We thank Drs Zhang and Yang for their comments. In their letter, Drs Zhang and Yang raised the possibility that the cardiotoxicity seen in our cohort may have been predominantly due to the effect of anthracyclines rather than trastuzumab. They cited three studies; a meta-analysis of 6 randomized control trials by Antonis Valachis et al. which showed a low rate of severe cardiac toxicity in patients treated with trastuzumab alone ; a study by de Azambuja et al. which demonstrated that the cardiotoxicity induced by trastuzumab was reversible ; and a study by Adamo et al. which reported that trastuzumab-based treatments were well-tolerated in a small cohort of 51 elderly patients > 70 years old with adverse events that were deemed only mild to moderate in severity ; to support their argument that trastuzumab caused very limited cardiotoxicity.

In response we would like to highlight a number of points. Firstly, we agree with Drs Zhang and Yang that it is hard to differentiate the effect of anthracycline from trastuzumab since both agents can cause cardiotoxicity independently. Used in combination, both these agents have a synergistic effect. However, just as Drs Zhang and Yang highlighted studies supporting limited cardiotoxicity with trastuzumab, there is also evidence indicating the contrary. Telli et al report that up to 4% of patients enrolled in adjuvant trastuzumab trials experienced severe CHF during treatment and point out that despite early cessation of trastuzumab, a large number of patients in these trials experienced some form of cardiotoxicity that ultimately required discontinuation of trastuzumab. They also highlighted that approximately 14% of patients in the National Surgical Adjuvant Breast and Bowel Project Protocol B-31 (NSABP B-31) trial had trastuzumab discontinued because of asymptomatic decreases in left ventricular ejection fraction (LVEF). Follow-up cardiac evaluations of patients diagnosed with any degree of cardiotoxicity in the NSABP B-31 trial documented that a clinically significant proportion of patients had sustained decrements in their LVEF to less than 50%. There is also recent data from animal studies, which report microscopic irreversible damage to the myocardium from trastuzumab.

Secondly, many of the trials that report minimal long-term cardiotoxic effects of trastuzumab are comprised of breast cancer patients who were treated with anthracyclines prior to initiating therapy with trastuzumab. As a consequence, many of these studies reporting the incidence and prevalence of trastuzumab cardiotoxicity have inherent selection bias particularly since an estimated 2–8% of patients are excluded from clinical trials with trastuzumab due to the development of cardiomyopathy or cardiac symptoms after completing anthracycline therapy.

Finally, a more compounding and wider problem is that the reported incidence and prevalence of cardiotoxicity varies greatly as comprehensive analysis of published data on cardiotoxicity is difficult to perform and frequently inadequate because oncology trials vary widely in their methods and definitions of cardiotoxicity. Many oncology trials, including the three that were cited by Drs Zhang and Yang, define cardiotoxicity as a significant drop in left ventricular ejection fraction during treatment with chemotherapy or clinical symptoms of heart failure. Despite the widespread use and acceptance of LVEF as the parameter to monitor anthracycline-induced cardiotoxicity, there is no consensus as to what decrease in or absolute value of LVEF constitutes cardiotoxicity. Evidence also indicates that both clinical symptoms and LVEF lack sensitivity as measures of cardiotoxicity. LVEF is often preserved even though cardiomyocyte damage is proven on biopsy or biomarkers. Additionally, changes in LVEF have been hypothesized to only occur late in the progression of the disease, after compensatory mechanisms are exhausted. Hence, patients treated with trastuzumab alone (without anthracyclines), who do not meet the classic definition of cardiotoxicity based on a decrease in LVEF or clinical symptoms, may still have subclinical cardiotoxicity that is not detected by these moderately sensitive indices.

Studies employing newer technologies such as speckle tracking echocardiography (STE) to measure myocardial deformation (strain) and deformation rate (strain rate) have demonstrated the feasibility of early detection of subclinical LV systolic dysfunction during chemotherapy by detecting abnormalities of regional function that do not result in decreased global LV ejection fraction. The decreases in longitudinal strain and strain rate seen in our cohort at the end of treatment, whether they are due to anthracyclines, trastuzumab or a combination of both, were consistent with other studies of patients treated with anthracyclines and trastuzumab. We also report that these decreases persisted two years after cessation of treatment in our cohort. The significance of the sustained decreases in strain and strain rate observed in our study even after a significant time following cessation of treatment, and significant changes in LV dimensions (though small), which have not previously been reported in the adult population, suggest persistent damage from chemotherapy. Ultimately, the value of changes in echo parameters of LV systolic function in predicting subsequent symptomatic heart failure and the significance of this lack of recovery in all three parameters in predicting future heart failure is still unknown due to a paucity of studies. We agree with Drs Zhang and Yang that larger multi-center studies are clearly needed to test the impact of these decreases in LV function parameters on symptoms, morbidity and mortality.