Echocardiography is recognized as the primary non-invasive tool for the evaluation of cardiac reserve because of its low cost, lack of ionizing radiation, and ability to assess global and regional cardiac motion without the need for a physiological steady-state (79). Most often, exercise echocardiography is performed immediately after peak exercise on a treadmill or cycle ergometer, where the patient is transferred into a supine position for imaging (45, 79). Less commonly, images are acquired during exercise on an upright or semi-recumbent cycle ergometer (45, 79). A number of echocardiographic techniques can be used to assess different indices of systolic and diastolic function. These include M-mode echocardiography, Doppler echocardiography, 2D and 3D echocardiography, and myocardial deformation imaging. An overview of each of these techniques will be provided below.

Radionuclide angiocardiography is a nuclear imaging technique that can evaluate ventricular performance during exercise. Most often, imaging is performed during supine or upright exercise on a bicycle ergometer. Radionuclide angiocardiography can be differentiated into two key techniques: (1) first-pass radionuclide angiocardiography (FPRNA) and (2) equilibrium radionuclide angiocardiography (ERNA).

FPRNA comprises a rapid bolus injection of radiotracer, preferably Technetium99m diethylamine triamine pentaacetic acid because of its renal excretion, hence minimizing radiation exposure (106). After bolus injection, radioactivity is measured during the first passage of the tracer through the heart. Generation of time-activity curves allows determination of LV end-diastolic and end-systolic counts, from which SV, EF, and CO can be inferred. ECG gating is required (especially for single-crystal gamma cameras) to allow reconstruction of 25–50 ms frames to calculate an EF. Ventricular volumes can be measured by a geometric or count-proportional technique. Similar to the dye-dilution methods, CO can also be assessed more directly by measuring the area under the time-activity curve. FPRNA provides temporal separation of the cardiac chambers with high target-to-background ratio, as radioactive counts are accumulated only during the initial passage of the isotope as it traverses the central circulation. Therefore, the radioactive bolus will have been ejected from the RV before entering the LV. This technique can be applied whilst exercising, but only in one projection, usually the anterior for LV function studies. For assessment of RV function, preferably a shallow right anterior-oblique view is recommended (106). Typically, data acquisition requires 20–50 s. Because of these relatively short acquisition times, FPRNA can be performed at high exercise levels (107).

In ERNA, the radioactive tracer is attached to patients' red blood cells or human serum albumin and allowed to reach equilibrium in the vascular system before imaging is performed (108). Image acquisition is triggered by the R-wave on the electrocardiogram such that the duration of each cardiac cycle can be identified as the interval between two successive R-waves. Each of the 200–600 cardiac cycles are summed and divided into 16–32 frames. Corresponding frames of multiple cardiac cycles are summed together during an acquisition period of at least 3 min. Therefore, in contrast to resting ERNA studies, acquisition time is often the limiting factor during exercise. However, previous research has demonstrated the feasibility of ERNA during peak exercise using short acquisition times (109). A time-activity curve is generated within a region of interest, involving the LV, to display the change in radioactive counts throughout the cardiac cycle (Figure 4). The peak and nadir of the time-activity curve represent the end-diastolic and end-systolic counts, respectively. Measurement of ventricular volumes using ERNA is susceptible to a number of problems. First, accurate assessment of ventricular volumes using ERNA requires adequate separation of cardiac chambers. Because of the considerable overlap between the RA and the RV, ERNA is not appropriate for evaluation of RV volumes or function. Furthermore, successful gating relies on regular heart rates, such that ERNA cannot be performed in patients with marked arrhythmias.

An important advantage of RNA techniques is that measurements of ventricular EF are not dependent on geometrical assumptions, but are based on the assumption that LV volumes are proportional to LV counts throughout the cardiac cycle. Ventricular volumes can be derived by a counts-based or geometrically based method, but these measurements are not routinely performed in a clinical setting. RNA techniques provide very reproducible measurements of EF both at rest and during incremental exercise (110, 111). Importantly, EF can be obtained in virtually all patients, regardless of body habitus, during maximal exercise (109, 111). However, one of the main limitations of RNA compared to cardiac magnetic resonance imaging (CMR) and echocardiography is its radiation exposure. This may be particularly important in populations who require ongoing cardiac monitoring such as childhood and adolescent cancer survivors. Furthermore, due to its intrinsic low spatial resolution, RNA contributes little to the characterization of anatomical structures or pressure gradients—both of which may also be important markers of CTRCD.

CMR is generally considered the gold standard technique for measurement of biventricular volumes and cardiac function at rest (22). Rather than depending on single image slices and multiple geometrical assumptions, the entire heart can be sampled without limitations in acoustic windows. Typically, CMR requires electrocardiographic gating and multiple breath holds. For each frame in the cardiac cycle, an image is acquired over several heartbeats. All acquired image frames can be displayed as a cine loop, with optimal spatial and temporal resolution. To measure RV and LV volumes, a stack of short-axis slices covering the ventricles from base to apex is acquired. Each slice contains a cine loop of image frames throughout the cardiac cycle. Endocardial borders from end-diastolic and end-systolic frames can be contoured in a manual or automated manner and volumes for both ventricles can be calculated using the slice summation method (Figure 5). The acquisition of a complete set of cine CMR images can be performed in 5–10 breath-hold periods of 10 s. However, the need to perform multiple breath holds and the relatively long acquisition times, limit the use of cine CMR imaging in exercise testing. Furthermore, the ECG signal is distorted by the magnetic field of the CMR system because of the magnetohydrodynamic effect which makes it extremely challenging to acquire reliable cardiac gating during the increased movement and blood flow associated with exercise.

To overcome these problems, real-time exCMR imaging has been developed (112, 113). In this mode, each image frame of the cardiac cycle is acquired independently and sequentially. Using techniques such as parallel imaging to improve temporal resolution, a complete image frame can be obtained within 35 ms, with only a slight compromise in spatial resolution. Because of such rapid data acquisition, the effects of cardiorespiratory motion are less pronounced and it becomes possible to visualize the beating heart during free breathing and without cardiac gating (112). As a result, real-time exCMR can be used to assess biventricular volume changes throughout exercise, without the need for geometrical assumptions or radiation exposure (112). A comparison with direct measurement (via catheterization) of cardiac function among athletes, arrhythmia patients and individuals with heart failure suggests exCMR is capable of measuring cardiac function with a high degree of accuracy (R = 0.96 for CO), precision (inter-observer CV for stroke volume: 1.9–5.0%) and reproducibility (Intra-class correlation coefficients: 0.86–0.94) (113). Simultaneous assessment of both ventricles also enables the study of ventricular interaction during exercise. This may be important as there is limited data regarding the RV response to exercise, and evolving evidence that it may be of importance to overall cardiac performance in health and disease (78, 114). Furthermore, the RV can also be impacted by cancer treatment such as anthracyclines (115, 116), however current research is primarily focused on the LV. Tools such as exCMR, which can provide a detailed quantification of combined LV and RV function may better elucidate the influence of cardiotoxic cancer treatments on RV function (and its interaction with LV function).

Another CMR method for determination of SV is CMR phase-contrast flow quantification (117). This technique uses application of bipolar magnetic field gradients to encode the velocity of moving blood. First, a region of interest is delineated around the vessel lumen to determine the cross-sectional area of the vessel. The spatial mean velocity for all pixels in this area is measured and plotted against time. The instantaneous flow volume across the vessel lumen can be calculated by multiplying the spatial mean velocity by the cross-sectional area. Finally, integrating the instantaneous flow volumes for all frames throughout the cardiac cycle yields the total flow volume per heartbeat. Systemic and pulmonary blood flows can be measured from measurements in the proximal ascending aorta and pulmonary trunk, respectively. However, similar to pulse wave Doppler echocardiography, CMR phase-contrast flow quantification is susceptible to aliasing if the flow velocities are too high. An advantage of CMR phase-contrast vs. Doppler echocardiography is the fact that it is not limited by acoustic windows. However, phase-contrast quantification is very dependent upon maintaining a stationary imaging plane and the respiratory translation and whole body movement during exercise can result in considerable error. Nevertheless, some groups have shown that this technique is feasible (118–121), and have demonstrated high correlations for exercise SV (R = 0.988) measured using real-time phase-contrast quantification and real-time assessment LV volumes (122).

Despite several important advantages over other imaging techniques, CMR also has its limitations. The key disadvantages of exCMR are its high cost and limited availability. Furthermore, exercise imaging with CMR is confined to exercise in the supine body position. Usually, this is performed on a supine cycle ergometer attached to the CMR table, which allows for in-magnet exercise testing. Some groups have used treadmill exercise outside the scanner to achieve maximal cardiovascular stress, followed by post-exercise breath-hold cine imaging (123). However, the rapid changes in physiology during recovery may render this a poor surrogate of real-time exercise values. Open-bore CMR systems might enable cardiac volume assessment during upright exercise in the future. Cheng et al. (124) demonstrated the feasibility of CMR phase-contrast flow quantification during upright exercise in an open CMR magnet. However, compared with conventional CMR scanners, open CMR scanners typically have a lower magnetic field strength, which compromises image quality and prolongs acquisition times. Currently, volumetric assessment of the heart using real-time CMR has not been performed during upright exercise.

Pharmacologic stress can be an alternative means of assessing cardiac reserve among patients who may be unable to exercise. This has primarily been used for the assessment of inducible ischemia and myocardial viability, either through inotropes such as dobutamine, or vasodilators such as adenosine. Dobutamine induces β₁-mediated increases in heart rate and contractility, and β₂-mediated peripheral vasodilatation, thereby simulating some of the physiologic changes seen during exercise. An advantage of pharmacologic stress over exercise imaging is the improved image quality due to lack of movement. The utility of pharmacologic assessment of cardiac reserve in the heart failure setting has primarily been investigated by stress echocardiography, with studies investigating the utility of pharmacologic stress CMR primarily limited to the diagnosis and prognosis of coronary artery disease (125, 126). Cardiac reserve, assessed by changes in contractility (either as EF or wall motion score index) measured during dobutamine echocardiography in patients with established heart failure has been associated with heart failure prognosis (127, 128) and response to medical therapy (129, 130). However, it is unclear whether pharmacologic assessments of cardiac reserve can identify at risk individuals who will go on to develop heart failure. Furthermore, it is important to note that whilst dobutamine induces physiological changes that are similar to exercise, these techniques should not be considered interchangeable. Compared to exercise, dobutamine infusion has been associated with a blunted increase in stroke volume, cardiac output and LV wall stress (131)—likely due to disparate effects on preload and afterload between the two techniques. Consequently, where possible, exercise should be the preferred modality given it is more likely to induce the physiology where heart failure symptoms such as fatigue and exertional breathlessness occur. This decision making should also consider the potential risks from pharmacologic stress such as arrhythmias, nausea, dizziness and vomiting.

There are also general considerations related to the assessment of cardiac reserve that warrant discussion. The type and timing of exercise are important as they can also influence the sensitivity for detecting cardiac dysfunction. These may be dictated by equipment availability, patient familiarity and the imaging modality being used (79). The pattern and type of muscle recruitment during treadmill exercise is often more familiar to patients than bicycle exercise, which improves the ability of the patient to achieve legitimate cardiovascular endpoints (as patients may be limited by peripheral muscle fatigue during cycling). Furthermore, a number of childhood cancer patients may be too small to effectively use a stationary bicycle, and so a treadmill may be the only method available for conducting exercise testing. However, acquiring images during walking or jogging using echocardiography is very technically demanding, and is not possible using existing CMR or nuclear imaging techniques. Therefore, it is really only feasible to perform imaging immediately following the exercise bout if treadmill is the preferred mode of assessment. This is an important consideration as cardiac loading can quickly return to normal following exercise (132), and so subtle dysfunction may be missed during the post-exercise image acquisition period. An advantage of supine or semi-supine bicycle ergometer is the ability to assess cardiac function during progressive exercise, and so cardiac function can be directly compared to the development of a patient's exertional symptoms (79). However, cycling in the supine position is uncomfortable and there are physiological differences to upright exercise that mean the maximal workloads and heart rates achieved with supine exercise are usually lower than in the upright position (133, 134). It is important to acknowledge that cardiac reserve assessment requires additional time, equipment, and expertise beyond standard resting cardiac function assessments. However, whilst being more demanding than resting assessment, there is potential for improved patient outcomes and subsequent cost savings if cardiac reserve proved to have additional predictive value over resting measures. Given the aim of continually improving patient care and the increasing importance of cardiovascular health in cancer survivors, there is a strong rationale for pursuing further research into the potential cost-effectiveness of measures of cardiac reserve. The timing of cardiac reserve assessment in relation to cancer treatment is another important consideration. There are no studies to guide the optimal timing of cardiac reserve assessment in the oncology setting, however following the standard guidelines for other assessments would be a reasonable starting point (10). Timing of assessment should consider the type, length, frequency, and dose of treatment, in addition to each patient's baseline cardiovascular risk profile and self-reported symptoms (10). It would be reasonable to conduct assessment of cardiac reserve prior to treatment, which may aid in identifying patients with pre-existing dysfunction, whilst also providing a baseline measurement from which to compare subsequent results. An assessment at the completion of standard doses of anthracycline-containing chemotherapy would also be logical, whilst further assessment may be needed for patients with increased risk of heart failure such as those scheduled to receive high doses of chemotherapy or other subsequent cardiotoxic treatments, or those for whom cardiac function is borderline. Testing too frequently may be problematic, and whilst treatment such as anthracycline chemotherapy is associated with dose-dependent cardiac injury it is unclear how long an interval is required for this to manifest into measurable dysfunction. This highlights the need for further work focusing on this question.

There are a number of studies (primarily cross-sectional) comparing cardiac reserve in pediatric cancer survivors treated with cardiotoxic therapies, largely anthracycline chemotherapy, to matched healthy controls (135–145). Compared to control subjects, blunted increases in LVEF or fractional shortening (135–137, 140, 146), attenuated increases in stroke volume and cardiac index (138, 141, 142, 144) or a blunted increase in LV longitudinal and circumferential strain (147) have all been reported. Another important question regarding the use of cardiac reserve in cancer patients is its ability to detect sub-clinical dysfunction. Unfortunately, the degree to which reductions in cardiac reserve precede impairments in resting function remain unclear, with some studies reporting resting function to be preserved (136, 137, 140, 142, 146, 147), whilst others report simultaneous reductions in resting function (138, 141, 144). This discrepancy between studies could be explained by the cross-sectional designs, differences in inclusion and exclusion criteria between studies, and the long duration since the completion of cancer treatment.

There are two studies that prospectively evaluated changes in exercise cardiac reserve among pediatric cancer survivors previously treated with anthracyclines (143, 145), however these studies have conflicting findings. Guimaraes-Filho et al. (145) found that whilst cancer survivors who were ~3 years from treatment completion had reduced exercise LVEF and increased end-systolic wall stress compared to control subjects, there was no change in cardiac function over the following 5 years (145). This makes it difficult to determine the benefit of longitudinal evaluation of cardiac reserve in this population. Sieswerda et al. (143) performed exercise echocardiography in 92 asymptomatic childhood cancer survivors 8.2 years post-chemotherapy, and whilst fractional shortening and LV end-diastolic dimension declined over the following 10 years of follow-up, the addition of baseline exercise fractional shortening measures did not provide additive benefit for predicting declines in fractional shortening beyond age, anthracycline dose and resting fractional shortening at baseline (143). Whilst these findings suggest cardiac reserve does not provide incremental value beyond resting measures in predicting future cardiac dysfunction, these findings should be interpreted with caution as exercise echocardiography was performed a number of years following treatment completion and only the most basic echocardiographic measures were employed.

The potential role of dobutamine stress has been investigated in several studies of anthracycline-treated pediatric cancer survivors. Cross-sectional studies of asymptomatic pediatric cancer survivors assessed with dobutamine stress echocardiography a number of years following anthracycline treatment (median 5.3–7 years) have shown cancer survivors have reduced posterior LV wall thickening, LVEF, and LV fractional shortening, decreased ratio of early to late peak mitral velocity (E/A), and increased LV wall stress during exercise compared to matched control subjects—demonstrating impairments in both systolic and diastolic reserve (148, 149). Furthermore, the degree of impairment in cardiac reserve appears to relate to the dose of chemotherapy received (150). In contrast, Lanzarini et al. found no differences in cardiac reserve in long-term (mean 7 years post-treatment) pediatric cancer survivors and matched sibling controls during low dose dobutamine stress echocardiography. One reason for this discrepancy could relate to differences in population characteristics, dobutamine infusion protocol and also the use of m-mode measurements, which as discussed previously, are limited by high variability. Furthermore, as these impairments are measured years following cancer treatment, a number of impairments in resting cardiac function were also present in a significant proportion of the cancer survivors assessed. Further prospective studies, using more contemporaneous measures of cardiac function are needed to better understand the trajectory and clinical significance of these differences.

Overall, it appears that there is sufficient evidence to suggest cardiac reserve is reduced in pediatric cancer survivors previously treated with cardiotoxic treatments (primarily anthracyclines), however more evidence is needed to prospectively evaluate changes in cardiac reserve following treatment and whether this can predict future CTRCD.

Compared to pediatric cancer survivors, there are few reports investigating changes in cardiac reserve among adult cancer patients and survivors. Kirkham et al. (151) performed a systematic review (which included 14 studies) investigating the role of stress testing for the early detection of cardiovascular disease in breast cancer survivors. However, the majority of studies focused on the role of stress imaging (such as myocardial perfusion imaging) or stress electrocardiography for the early detection of coronary artery disease, rather than the prediction of CTRCD. Two cross-sectional studies (n = 30–57) included in this systematic review compared cardiac reserve measured via 2D echocardiography among early stage breast cancer survivors previously treated with anthracyclines to age- and gender-matched controls (152, 153). Whilst none of the breast cancer survivors met the criteria for cardiotoxicity determined by resting LVEF or global longitudinal strain, both studies found that breast cancer survivors had reduced cardiac reserve. Specifically, this included a lower LVEF at peak exercise (153), or a lower cardiac index and SV index at peak exercise (152). Recent work from our laboratory has shed some light on the evolution of cardiac reserve impairment in breast cancer patients undergoing anthracycline chemotherapy using exCMR (154, 155). In this series of studies, we found that exercise stroke volume tended to be lower shortly following anthracycline chemotherapy (P = 0.06) (155), however over the following 12-months this evolved into a blunted augmentation of LVEF and SV during exercise, resulting in a reduction in peak exercise CO (154). One of the more intriguing findings comes from a small (n = 48) prospective analysis of mixed adult and pediatric patients with advanced cancer undergoing high-dose anthracycline chemotherapy (156). This study assessed cardiac reserve using resting and exercise RNA LVEF prior to-, after completion of low-dose and high-dose anthracycline chemotherapy. Both resting and peak LVEF declined progressively during chemotherapy, with a significantly blunted increase in LVEF from rest to peak exercise at completion of low- and high-dose chemotherapy. Notably, four patients developed congestive heart failure 1–6 months following treatment, all of whom had normal pre-treatment resting LVEF, but an abnormal increase in LVEF during exercise (<4% increase), suggesting pre-treatment cardiac reserve may be a predictor of chemotherapy-induced heart failure. However, given the small numbers and high dose of treatment, this requires further validation in larger cohorts, and with chemotherapy does that would be more applicable to the broader population of adult cancer survivors.

There have also been a handful of studies investigating the role of pharmacologic stress to assess cardiac reserve in adult cancer survivors. Civelli et al. (157) investigated the prospective changes in resting and dobutamine stress LVEF throughout and shortly following anthracycline chemotherapy, and ability for resting and peak LVEF to predict late CTRCD 18-months following chemotherapy in 49 women with advanced breast cancer undergoing high-dose chemotherapy. When subjects were differentiated into those with normal (>50%) or impaired (<50%) resting LVEF at 18-months, it was found that whilst resting LVEF was similar at all previous time points, peak stress LVEF was significantly lower from the third cycle of anthracycline chemotherapy onwards. Importantly, a >5 point reduction in LVEF reserve was a significant predictor of impaired resting LVEF at 18-months. Cottin et al. (158) assessed changes in cardiac reserve using low-dose dobutamine stress echocardiography in a mixed population of adult cancer patients undergoing standard doses of anthracycline chemotherapy. These authors found that peak LVEF did not change from baseline to the end of chemotherapy, however at the end of chemotherapy there was a blunted increase in peak mitral flow velocity (E wave) and a decrease in the ratio between peak early and late mitral flow velocities (E/A) which may be indicative of poor diastolic reserve. The lack of change in peak LVEF may reflect the lower chemotherapy doses and shorter follow-up used in this study (158) compared to that of Civelli et al. (157). It would be interesting understand whether the subtle diastolic impairment demonstrated by Cottin et al. (158) evolves into more significant dysfunction over the months following therapy. In contrast, Bountioukos et al. (159) found no additional value of contractile reserve assessed by dobutamine stress echocardiography for predicting RNA-assessed reductions in LVEF among adult patients with hematological cancers. However, it should be noted that contractile reserve in this study was quantified using the change in wall motion score index, which is a relatively blunt method for quantifying cardiac reserve, and may lack the necessary sensitivity for assessing subtle changes in peak cardiac function in this setting.

Overall these results suggest cardiac reserve is impaired in adult cancer survivors treated with anthracyclines, and may be a more sensitive marker of subclinical dysfunction given resting function was often relatively preserved. However, due to the relatively small sample sizes, these findings require validation in larger cohorts, which also investigate their potential relationship with other important cardiovascular and health outcomes. Furthermore, the utility of cardiac reserve to detect cardiac dysfunction induced by other cardiotoxic cancer therapies requires further evaluation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.