Longitudinal rotation: a new way to detect the cardiotoxicity of anthracycline-based chemotherapy in breast cancer patients

Background and aims The study was to compare cardiac parameters before and after anthracycline-based chemotherapy and identify a parameter for detecting cardiotoxicity in breast cancer patients. Methods Cardiac function in 43 female breast cancer patients was evaluated at three time points: baseline, 1-3 days before the initiation of anthracycline-based chemotherapy; 3 weeks and 6 months after the final cycle of chemotherapy. At each visit, the peak longitudinal velocity; strain rate; peak systolic strain; peak systolic longitudinal displacement, and segmental and global longitudinal rotation degrees of the left ventricular were measured. Results The peak early-diastole left ventricular wall velocity at baseline was significantly higher than the values at 3 weeks and 6 months after the final cycle of chemotherapy. The absolute value of the lateral wall peak systolic longitudinal rotation degrees was significantly higher at baseline than at 3 weeks and 6 months after the final cycle of chemotherapy, whereas the absolute value of the global peak systolic longitudinal rotation degrees at baseline was significantly lower than the values at 3 weeks and 6 months after the final cycle of chemotherapy. None of the measured parameters differed significantly between the 3 weeks and 6 months after the final cycle of chemotherapy. Conclusions Cardiac diastolic and systolic dysfunction was found after anthracycline-based chemotherapy in this study, and the peak systolic longitudinal rotation degrees can be used to detect dysfunction after chemotherapy. The cardiotoxicity of epirubicin-based chemotherapy is stronger than that of therarubicin-based chemotherapy.


INTRODUCTION
Breast cancer is one of the most common malignant tumors, and now it is becoming a leading cause of cancerrelated death in women worldwide [1][2]. The treatment for breast cancer involves a combination of surgery, chemotherapy, and radiotherapy [3]. Over the past decades, the mortality rate among breast cancer patients has decreased. However, cardiotoxicity induced by the chemotherapy has become a leading cause of morbidity and mortality in survivors [4][5]. To prevent heart failure caused by cancer therapy, consideration of potential cardiotoxicity is very important.
Epirubicin and therarubicin, as common anthracyclines, are widely used chemotherapy drugs for reducing the recurrence of breast cancer in China. However, the use of these drugs is limited by their cardiotoxicity. The cardiotoxicity of anthracycline-based chemotherapy may be associated with the cumulative dose, other associated cytotoxic drugs, or the advanced age of patients [6][7]. Previous studies found that the cardiotoxicity can appear during or after the chemotherapy www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 41), pp: 70072-70083 Research Paper www.impactjournals.com/oncotarget in breast cancer patients [6,8]. Left ventricular ejection fraction (LVEF) measurement by two-dimensional echocardiography is the most commonly used method for detecting cardiotoxicity, which is most commonly defined as a ≥5% or ≥10% reduction in the LVEF from baseline to an LVEF <55% in symptomatic patients and asymptomatic patients, respectively [9]. However, the use of LVEF has limitations for early detection of cardiotoxicity because it cannot reflect the early dysfunction of the heart [5].
With the development of various imaging techniques, such as tissue Doppler imaging (TDI), speckle tracking echocardiography (STE, two-dimensional or three-dimensional), and cardiac magnetic resonance imaging (CMRI), the cardiotoxicity of anthracyclinebased chemotherapy can be detected with greater convenience. However, the angle dependency of TDI makes these results difficult to reproduce [10]. Although CMRI is considered the gold standard for evaluating LV volumes and EF, its lack of availability and high cost limit its routine use [11][12]. STE is an ideal method for detecting cardiotoxicity caused by chemotherapy, and previous studies have demonstrated that STE is feasible and convenient for early detection of cardiac dysfunction [13][14][15].
The aim of the present study was to identify a parameter for reliable detection of cardiac dysfunction via STE after chemotherapy in breast cancer patients. We hypothesed that longitudinal rotation (LR) motion could be detected in the hearts of breast cancer patients after chemotherapy, then tested the hypothesis by measuring the global peak systolic LR (PSLR) degrees in breast cancer patients after anthracycline-based chemotherapy. Cardiac dysfunction was then evaluated to confirm whether LR could serve as a new way to detect cardiac dysfunction after chemotherapy in breast cancer patients.

General characteristics of the patients
The general characteristics of the study group are listed in Table 1. The cumulative dose of epirubicin given was significantly higher than that of therarubicin.

Conventional two-dimensional echocardiographic parameters of the study groups
The conventional two-dimensional echocardiographic parameters of the study groups at the three time points are listed in Table 2. There were no significant differences among the study groups at the three separate time points, except that the e' sep at baseline was higher than at 3 weeks and 6 months after the final cycle of chemotherapy. Although there were no significant differences among E, E/A, and e' lat between the groups, the baseline values were higher than those at 3 weeks and 6 months after the final cycle of chemotherapy.

STE parameters
The STE parameters of the study groups at the three time points are listed in Table 2. The peak earlydiastole LV wall velocity at baseline was significantly higher than at 3 weeks and 6 months after the final cycle of chemotherapy ( Figure 1). Although there were no significant differences among the peak systolic longitudinal strains of the different myocardial layers, the values at baseline were higher than those at 3 weeks and 6 months after the final cycle of chemotherapy (Figure 2A, 2B and 2C). There were no differences among the peak longitudinal strain rates and displacements.

Segmental and global PSLR
The segmental and global PSLR values of the study groups at the three time points are shown in Table 2. In the apical 4-chamber view, the lateral wall rotated counterclockwise, whereas the septal wall rotated clockwise. The global PSLR at baseline was small, and significant differences were detected between the lateral wall and global PSLR values at baseline and those at 3 weeks and 6 months after the final cycle of chemotherapy ( Figure 2D, 2E and 2F). The absolute value of the lateral wall PSLR at baseline was higher than the values at 3 weeks and 6 months after the final cycle of chemotherapy ( Figure 3A), whereas the absolute value of the global PSLR at baseline was lower than the values at 3 weeks and 6 months after the final cycle of chemotherapy ( Figure 3B). The values of the lateral wall and global PSLR did not change significantly from 3 weeks to 6 months after the final cycle of chemotherapy.

Correlations between the global PSLR and LVEF, E, A, E/A, e' sep , a' sep , e' lat , a' lat , E/ e' sep , E/ e' lat and cumulative dose of anthracycline
The results of correlation tests are listed in Table 3. At baseline, no correlation was found between the global PSLR and LVEF, E, A, E/A, e' sep , a' sep , e' lat , a' lat , E/ e' sep and E/ e' lat .However, at 3 weeks after the final cycle of chemotherapy, the global PSLR was positively correlated with the e' lat . Breast cancer patients with a higher e' lat were more likely to have a higher global PSLR ( Figure 4A). At 6 months after the final cycle of chemotherapy, the global PSLR was positively correlated with e' sep and e' lat , and negatively correlated with E/e' sep and E/e' lat . Breast cancer patients with a higher e' sep and e' lat and lower E/e' sep and E/e' lat were more likely to have a higher global PSLR ( Figure 4B, 4C, 4D and 4E). www.impactjournals.com/oncotarget

ROC analysis for detecting the accuracy of PSLR for detecting cardiotoxicity in breast cancer patients
The areas under ROC curves (AUCs) were measured to determine the cut-off values for the global PSLR with the optimal sensitivity, specificity, and accuracy for assessing cardiotoxicity with regard to LV function. The AUC values at 3 weeks and 6 months after the final cycle of chemotherapy in breast cancer patients were 0.649 and 0.665, respectively. The optimal cut-off values for global PSLR at 3 weeks and 6 months after the final cycle of chemotherapy were -2.58° and -1.50°, respectively. The sensitivity and specificity for the global PSLR at 3 weeks after the final cycle of chemotherapy were 55.8% and 72.1%, respectively, and the corresponding values at 6 months after the final cycle of chemotherapy were 67.4% and 60.5%, respectively ( Figure 4F and 4G).

Differing effects of epirubicin-based and therarubicin-based chemotherapy
The STE parameters of patients who received epirubicin-based chemotherapy are compared to those of patients who received therarubicin-based chemotherapy in Table 4. In the group that received epirubicin-based chemotherapy, the peak early-diastole LV wall velocity, and lateral wall and global PSLR differed significantly The values at baseline compared with those at 3 weeks after the final cycle of chemotherapy, *P<0.05. IVSd: interventricular septal thickness in end-diastolic period, LVDd: left ventricular diameter in end-diastolic period, LVPWd: left ventricular posterior wall thickness in end-diastolic period, LVSd: left ventricular diameter in end-diastolic period, LVEDV: left ventricular end-diastolic volume, LVESV: left ventricular end-systolic volume, LVEF: left ventricular ejection fraction, E: peak velocity during early diastole of anterior mitral leaflet, A: peak velocity during late diastole of anterior mitral leaflet, e' Sep : peak early diastolic annular velocities at septum positions using TDI, a' Sep : peak late diastolic annular velocities at septum positions using TDI, e' lat : peak early diastolic annular velocities at lateral positions using TDI, a' lat : peak late diastolic annular velocities at lateral positions using TDI.

4-CH: Apical 4-chamber view
The values at baseline compared with those at 3 weeks after the final cycle of chemotherapy, *P<0.05.
The values at baseline compared with those at 6 weeks after the final cycle of chemotherapy, #P<0.05.
"+" indicates counter-clockwise rotation and "-" indicates clockwise rotation. among the three time points; however, these values did not change significantly over time after therarubicinbased chemotherapy. The absolute value of the peak early-diastole LV wall velocity and the lateral wall PSLR at baseline were higher than those at 3 weeks and 6 months after the final cycle of chemotherapy, whereas the absolute value of global PSLR at baseline were lower than the values at 3 weeks and 6 months after the final  cycle of chemotherapy in both the epirubicin-based and therarubicin-based chemotherapy groups. The values of these parameters did not differ significantly between 3 weeks and 6 months after the final cycle of chemotherapy ( Figure 5A, 5B and 5C).

Reproducibility and repeatability
The results for the intraobserver and interobserver variabilities for the global PSLR upon repeated measurements in all study patients are shown in Table 5.

DISCUSSION
As is well known, chemotherapy can reduce the recurrence of breast cancer, especially when used in combination with anthracyclines. However, the cardiotoxicity of anthracyclines is becoming a major concern. Early detection of cardiotoxicity is becoming increasingly important. The main findings of this study were: ① cardiac diastolic and systolic dysfunction was found in breast cancer patients after anthracycline-based chemotherapy; ② LR motion is a new parameter for detecting cardiotoxicity at 3 weeks and 6 months after the final cycle of chemotherapy; and ③ subgroup analysis indicated that the cardiotoxicity of epirubicin-based chemotherapy maybe stronger than that of therarubicinbased chemotherapy.
In clinical practice, LVEF is quantified as a measure of cardiac function. In the present study, we found that there were no significant differences in LVEF among the three time points. Thus, LVEF cannot be used in clinical practice for early detection of cardiotoxicity. With its angle independency and high reproducibility, STE is widely used for detection of cardiac function [16][17][18]. Ho et al [19] used STE to examine the longterm effects of standard chemotherapy on myocardial function in asymptomatic breast cancer survivors and found that the E/A ratio, global E', and global longitudinal 2D strain were reduced in comparison The absolute value of the lateral wall PSLR was higher at baseline than at 3 weeks and 6 months after the final cycle of chemotherapy, while the absolute value of the global PSLR at baseline was lower than at 3 weeks and 6 months after the final cycle of chemotherapy. www.impactjournals.com/oncotarget with those of controls. Fallah-Rad et al [3] used cardiac biomarkers, tissue velocity (TVI), strain imaging, and CMRI to predict early LV dysfunction in breast cancer patients treated with adjuvant trastuzumab therapy. They found that the peak global longitudinal and radial strain decreased early in the group with trastuzumabmediated cardiotoxicity. Portugal et al [16] used speckle tracking imaging to determine the relationship between global and regional longitudinal strain (GLS) and chemotherapy-induced cardiotoxicity in breast cancer patients and concluded that impaired GLS in patients with preserved LVEF was independently associated with an increased incidence of chemotherapy-induced cardiotoxicity. Our results were consistent with those of this previous research. Cardiac diastolic and systolic dysfunction was found after administration of anthracycline-based chemotherapy, based on lower values of e' Sep , peak early-diastolic LV wall velocity, and longitudinal strain in different myocardial layers at 3 weeks and 6 months after the final cycle of chemotherapy in comparison with the baseline values. The reasons for these results may be related to LV wall  PSLR (B, C, D, E). ROC analysis for determining the accuracy of PSLR for identifying cardiac dysfunction in breast cancer patients. The sensitivity and specificity for a PSLR cut-off value of -2.58° at stress and local differences in the activation of signaling pathways involved in apoptosis or fibrosis.
LR, was first reported by Popović [20], represents the rotational motion in the long axis of the heart. In ischemic cardiomyopathy, dilated cardiomyopathy, and hypertension patients, as well as in acute myocardial ischemia animals, LR motion in the heart has been reported [21][22][23]. A normal myocardial fiber consists of longitudinal and circumferential fibers. Subendocardial and subepicardial myocardium are mainly composed of longitudinal fibers, and the middle layer is composed of circumferential fibers [24]. During systole and diastole, two motions are produced: short and long axis motions. In systole, when viewed from the apex, the apex rotates counter-clockwise, while the base rotates clockwise. This twisting motion of the left ventricle can be described as ''the wringing of a linen cloth to squeeze out water'' [15]. When viewed from its long axis, in systole, shortening of the long axis and thickening of the walls can be observed.
Previous studies indicated that the degree of PSLR in a normal heart is small, and our results agree with this [21][22][23]. The global PSLR at baseline in our study was -0.23±3.26°. At 3 weeks and 6 months after the final cycle of chemotherapy, the absolute values were higher than that at baseline. However, the value of the lateral wall PSLR at baseline was lower than those at 3 weeks and 6 months after the final cycle of chemotherapy. These results may demonstrate systolic dysfunction after anthracycline-based chemotherapy. The reason for such LR may be related to changes in the fiber arrangement by 3 weeks and 6 months after the final cycle of chemotherapy. Anthracyclines induce the formation of reactive oxygen species, which in excess can result in DNA and cytosolic damage, with subsequent loss of cellular integrity and apoptosis, inducing disarray and imbalance of the myocardial fibers [25].
In the present study, no correlations were found between the baseline values of global PSLR and LVEF, E, A, E/A, e' sep , a' sep , e' lat , a' lat , E/e' sep , E/e' lat and cumulative dose of anthracycline. At 3 weeks after the final cycle of chemotherapy, breast patients with a higher e' lat were more likely to have a higher global PSLR. At 6 months after the final cycle of chemotherapy, breast patients with higher e' sep and e' lat and lower E/e' sep and E/e' lat values were more likely to have a higher global PSLR. These findings indicated that although the LV filling pressure did not differ among the three time points, a lower LV filling pressure may correspond to a higher global PSLR.
The AUC values for PSLR at 3 weeks and 6 months after the final cycle of chemotherapy in breast cancer patients were 0.649 and 0.665, respectively. The sensitivity and specificity values for a PSLR cut-off value of -2.58° at 3 weeks after the final cycle of chemotherapy were 55.8% and 72.1%, respectively, and those for a PSLR cut-off value of -1.50° at 6 months after the final cycle of chemotherapy were 67.4% and 60.5%, respectively. From these results, we found that use of the global PSLR to detect cardiotoxicity after anthracycline-based chemotherapy was reliable.
When we compared parameters of cardiac function between the patients who received epirubicin-based and therarubicin-based chemotherapy, we found that after epirubicin-based chemotherapy, the peak early-diastole LV wall velocity and lateral wall and global PSLR differed significantly among the three time points. However, no differences between the time points were observed in patients who received therarubicin-based chemotherapy.
The absolute values of peak early-diastole LV wall velocity and the lateral wall PSLR at baseline were higher than those at 3 weeks and 6 months after the final cycle of chemotherapy, whereas the absolute value of global PSLR at baseline was lower than those at 3 weeks and 6 months after the final cycle of chemotherapy, in both the epirubicin-based and therarubicin-based chemotherapy groups. From this, we found that the cardiotoxicity of epirubicin-based chemotherapy may be stronger than that of therarubicin-based chemotherapy. Finally, the lack of significant differences in the values of cardiac function parameters between 3 weeks and 6 months after the final cycle of chemotherapy illustrated Figure 5: Differences in peak early-diastole LV wall velocity (A) and lateral wall (B) and global PSLR (C) after epirubicinbased chemotherapy versus therarubicin-based chemotherapy at baseline and 3 weeks and 6 months after the final cycle of chemotherapy. The absolute values of the peak early-diastole LV wall velocity and the lateral wall PSLR at baseline were higher than those at 3 weeks and 6 months after the final cycle of chemotherapy, whereas the absolute value of global PSLR at baseline was lower than at 3 weeks and 6 months after the final cycle of chemotherapy in both epirubicin-based and therarubicin-based chemotherapy groups. that the anthracycline-induced cardiotoxicity is irreversible. But to state this surely, more breast cancer patients should be increased and also it needs longer follow up.

Study sample
This study was approved by the ethics committee of ChangZhou No. 2 People's Hospital, affiliated to NanJing Medical University. All patients provided consent for participation in the study.
Forty-three female breast cancer patients were enrolled. The inclusion criteria were: ① treatment with epirubicin or therarubicin; and ② LVEF >50%. None of the patients had a history of cardiovascular disease, hypertension, or diabetes mellitus, all of them were asymptomatic before the chemotherapy; and ③ All enrolled breast cancer patients did not perform radiotherapy.
In total, the patients were evaluated at three separate time points: ① baseline: 1-3 days before the initiation of anthracycline-based chemotherapy; ② 3 weeks after the final cycle of chemotherapy; and ③ 6 months after the final cycle of chemotherapy.

Conventional 2D echocardiography
At each visit, all patients underwent conventional 2D echocardiography (Vivid E9, GE Healthcare). Interventricular septum thickness at end-diastole (IVSd), LV posterior wall thickness in end-diastole (LVPWd), and LV diameter at end-diastole (LVDd) and end-systole (LVSd) were measured in the parasternal long axis view of the LV by M-mode. Biplane Simpson's method was used to measure the LV end-diastolic and systolic volume (LVEDV and LVESV) for calculation of the LVEF. The peak early and late diastolic mitral annular velocities (E and A, respectively) were measured by pulsed-wave Doppler, and the ratio of E/A was calculated. The peak early (e') and late (a') diastolic annular velocities were obtained at the septum (e' sep ) and lateral (e' lat ) positions using TDI, and E/ e' sep and E/ e' lat were measured.
In each group, ECG leads were connected to the patients. For offline analysis, the standard high frame rate (>36 /s) of the apical 3-, 4-, and 2-chamber views of three consecutive cycles were measured while patients held their breath.

Analysis of LV function
The apical 3-, 4-, and 2-chamber views were analyzed using 2D-STE software (2D-Strain, EchoPac PC 113, GE Healthcare, Horten, Norway) by one experienced cardiologist. The peak systolic and diastolic longitudinal velocities and strain rates as well as the peak systolic longitudinal strain of three different layers and displacement of LV were measured.
We defined LR as the global rotation of the LV cross section. Using the SAX-MV option of the Echopac in the apical 4-chamber view, the subendocardial layer was displayed. The software automatically created a region of interest (ROI) that included the subendocardial, midmyocardial, and subepicardial layers. The LV region was divided into five segments: base-septal, middle-septal, apex, middle-lateral, and base-lateral. The segmental and global PSLR of the LV was assessed via the software.