What Is the Cardiac Impact of Chemotherapy and Subsequent Radiotherapy in Lymphoma Patients?

Introduction

Anthracyclines are widely used against cancer, but can induce cardiotoxicity, usually by means of irreversible damage to cardiomyocytes that can manifest with left ventricle (LV) remodeling, dilation, and eventually heart failure (HF). The mechanisms behind anthracycline cardiotoxicity involve oxidative stress, alterations in mitochondrial function, and activation of the DNA damage responses, which can eventually lead to cardiomyocyte apoptosis and necrosis. The surviving cells can develop maladaptive changes to face this form of cardiac injury (6, 8) (Fig. 1).

OPEN IN VIEWER

FIG. 1.

Schematic representation of the main mechanisms of anthracycline-induced injury to cardiac cells. The classic model of ANT cardiotoxicity involves the generation of ROS by the quinone moiety common to all ANTs. ROS and reactive nitrogen species hyperproduction results in damage to DNA, protein carbonylation, and lipid peroxidation leading to cellular dysfunction and cardiomyocyte death. ANTs can also bind and block the functions of both topoisomerases 2A (TOP2A) and 2B (TOP2B). Tumor cells express high levels of TOP2A, whereas TOP2B is ubiquitously expressed. Cardiomyocytes express TOP2B, but not TOP2A. ANTs form a complex with TOP2B inhibiting its enzymatic activity. Without functional TOP2B, DNA breaks accrue, leading to the activation of p53 tumor-suppressor protein, mitochondrial dysfunction, and the generation of ROS that result in cardiomyocyte death. Another mechanism underlying doxorubicin-dependent oxidative stress is linked to the ability of the drug to directly interfere with the activity of NADPH oxidase and nitric oxide synthase (NOS). Both NADPH oxidase and NOS can transfer electron from NADPH to doxorubicin, causing the formation of SQ-DOX. SQ-DOX in turn transfers electron to O₂ and generates O₂ ⁻. In the NOS compartment, O₂ ⁻ can react with NO to form peroxynitrite (ONOO⁻), a powerful oxidant that can generate free radicals. An alternative mechanism by which ANTs exert their cardiotoxic effects is the inhibition of neuregulin-1-HER-2 in cardiomyocytes. Doxorubicin also induces necrosis of immune (i.e., macrophages) and cancer cells releasing HMGB1, which activates TLR-2 and TLR-4 in cardiomyocytes and inflammatory cells inducing the release of proinflammatory cytokines. These primary effects induce a plethora of secondary effects in cardiomyocytes (e.g., DNA damage, lipid peroxidation, and mitochondrial dysfunction), resulting in cell dysfunction and death (8). Reproduced with permission from Varricchi et al. (8). ANT, anthracycline; NADPH, nicotinamide adenine dinucleotide phosphate; NOS, nitric oxide synthase; ROS, reactive oxygen species; SQ-DOX, semiquinone doxorubicin.

Radiotherapy (RT) is used in many anticancer protocols, but unfortunately can also damage the heart [radiation-related heart disease (RRHD)] and induce cardiomyopathy, coronary artery disease, and valvular disease. RRHD can be acute, lasting only a few days after irradiation, or chronic, leading to myocardial fibrosis. Chronic cardiac damage after RT can manifest many years after irradiation, and hence, it is difficult to estimate the real incidence of RRHD (9) (Fig. 2).

OPEN IN VIEWER

FIG. 2.

Mechanisms of radiation-related heart disease.

At present, LV ejection fraction (EF) measured by two-dimensional (2D)-echocardiography is widely used to detect cardiac dysfunction during and after anticancer treatments (9). However, left ventricular ejection fraction (LVEF) measurements could underestimate cardiac injury, since a reduction in LVEF is detectable by echocardiography when heart dysfunction is already consistent; therefore, this parameter is not really suitable to identify early, subclinical damage. Moreover, intra- and interoperator reproducibility of LVEF measurements is low. New echocardiographic approaches, such as assessment of myocardial deformation with global longitudinal strain (GLS) in speckle tracking echocardiography (STE), allow for detection of early cardiac motion abnormalities (5, 7).

GLS and STE have been investigated mainly in breast cancer, and are nowadays recommended by the American Society of Echocardiography and the European Association of Cardiovascular Imaging (5). It has been recently shown that GLS predicts cardiotoxicity in lymphoma patients in a retrospective study (3). Here, in a small prospective study, we used traditional and STE to detect early anthracycline cardiotoxicity in adult patients with Hodgkin's lymphoma (HL) and non-Hodgkin's lymphoma (NHL), aimed at verifying whether patients who underwent mediastinal irradiation for RT after anthracycline-based chemotherapy (CT) develop further impairment in GLS, and we checked whether the alterations in GLS induced by radiations predominantly affect the midventricular and apical regions of the LV, more exposed to radiations.

Results, Discussion, Limitations, and Future Perspectives

Twenty-nine adult patients (19 females; 66%) newly diagnosed with lymphoma to be treated with anthracycline-based CT (with or without RT) were evaluated from November 2015 to March 2018. One patient did not complete all the CT cycles, one had a diagnosis of relapse and underwent a second-line treatment with rituximab and bendamustine, and five were lacking a baseline cardiac evaluation. Hence, 22 patients met the inclusion criteria and were eligible for the study (see Fig. 3 for study design). Our final cohort consisted of 14 (63.6%) females, with a median age of 41 (20; 88) years; 86.4% had HL. All patients underwent anthracycline-based regimens (Table 1), median cumulative dose was 378.00 (100.00; 600.00) mg/m². Fourteen out of 22 patients (subgroup S2) were also treated with involved-field RT (IFRT), mostly directed to the anterior mediastinum with a median prescribed dose of 30.00 (30.00; 36.00) Gy, fractionated in 20 sessions, while subgroup 1 (S1) identified patients treated with CT only.

OPEN IN VIEWER

FIG. 3.

Study design flow chart.

Baseline echocardiographic data [baseline assessment (BA)] were compared with echocardiographic data acquired at the end of CT [chemotherapy assessment (CA)], and within 6 months from the end of hematologic treatments [final assessment (FA)]. CA was performed after 4.5 months (2; 8) from BA [5 months (2; 8) for S1, 4 months (2; 7) for subgroup 2 (S2)]. FA was performed after 9.5 (6.0; 13.0) months from BA; in particular, the median follow-up for the S1 was 9 (8; 12) months, while for the S2 was 10 (6; 13) months.

First, we assessed cardiac function by analyzing echocardiographic differences from BA to CA in the whole population. Importantly, at the end of CT, we observed a significant reduction in LVEF, from 60% (54; 65) at BA to 55% (53; 65) at CA. No other conventional echocardiographic parameter changed, as shown in Table 2. These results were confirmed when we excluded 4 patients older than 65 years: EF declined from 60% (55; 63) to 57% (53; 65), p = 0.033.

We then divided our population according to whether patients underwent RT or not. The subgroup analysis revealed that in S1 no conventional parameter changed from CA to FA (final assessment), while in group S2, we found a significant increase in E/E’ from CA [6.2 (4.2; 14.2)] to FA [6.9 (5.5; 15.2), p < 0.05] (Table 3), suggesting an increase in LV diastolic pressures due to RT after CT. The same was true when we excluded the three elderly patients: E/E′ increased from CA [6.0 (4.2; 13.0)] to FA [6.8 (5.5; 13.6), p = 0.01].

We then aimed at further assessing heart function by using the more sophisticated STE. We found that in the whole population, GLS significantly worsened from BA to CA [from −17.17 (−23.33; −11.33) to −16.17 (−19.61; −11.94), p < 0.01], and the same was true when we excluded the four elderly patients [from −17.58 (−20.44; -13.50) to −16.19 (−19.61; −11.94), p < 0.01]. Such impairment persisted at FA, with no significant differences between CA and FA (Table 2). At subgroup analysis, GLS in S1 had the same behavior (Table 4).

Interestingly when we analyzed group S2, GLS worsened, although not significantly, from BA to CA, while the change from BA to FA was significant (Table 4), also when we excluded the three elderly patients. In particular, the entity of worsening in GLS from CA to FA was significantly more relevant in S2 than in S1, suggesting that RT after CT significantly adds on an impairment in cardiac function (Table 5). In S2, the impairment in GLS at FA was more evident in the midventricular and apical regions of the LV (Table 4).

Recent studies have shown that baseline GLS predicts long-term LV dysfunction after anthracycline CT and HF hospitalization in lymphoma patients (3). In our work, we observed a slight but significant reduction of EF right after CT in our population (Table 2), even though no patient had a decrease in EF sufficient to meet the diagnostic criteria for cardiotoxicity (1, 5, 9). In addition, we checked for differences between patients treated with CT only (S1) and patients treated with RT after CT (S2). In S1 patients, GLS analysis detected a decrease in myocardial deformation after CT compared with baseline (CA vs. BA). No further changes in GLS were found at FA versus CA, suggesting that the subclinical damage induced by anthracyclines has an early onset right after CT and does not revert, at least during the time of observation (Table 2). A longer follow-up will reveal whether such alterations in GLS will recover or whether they are predictive of overt cardiac deterioration, with adverse remodeling and further decrease in EF.

In S2 patients, when comparing CA and FA, by means of standard 2D echocardiography, we found an increase in E/E′ (Table 3), showing that the addition of RT upon previous CT administration can worsen diastolic heart function. Further studies are warranted to assess this observation on a longer time period. Moreover, STE analysis in S2 patients revealed that GLS worsened at FA compared with BA (Table 4). These alterations were mainly due to the addition of RT on CT, since there was no significant change of GLS at CA versus BA, while the percentage of drop in GLS between CA and FA was significantly more pronounced in S2 than in S1 (p = 0.04, Table 5). These alterations were mainly localized at the apical and midventricular regions of the LV. We hypothesize that this happens because these are the heart portions more exposed to mediastinal irradiations for lymphoma (Table 4). On the contrary, the reduction in LS observed in S1 is more uniform and does not show patterns of regionality. Seven out of the 22 patients (31.82% of the whole sample) had a reduction of GLS ≥15% from BA to FA (Table 1) (1, 5, 9). Among these seven patients, six belonged to S2, and this seems to confirm our hypothesis that RT after CT can further worsen cardiac damage already induced by CT alone.

There are some limitations to our study. First, this is a single-center study with a small sample size. We plan to keep on enrolling additional patients prospectively, in a multicenter study, to expand our sample and have a better understanding of the echocardiographic changes, and whether such changes are predictive of future clinical events. Second, some patients were already being treated with β-blockers and angiotensin converting enzyme (ACE)-inhibitors, which may have protected from further cardiac deterioration after CT and/or RT, thus masking the real impact of these anticancer treatments. Indeed, β-blockers can reduce oxidative stress and calcium overload in myocardial cells (8). In particular, carvedilol demonstrated a preventive role against LV dysfunction in patients treated with anthracyclines, reducing reactive oxygen species (ROS) production, apoptosis of cardiomyocytes, and mitochondrial alterations (8). The renin/angiotensin/aldosterone system also plays a key role (8). In patients treated with anthracyclines, enalapril reduced the incidence of LV dysfunction, while candesartan ameliorated in vitro anthracycline cardiotoxicity and telmisartan reduced acute LV dysfunction induced by doxorubicin in rats. Coadministration of ACE-Is and carvedilol can reduce cardiac damage induced by anthracyclines (8).

Correlating serum biomarkers to our imaging markers would sure help in the assessment of cardiotoxicity (1, 8, 9), but unfortunately, for most patients we were not able to collect NT-proBNP or troponin values. Importantly, in the setting of oxidative damage, such as the one from anthracyclines and RT, markers of oxidative stress could play a key role in early detection of cardiac damage (8). Unfortunately, ROS are very unstable and their detection in patients treated with anticancer therapies is almost impossible to use in clinical practice (8). Metabolomics may play an important role, with the quantification of succinate that could be used as an alternative biomarker of oxidative or nitrosative cardiotoxicity (8). Furthermore, the decrease in NAD(P)H:quinone oxidoreductase 1 activity and the increase of ROS produced by the activity of NAD(P)H oxidase have also been proposed as early markers of cardiac damage in patients treated with anthracyclines (8).

Another limitation concerns the echocardiographic technique, since STE depends on 2D image quality; nevertheless, none of the patients was excluded from the analysis because of poor quality of echocardiographic images. Furthermore, there is vendor-specific variability in strain measures that may affect reproducibility of GLS findings.

All things considered, with the abovementioned limitations, the data we have obtained from our small prospective study seem to suggest that in adult lymphoma patients, EF declines right after CT. Also, RT seems to have an impact on cardiac function that is additional to CT alone, and this further damage can be detected by STE. Further high-volume studies and longitudinal observations are needed to reveal whether these subclinical changes induced by anthracyclines combined with RT are reversible or will translate into overt cardiotoxicity.

Notes

Speckle tracking echocardiography

Additional 2D-echocardiographic cine loops, using frame rates between 50 and 70 fps, from four-, two-, and three-chamber views were acquired for subsequent offline speckle tracking-derived analysis using a commercially available vendor-independent strain software (Epsilon, Echolnsight, Milwaukee, WI). The analysis was based on an 18-segment model and GLS was obtained from the mean of the values of the 18 segments. Regional strain values from the six basal, six midventricular, and six apical segments of the LV were averaged to obtain three regional longitudinal strain (LS) values (basal, midventricular, and apical, respectively). LS analysis allows for evaluation of shortening of a specific portion of the heart obtained by tracking the endocardial border of the LV, compared with its original length, and is expressed as a negative percentage (6). This evaluation is important to estimate myocardial deformation, defined as peak longitudinal systolic strain and expressed as percentage of the longitudinal shortening of systole compared with diastole. A worsening in strain is defined by a less negative number or a smaller deformation in the longitudinal axis, while an improvement is defined by a higher negative number (in its absolute value) or a greater deformation in the same axis, referred to a given segment (Fig. 4).

OPEN IN VIEWER

FIG. 4.

Example of global longitudinal strain analysis of the left ventricle. Bulls-eye plot of GLS obtained after tracking the LV end-systolic endocardial border in apical 4-chamber (A), apical 2-chamber (B), and apical 3-chamber (C) views. GLS, global longitudinal strain; LV, left ventricular.

A clinical cardiologist expert in echocardiography (V.M.) blinded to patient status analyzed each study for standard echocardiography and strain measurements. To assess for intraobserver variability, eight studies with adequate image quality were randomized for reanalysis 1 week after initial analysis by the same physician, and, to assess interobserver variability, by a second independent clinical cardiologist expert in echocardiography (D.C.), again blinded to patient status.