Two-dimensional speckle tracking echocardiography predicts early subclinical cardiotoxicity associated with anthracycline-trastuzumab chemotherapy in patients with breast cancer

For this retrospective study, we enrolled women newly diagnosed with breast cancer between December 1, 2004, and June 1, 2012, who were treated with an anthracycline (doxorubicin and epirubicin), or a HER 2 inhibitor (trastuzumab), or both. Though the majority of the study population (40/66; 61%) received adjuvant therapy, a substantial number (26/66; 39%); received neoadjuvant therapy. Of the total cohort, one woman was found to have stage IV disease at diagnosis. Twenty three (35%) had stage III disease and 6 of those had inflammatory breast cancer. None of the study population received concurrent anthracyclines and trastuzumab. The average age was 52 (±9), mean anthracycline dose of 252 (±45) mg/m² and trastuzumab was administered for an average of 11 (±2.7) months in 3 week cycles and a mean dose of 5.4 (±2.7) gr. All patients had to have a normal pretreatment (T0) LV ejection fraction (LVEF), had to have been followed up for 1 year, as well as an echocardiogram at baseline. Patients who met the inclusion criteria and had images of adequate quality, as well as follow-up, were included (N = 66). Because of lack of consensus, we followed the definition of cardiotoxicity used in the clinical trials for combined chemotherapy in HER 2 positive patients, where it was defined as an absolute decrease in LVEF of 10 or more percentage points from baseline echocardiogram [24, 25]. Traditional cardiovascular risk factors, such as age, hypertension, diabetes mellitus, hyperlipidemia, family history of premature coronary artery disease, and smoking status were also considered in the analyses [26‐28]. The study was approved by the Mayo Clinic Institutional Review Board, and written informed consent was obtained.

To be included, patients needed to have at least 3 standard echocardiographic examinations: at baseline (T0); from the start of chemotherapy to first echocardiogram (T1) and from the start of chemotherapy to the second echocardiogram (T2) following the cardio oncology clinic protocol [19]. Echocardiographic examinations were performed using a GE Vivid 7 system (General Electric Company) with an M4S transducer (1.5-4.3 MHz). Mean frame rates were 55 Hz for grayscale imaging. Studies were performed and reported according to the guidelines of the American Society of Echocardiography [29]. LVEF was calculated using the biplane Simpson method [13]. Digital images were saved for subsequent, blinded off-line analysis using the Syngo Velocity Vector Imaging software, version 3.5 and 2D-STE analyses were performed.

For all patients, the region of interest analyzed was adjusted to cover at least 90% of the myocardial wall thickness for myocardial strain and SR. LV longitudinal parameters were measured from the apical 4-chamber, 2-chamber, and 3-chamber views, and the myocardium was divided into 6 segments per view. Care was taken to ensure that the long axis of the ventricle was perpendicular to the plane of the mitral annulus in the LV apical views. Circumferential and radial parameters were measured using the parasternal short-axis plane at the midventricular level. RV longitudinal parameters were measured from the apical 4-chamber view. Global and segmental myocardial deformation parameters, including strain (S), peak systolic strain rate (SRs), and peak early diastolic strain rate (SRe) were measured for each patient.

To determine intraobserver variability, echocardiograms from 20 randomly assigned patients were reanalyzed by the same observer (X.H.) 2 months after the initial analysis. For interobserver variability, the same patients and the same cardiac cycles were analyzed by a second observer (Z.Y.).

Continuous data are presented as mean (SD) and categorical data as frequencies (percentages). Deformation parameters were compared among T0, T1, and T2 using 1-way ANOVA and paired t tests. Differences among age subgroups were assessed using the Tukey-Kramer multiple comparisons test. The first(T1) and second time point(T2) were used to construct a receiver operating characteristic (ROC) curve, which was used to predict cardiotoxicity. The best cutoff value was defined as the point with the highest sum of sensitivity and specificity. Univariate and multivariate logistic regression analyses were used to determine predictors of a significant decrease in LVEF. Intraobserver variability, interobserver variability, and intraclass correlation coefficients (ICCs) with 95% CIs were calculated to evaluate test reliability [30, 31]. All statistical analyses were performed using SAS software, version 9.3 (SAS Institute Inc). A P value <.05 was considered statistically significant.

Sixty-six patients who completed anthracycline-trastuzumab treatment for breast cancer were included in the study (mean [±SD] age, 52 [±9] years; median [range] age, 51 [34-72] years). The patients’ clinical characteristics are summarized in Table 1. Of the 66 patients, 13 (20%) had cardiotoxicity defined by a decrease in LVEF of 10 or more percentage points from baseline [24, 25]. Forty-six percent of those patients developed cardiotoxicity at T1 and 54% at T2. Both groups were followed up for the same amount of time. The median cumulative doses of anthracycline-trastuzumab are shown in Table 1. The patients LVEF in whom cardiotoxicity developed vs. patients LVEF with no cardiotoxicity is shown in Table 2. Thirty-one patients had baseline cardiac risk factors, including hypertension, 13 patients; diabetes mellitus, 3 patients; hyperlipidemia, 12 patients; family history of premature coronary artery disease, 4 patients; and smoking history, 13 patients. There was no difference in cardiotoxicity for patients with more than or less than 3 risk factors. In addition, 50 patients (76%) began radiotherapy 5.3 (2.2) months after the start of chemotherapy.

Echocardiographic follow up was obtained in three time points: baseline (T0) at the time of diagnosis; (T1) from the start of chemotherapy to first echocardiogram (median [interquartile range {IQR}], 2.25 [1.84-2.99] months); and (T2) from the start of chemotherapy to the second echocardiogram (T2) (median [IQR], 5.44 [4.61-6.47] months).

Serial 2D-STE parameters at T0, T1, and T2 are summarized in Fig. 1. Compared with T0, GLS at T1 and T2 and GCS at T1 and T2 were significantly reduced (P < .01 for all). Also significantly reduced were RV GLS at T1, SRs at T1, RV GLS at T2, SRs at T2 and SRe at T1 (P < .001 for all). There was a significant decrease with increasing age in LV GLS both at T1 and T2, in LV longitudinal SRe at T2, and in RV longitudinal SRe at T1 (P < .01 for all). There was no significant decrease in any parameter measured with increasing age (Table 3).

Combining both RV GLS at T1 and LV GLS was the strongest predictor of cardiotoxicity (area under the curve [AUC], 0.91; sensitivity, 100%; specificity, 73%; P < .001). LV GLS at T1 (AUC, 0.85; cutoff, − 14.06; sensitivity, 91%; specificity, 83%; P = .003) was also a strong indicator of subsequent cardiotoxicity. Combining LV GLS with longitudinal early diastolic strain rate (LSRe) at T1 (AUC, 0.90; sensitivity, 91%; specificity, 86%; P < .001) and combining GLS with radial early diastolic strain rate (RSRe) at T1 (AUC, 0.88; sensitivity, 91%; specificity, 78%; P < .001) were also strong predictors of subsequent cardiotoxicity (Table 4).

The intraobserver and interobserver agreement are shown in Table 5. Both measurements of agreement were lowest for radial S, SRs, and SRe values.

This study was limited by its retrospective design and small sample size. However, it was able to show that RV mechanical parameters should be studied in a larger population. Our data shows increased risk of cardio toxicity in the older population, however there is not enough power when the data is divided into age groups to draw a definitive conclusion.