Defining Cancer Treatment-Related Cardiac Dysfunction (CTRCD)
The condition that we are seeking to avoid is the development of HF in cancer survivors. While this may occur acutely, it is more commonly detected years after chemotherapy. Of course, awaiting the clinical presentation of HF is not a prudent approach to this diagnosis, because it takes so long to develop and because the patients at that stage have disease that is too far advanced make a therapeutic impact. Consequently, the definition of CTRCD is based upon a meaningful drop of ejection fraction (10% if asymptomatic, 5% if symptomatic) to below the normal range, albeit with some differences in exact definitions [
3].
The cause of LV dysfunction in cancer patients is inherently multifactorial, with significant contributions from not only chemotherapy, but also risk factors that are shared between cancer and cardiac disease [
4]. Nonetheless, patients treated with potentially cardiotoxic chemotherapy represent an important group for surveillance both acutely at the time of chemotherapy, as well as during long-term follow-up.
CTRCD may arise from exposure to cytotoxic agents or biological agents. While the classification into types I and II cardiotoxicity, respectively, has been criticized as being too simplistic, it has some value. CTRCD from biological agents is usually reversible, and uncommon — most studies suggest frequencies of 2% with lapatinib, imatinib, trametinib, and bevacizumab, although up to 11% is reported with sunitinib, and up to 27% with trastuzumab [
5], possibly reflecting other influences. In contrast, myocardial injury due to cytotoxic agents, related to oxidative stress, free radical formation, and cell death, is associated with ultrastructural changes at biopsy, and is usually irreversible. CTRCD may occur at three stages in relation to chemotherapy — (i) acutely (a dose-dependent and usually reversible decline in LV function during/after infusion), (ii) subacutely (a rare, myocarditis-like presentation with edema, wall thickening, and diastolic dysfunction within a few weeks), and (iii) late (usually in 1st year but occurring in up to 10–20 years, with the dose-response influenced by age, sex, underlying CVD, hypertension, and smoking) [
5]. As the development of HF or LV dysfunction is not purely a reflection of chemotherapy, but is also influenced by the background risk factor milieu, possible direct effects of cancer on the heart, and the impact of other treatments including radiotherapy, this needs to be approached on a probabilistic rather than a deterministic basis.
Clinical Prediction
A Bayesian approach is necessary whenever imaging is used in risk assessment, and patients over the age of 65, those with previous anthracycline use and radiotherapy, those with coronary or other heart disease, and hypertension are particularly at risk of developing anthracycline-related cardiotoxicity [
6]. The development of LV dysfunction and cardiotoxicity is dose-dependent — the frequency of HF attributable to anthracyclines, at least acutely, has decreased from the original report by Von Hoff [
7]. Paradoxically, patients administered low doses of anthracycline account for the majority of cases, as although the rate of occurrence is low, these account for a very large number of patients, who are at some risk, even at doses < 250 mg/m
2 [
8]. The use and dose of anthracycline has been incorporated into various risk prediction tools, but although these do identify the spectrum of risk, they have been unable to reliably identify a very low risk group, who could forego imaging [
9,
10].
The fundamental need for HF surveillance in this group is a solution that will work in large numbers of people (which has implications for cost and availability), at a high level of accuracy (taking note of the risks of misdiagnosis), at a high level of test-retest consistency, and irrespective of differing baseline risk. Three successive stages precede the development of HF — an initial phase where the causative pathway is identified, abnormal myocardial deformation, and asymptomatic LV dysfunction (Table
1). The use of more specific and less sensitive tests may detect disease at a late stage, whereas more sensitive and less specific tests have the potential of detecting early disease, but at the risk of mis-labeling patients as having disease.
Table 1
Identification of CTRCD at various clinical stages before the development of HF
Asymptomatic LVD | EF < 53% | BNP |
Abnormal deformation | Abnormal LV strain | BNP |
Causative factors | Myocyte injury | CMR | HsTn |
Oxidative stress | | MPO |
Fibrosis | | ST-2, galectin |
Inflammation | | CRP, IL-6 |
MicroRNA | | miR-1, miR-29b, miR499 |
Use of Ejection Fraction to Guide Therapy
Ejection fraction (EF) is part of the current definition of CTRCD, and is uniformly referenced in the guidelines [
3,
11]. However, EF is not without its problems. EF is a useful marker of prognosis in patients with HF, particularly with an EF < 40%, but its association with outcome in individuals with preserved EF is rather poor. The diagnosis of CTRCD is based upon a meaningful
reduction of EF to below the
normal range, and there are challenges to both defining normal and defining reduction. The ability to interpret change can be confounded by alterations of loading conditions, implying that it may change despite the presence of stable myocardial function if BP changes in the course of serial follow-up. High and low heart rate, mitral regurgitation, and left bundle branch block all provide technical challenges.
Two-dimensional echocardiography is the most widely used technique for surveillance, and calculations of LV volumes using this technique are dependent on geometry, with changes from one visit to the next being influenced by different imaging planes. These relatively common differences in imaging planes lead to wide confidence intervals for the detection of sequential changes — absolute EF differences of up to 10% cannot be attributed to a change of myocardial status and may simply reflect measurement variation [
12]. The definition of normal with different imaging modalities is different [
13], and both sex and ethnic background influence findings [
14].
EF may be calculated without geometric assumptions when a full 3-D data set is obtained, most commonly using 3D echocardiography or cardiac magnetic resonance. In contrast to the 10% confidence intervals of 2D echo, a meaningful change of 3D-EF is 5–6% [
15]. However, although 3D-EF is recommended in guidelines, it is often not performed because of inadequate imaging windows and suboptimal image quality.
EF is widely assessed at baseline in patients undergoing anthracycline chemotherapy, with a recent data-linkage of chemotherapy for breast cancer with administrative data showing its use in > 85% at baseline. In contrast, only about 50% of patients have an echocardiogram performed during follow-up, and fewer than 50% have monitoring is recommended in guidelines [
16,
17].
Response to Abnormal Test
There are two potential responses to an abnormal test — initiation of cardioprotection, or modifying or interrupting cardiotoxic therapy. The latter is very much a last option because it may have a detrimental effect on treatment of the cancer, and is rarely required unless the patient has developed overt heart failure or other steps have failed.
Four groups of medications have been used for cardioprotection. Dextrazoxane has been known to have a cardioprotective effect during anthracycline chemotherapy for nearly 50 years, but its uptake has always been impeded by concerns that it may limit the efficacy of anthracyclines, and thereby inhibit successful cancer remission. While its activity was traditionally attributed to protection from oxidative stress (through iron chelation), it seems more likely that its efficacy pertains to differential expression and/or regulation of topoisomerase II isoforms in cardiac and cancer cells, leading to the selective modulation of anthracycline action [
33]. Because this agent is thought to provide protection from injury, the model would be uniform use from time of initiation of chemotherapy, rather than selective use based upon abnormal imaging findings (see below).
In a systematic review of 25 studies with neurohormonal blockade using angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, and beta-blockers, Elghazawy [
34] showed similar levels of efficacy between the groups, with a 2.4% absolute difference in EF between treated and untreated groups. Their effectiveness for preserving EF was similar immediately after therapy, and at 6 and 12 months following treatment. These medications are presumed to work by unloading the LV, as they do in heart failure, as well as potentially through antioxidant and other metabolic effects.
The third group of agents that have been shown to be effective are statins, which are thought to limit the oxidative stress through which anthracyclines mediate cardiotoxicity. A recent meta-analysis of six studies [
35] showed a 6% average difference in ejection fraction between treated and untreated groups, with a 60% odds reduction for developing cardiotoxicity. In a network meta-analysis, statins appeared to be more effective than neurohormonal blockade in protecting patients against a drop in EF [
36].
The final class of cardio-protective treatments are modern heart failure drugs such as Entresto and SGLT2 inhibitors, but this evidence base is small and not yet ready for routine use.
The use of GLS surveillance is based on a fundamental decision to pursue selective rather than universal cardioprotection. A number of medications — angiotensin converting enzyme inhibitors, angiotensin receptor blockers, beta-adrenoceptor blockers and statins — have proven benefit in the prevention of cardiotoxicity. The use of cardioprotection in all patients at risk would be simpler than an imaging-based approach, and it avoids concerns about timing, test accuracy, and interpretation. However, the fundamental problem is that such an approach is dependent on acceptance that 80% of patients would receive this treatment unnecessarily, as cardiotoxicity develops in < 20%. In the placebo groups of randomized controlled trials, the average change of EF — while variable — is most commonly < 5% [
37]. Decision analytic models have shown that a selective approach is more cost-effective than universal approach [
38].
If a selective (image-guided) strategy is used, there is some time urgency in the response to an abnormal imaging surveillance test. In a study of over 200 patients with anthracycline related CTRCD, Cardinale reported 64% to respond to ACE inhibitors and beta-blockade administered within 2 months, < 30% to respond to treatment administered between 2–4 months, and < 10% to respond to treatment provided thereafter [
39]. This responder status was important in predicting subsequent event-free survival, which was < 40% at 2 years in non-responders (EF < 50% and ΔEF < 10%) and partial responders (EF < 50% and Δ EF ≥ 10%).
Conclusion
Anthracycline chemotherapy continues to carry a risk of myocardial dysfunction and heart failure. While neurohormonal antagonists and statins reduce the risk of cancer treatment-related cardiac dysfunction (CTRCD), the frequency of this problem is too low to justify the uniform use of these cardioprotective therapies. Consequently, most centers use a risk-guided strategy of cardioprotection — most commonly using 2D-EF. The problem is that 2D-EF measurements have broad confidence intervals, so substantial changes are required to designate the presence of LV dysfunction. GLS provides a more reliable and reproducible means of assessing global cardiac function, and observational studies and a randomized trial have shown the most reliable approach is to use a 10–15% relative change of GLS with therapy. GLS is a feasible, robust, and reproducible technique that requires limited training and is ready for wide adoption.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.