Variations of circulating cardiac biomarkers during and after anthracycline-containing chemotherapy in breast cancer patients

The cancer burden is a worldwide major public health problem. Fortunately, the outcome, including the cancer death rate, can be improved by earlier diagnosis and better treatment [1].

Because more patients are cured, the attention is now focusing on quality of life and long-term outcome of cancer survivors. Cardiovascular disease is the leading cause of late mortality among survivors of childhood and adolescent cancer. The risk of cardiovascular death is higher than the actual risk of cancer recurrence in many adult cancer patients in complete remission. Cancer survivors have a ten-fold higher mortality than the general population, with a fifteen-fold increased risk of developing a congestive heart failure (CHF) and a ten-fold increased risk of coronary artery disease [2, 3]. These patients also have a higher risk of atherosclerosis, hypertension, pericardial disease, valvular heart disease and dyslipidemia [4, 5]. The higher cardiovascular death rate in cancer survivors is secondary to a combination of cancer treatment-related risk effects (ionizing irradiation, cytotoxic and targeted agents), familiar risk factors and health behavior. Consequently, cancer survivors need appropriate surveillance in order to early detect long-term side effects of cancer therapy, allowing appropriate treatment before the toxicity becomes irreversible [4, 5].

In this respect, easily accessible circulating biomarkers could be seen as highly valuable diagnostic tools for early detection of cardiotoxicity related to cancer treatments.

The cardiac-specific isoenzymes of troponins T and I (cTnT, cTnI) are released into the blood when cardiomyocytes are damaged. Troponin levels rapidly increase after chemotherapy and might predict late cardiac events [6]. However, troponins release is related to lysis of myocardial cells, which may be initially absent in case of type II heart damage chemotherapy. In addition, their half-life is relatively short (2 h) [7], requiring the collection of multiple blood samples during the treatment, which is hardly feasible during outpatient’s treatments.

N-terminal brain natriuretic peptides (NT-proBNP) are released by left ventricular cardiomyocytes in response to wall stress. This biomarker has well-established clinical utility in CHF. NT-proBNP elevation during chemotherapy in breast cancer patients has been related to asymptomatic decline in left ventricular ejection fraction (LVEF) [8‐11]. Nevertheless, studies evaluating the predictive role of NT-proBNP in the detection of chemotherapy-related cardiac dysfunction (CRCD) gave conflicting results and the threshold of positive tests has not been determined yet [12, 13].

Soluble ST2 (sST2), a member of the interleukin-1 receptor family, is a more recently discovered biomarker of cardiovascular stress. Two studies have demonstrated that high levels of sST2 may be a strong predictor of cardiovascular death in CHF patients [14, 15]. Furthermore, sST2 has a higher discrimination power than NT-proBNP in CHF patients [16], and in community-based populations free of cardiovascular disease [17].

MicroRNAs (miRNAs) are approximately 22-nucleotide long RNAs that regulate gene expression by binding to and consequently silencing target messenger RNAs. They are involved in multiple biological processes including cell proliferation, differentiation and apoptosis. All cell types release miRNAs in peripheral blood under both normal and pathological conditions. Therefore, circulating miRNAs are promising biomarkers for the early and minimally invasive diagnosis of cancer and its treatment-related cellular toxicity [18‐24].

The use of anthracyclines in the treatment of breast cancer is limited by dose-dependent cardiotoxicity, which eventually may lead to CHF. The aim of the present study is to evaluate the variations of innovative biomarkers during and after anthracycline-containing chemotherapy. Plasma levels of cTnT, NT-proBNP, sST2 and 10 selected miRNAs were measured in a total of 45 breast cancer patients receiving anthracyclines as part of their neoadjuvant chemotherapy (NAC). The neoadjuvant setting was selected because patients were treatment-naïve, so that markers levels were not modified by other therapies such as surgery or radiotherapy.

cTnT and NT-proBNP are both important biomarkers in heart diseases. Our study finds a significant increase in these biomarkers at the end of the NAC for breast cancer, and potentially identifies a group of patients at risk of CRCD.

Compared with cTnT and NT-proBNP, sST2 levels increase in a higher percentage of patients directly after anthracyclines-based chemotherapy (NA2 time point). As sST2 is strongly associated with CHF severity and outcome [14, 15], its blood concentration during chemotherapy may be an important predictor of long-term CRCD. Nonetheless, sST2 lacks tissue specificity and its levels could be elevated in case of breast cancer [36, 37]. In fact, Lu et al. previously reported elevated sST2 levels in the serum of breast cancer patients, with a decrease after tumor surgery [36]. All our patients experienced a complete - or at least a partial - pathologic response to the NAC, and were free of disease after 3 months. If tumor cells indeed secreted plasma sST2, we would expect a progressive decrease in its levels as the tumor responds or after the surgery. On the contrary, plasma levels of sST2 remain significantly higher, compared with baseline, after the chemotherapy and even 3 months after breast surgery (D8 and 3 M time points, Fig. 2). Apoptotic cancer cells could also have released plasma sST2. If this was the case, we should detect a correlation between sST2 upregulation and tumor response to chemotherapy, but such a correlation was not found. Based on these results, we think that sST2 elevation is a consequence of the chemotherapy rather than a reflection of the tumor presence and/or response.

Among selected miRNAs, we demonstrated the increase of 3 miRNAs related to the diagnosis and prognosis of CHF: miR-126-3p, miR-199a-3p, miR-423-5p. Vascular endothelium-enriched miR-126 has been associated with coronary artery disease and CHF. Reduced levels were observed in the acute phase [26, 28, 38], followed by normalization after clinical improvement [26, 28]. CRCD does not necessarily imply tissue ischemia [18] and an increase, rather than a decrease, of miR-126-3p concentrations was found after the chemotherapy. As miR-126-3p is enriched in endothelial cells and promotes blood vessels formation [39], its upregulation could be a response to cellular stress and anti-angiogenic activity of the chemotherapy [40]. In the same way, miR-199a-3p levels also decreased in acute heart failure [41], while our results have shown an elevation after chemotherapy. miR-199a-3p is largely expressed in cardiomyocytes and its myocardial levels are upregulated in hypertrophic hearts [42], which is sometimes observed in CRCD [18]. After myocardial infarction in mice, miR-199a promotes cardiomyocytes regeneration and recovery of cardiac functional parameters [43]. Cardiomyocytes could therefore release miR-199a-3p under the effect of the chemotherapy to play a cardio-protective role, though this hypothesis requires further explorations. Plasma levels of miR-423-5p were highly increased immediately after anthracyclines (NA2 time point), especially in patients with decreased LVEF (Fig. 4). Two recent studies have unveiled that miR-423-5p was enriched in the blood of patients suffering from CHF, with a high diagnosis power and a significant correlation with CHF prognostic markers [27, 44]. In another study, circulating miR-423-5p levels were able to predict long-term mortality of CHF patients [45]. Based on these reports and our findings, we believe that the diagnostic, prognostic and predictive roles of miR-423-5p in CRCD should be further explored.

Circulating miR-34a-5p was also studied for multiple reasons. Boon et al. demonstrated that miR-34a was involved in the alteration of cardiac contractile function after AMI, by inducing telomere attrition [29]. Desai et al. evaluated miR-34a expression in myocardial tissues of mouse exposed to increasing doses of doxorubicin. Unlike troponins, upregulation of miR-34a was an early event inside the cardiac tissue and did not involve prior necrosis of cardiomyocytes [20]. Our study indicates a strong increase in miR-34a plasma levels immediately after anthracycline-based chemotherapy, followed by a gradual decrease. The tumor suppressor p53 is known to promote both growth arrest and apoptosis upon DNA damage [46]. miR-34a is directly induced by p53 to exert anti-tumor functions [47]. This might explain the miR-34a upregulation observed after anthracyclines, which cause DNA breaks and p53 activation [48]. The subsequent tubulin-binding agent based chemotherapy does not act via DNA breaks but by the disruption of microtubule function. Plasma miR-34a increase could therefore be specific of anthracyclines chemotherapy, explaining the gradual decrease of miR-34a levels after this part of the treatment. The concentration of miR-34a-5p may help for the identification of patients at risk of developing CRCD, however, the threshold of a positive test is still unknown. One problem may be the significant elevation in a high percentage of patients although most of them will not develop any cardiotoxicity. Furthermore, miR-34a is broadly expressed in normal tissues and this miRNA may thus lack specificity. The increase of miR-34a levels after anthracyclines was not correlated with the increase of troponins. This observation is in contradiction with our previous study that was performed on a smaller number of patients [23], probably because of the limited statistical power. Globally, the elevation of the other markers (cTnT, NT-proBNP, sST2) was poorly correlated with that of CHF-related miRNAs, which could imply that these molecules provide distinct information about CRCD.

Interpretation of our results is limited for several reasons. Firstly, the small number of patients, with only one who developed a clinical CRCD, does not allow definitive conclusions to be drawn. In addition, majority of patients experienced an asymptomatic decline in LVEF in the context of anti-HER2 therapy, and we know that the cardiotoxicity mechanism is different from that of anthracyclines [18]. As cardiomyopathy is a late side effect of the chemotherapy, a long-term follow-up is required. The biomarkers we identified (sST2, miR-126-3p, miR-199a-3p, miR-423-5p and miR-34a-5p) should therefore be studied in a larger prospective trial with regular and prolonged cardiac monitoring. These biomarkers may be useful either as a predictive marker of cardiotoxicity at the time of treatment or as a new tool for the early identification of late side effects before the patients have clinical symptoms. Importantly, appropriate treatment of CRCD may prevent irreversible consequences if initiated early.

We identified sST2, miR-126-3p, miR-199a-3p, miR-423-5p and miR-34a-5p as innovative biomarkers for potential early and sensitive detection of the cardiomyopathy associated with anthracycline-based breast cancer chemotherapy.