ABC | Volume 110, Nº2, February 2018

Original Article Gripp et al Global Longitudinal Strain Accuracy for Cardiotoxicity Arq Bras Cardiol. 2018; 110(2):140-150 incidence of 10%, lower than that reported by those studies. That could be explained by the low morbidity profile of our population, composed only by patients with breast cancer, with similar treatment protocols. In the study by Baratta et al., 17 if the cardiotoxicity incidence would be calculated only among patients with breast cancer, a 12% rate would be found, similar to that of our population. Characteristics of the population that developed cardiotoxicity In our study, cardiotoxicity showed no statistically significant risk association with the clinical and anthropometric variables, histological type of tumor and treatment instituted. However, some variables evidenced clinically relevant information. The first was age, which was higher in the group that developed cardiotoxicity (mean of 56 years versus 49 years, in the group without cardiotoxicity), and could lead to a higher risk of events according to the literature. Another interesting variable, the total dose of anthracyclines and trastuzumab administered, which was lower in the group with cardiotoxicity, might be justified by the suspension or dose reduction of the antineoplastic drug by the oncology team in face of the drop in EF. Choosing the best time for doppler echocardiography There is no consensus between the European and American Societies of Cardiology about the time during the treatment in which the echocardiographies should be performed. In our population, two patients had cardiotoxicity on the third month. Analyzing retrospectively, if theDoppler echocardiography would be performed after each anthracycline cycle, the drop in LV GLS might have occurred before the reduction in EF on the thirdmonth. Therefore, Doppler echocardiographywould ideally be performed after the end of each anthracycline cycle. Marker of cardiotoxicity: 2D strain Ejection fraction is not considered a good predictor of cardiotoxicity, because it does not detect early myocardial contractile function changes. Recent studies have demonstrated that strain changes precede EF changes in patients undergoing antineoplastic treatment. 14,16,18-21 However, no consensus has been reached regarding the specific cutoff point of that variable that should be used as a predictor of cardiotoxicity. The results of our study confirm LV GLS as an excellent independent predictor of cardiotoxicity, which can be assessed by use of the data from Cox regression (p = 0.004, HR = 2.77; 95%CI: 1.39-5.54). None of the patients assessed showed the LV GLS drop after the EF drop. The LV GLS change occurred from the third month onward, while EF (Simpson’s method) changed only on the sixth month. There is no consensus in the literature regarding the LV GLS value that can predict cardiotoxicity. Some articles have mentioned that, using the Speckle Tracking technique, a 10% to 15% reduction could predict that outcome. The last European recommendation from 2016 states that a reduction > 15% could predict cardiotoxicity, while a reduction < 8% could exclude its diagnosis. However, there is a grey zone between those values. 12,22 Because of data inconsistency, our study aimed at finding the best cutoff point of the absolute value and percentage reduction of LV GLS to predict cardiotoxicity. The five events that occurred in our study enabled the construction of ROC curves to assess the diagnosis of cardiotoxicity on the sixth month. The need to define the ideal cutoff point of the GLS drop percentage capable of preventing cardiotoxicity has also been approached by some authors in recent years. According to Sawaya et al., 16 a 10% GLS drop on the third month of assessment could predict ventricular Figure 3 – ROC curves to assess the cutoff point of the absolute value of left ventricular global longitudinal strain (GLS) (A), and the cutoff point of the percentage reduction in left ventricular GLS (B) as predictors of cardiotoxicity. A B 1.0 1.0 1.0 1.0 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.0 0.0 0.0 0.0 Sensitivity Sensitivity 1 – Specificity 1 – Specificity AUC: 0.95 95%CI: 0.87-1.0 AUC: 0.97 95%CI: 0.92-1.0 ROC curve ROC curve 147