiation with subsequent cardiotoxicity: 2 found no association, 23,24 1 reported a significant association (odds ratio per 1% decrease, 1.48; 95% CI, 1.15-1.89)and 1 reported an AUC of 0.76 (95% CI, 0.58-0.88),with an optimal cutoff value of -19.95% (sensitivity, 83%; specificity, 72%) for cardiotoxicity. 11her biomarkers have been studied as a strategy for the early detection and monitoring of cardiotoxicity.The most studied biomarkers in cardiotoxicity include troponin and BNP.Regarding troponin, strong evidence favors its ability to predict cardiotoxicity and cardiac events. 15,25However, the utility of BNP for chemotherapy-related cardiotoxicity remains controversial, with many studies reporting no prognostic value in this scenario. 26Our study showed that the combination of the LV-GLS pre-chemotherapy with BNP during follow-up (until 24 weeks) could be a greater predictor of a >10% decrease in LVEF.However, the combination of LV-GLS with troponin showed no prognostic value regarding cardiotoxicity.

Limitations
This was a post-hoc analysis of the CECCY trial, so its findings were not pre-specified.We also included a small sample size due to difficulties obtaining speckle tracking images.We did not have a more robust modality for assessing EF such as planar equilibrium radionuclide angiocardiography or magnetic resonance imaging.Moreover, we noted a low incidence of cardiotoxicity that could have impaired the results.

Conclusion
LV-GLS is a potential predictor of ChC in patients with a low prevalence of cardiovascular comorbidities and risk factors for cardiovascular disease.LV-GLS was a better predictor of cardiotoxicity than LVEF, while the combination of LV-GLS and BNP during follow-up could be a predictor of cardiotoxicity.Overall, our findings confirm the ability of LV-GLS to detect subclinical cardiotoxicity and emphasize the need for early evaluations of LV-GLS to detect cardiotoxicity.

Figure 1 -
Figure 1 -Receiver operating characteristic curve analyses of LV-GLS measured before the onset of chemotherapy (A) and BNP after 2 weeks of chemotherapy (B) and their association with cardiotoxicity.AUC: area under the curve; BNP: brain-type natriuretic peptide; LV-GLS: left ventricular global longitudinal strain (Central Illustration).

Table 1 -Baseline characteristics of the study population.
Data are expressed as mean ± SD or number.LV-GLS: left ventricular global longitudinal strain; NS: not significant.Global Longitudinal Strain as a Predictor of Cardiotoxicity Arq Bras Cardiol: Imagem cardiovasc.2022;35(4):eabc340