Cancer and cardiovascular disease are the leading causes of death and disability worldwide.1 Heart disease and cancer share common risk factors in an ageing population and are further linked through cardiotoxic effects of contemporary cancer treatment.2–4 Adverse cardiovascular sequelae of cancer treatment, such as cardiac dysfunction, myocardial ischaemia and thromboembolic events, can increase cardiovascular mortality in cancer patients.5 It has been estimated that 25% of patients with cancer go on to develop cardiac dysfunction as a result of their cancer therapy.6 The primary players in chemotherapy-induced cardiomyopathy are anthracyclines, such as doxorubicin and idarubicin, which are used to treat a variety of solid tumours and haematological malignancies.7
Several definitions for cancer therapy-related cardiac dysfunction (CTRCD) have been proposed over the past decades and the 2021 European Society of Cardiology cardio-oncology consensus document recently aimed to provide a harmonised definition based on the presence or absence of symptoms and changes in left ventricular ejection fraction (LVEF), global longitudinal strain and/or rise in cardiac biomarkers. 8,9 The time course of anthracycline-induced cardiomyopathy has been classified into early and late, with a cut-off of 1 year separating the two phases, and with the cumulative incidence peaking at 1 year.10 The actual incidence of late-onset anthracycline-induced cardiomyopathy is unclear. Longitudinal studies demonstrate that chemotherapy-induced heart failure has a high mortality – 60% at 2 years11,12 – and subclinical cardiac dysfunction may occur in up to 50% of all patients at 20 years.6 However, more contemporary studies have shown much lower incidence of anthracycline-induced cardiotoxicity.13,14
We sought to understand the effect of these molecular mechanisms on myocardial metabolism, which can be indirectly assessed in vivo using fluorine-18 fluorodeoxyglucose positron emission tomography (18F-FDG-PET) imaging.15 The availability of hybrid simultaneous PET/MRI of the heart allows the advantages of MRI in myocardial tissue characterisation to be combined with the unique metabolic insights provided by PET without requiring a patient to undergo two separate scans.16–19 As such, hybrid simultaneous PET/MRI imaging permits concurrent data acquisition and allows correlation of changes in cardiac metabolism with changes in myocardial function and tissue characteristics. Hybrid PET/MRI imaging of the heart has previously been used in patients with acute coronary syndrome and sarcoidosis.20,21 In the Multimodality Imaging to Detect Anthracycline-induced Cardiotoxicity (MIDAC) pilot mechanistic study in breast cancer patients, we aimed to use hybrid PET/MRI imaging to correlate metabolic PET parameters with cardiac MRI readouts of left ventricular function and myocardial tissue characteristics to provide further mechanistic insights into the pathophysiology of anthracycline-induced cardiotoxicity.
Methods
We prospectively enrolled patients with breast cancer from the National University Cancer Institute, Singapore, who were scheduled to receive a standard course of anthracycline chemotherapy between January 2017 and April 2019. To select a higher-risk cohort, patients aged >50 or ≤50 years and with either hypertension and/or ischaemic heart disease and with a diagnosis of breast carcinoma, scheduled for anthracycline chemotherapy, were eligible for inclusion. The main exclusion criteria were AF (excluded because this cardiac arrhythmia will interfere with cardiac imaging scan quality), diabetes, and contraindications to cardiac MRI: significant claustrophobia, allergy to gadolinium chelate contrast; renal insufficiency with an estimated glomerular filtration rate <30 ml/min/1.73 m2; presence of MRI-contraindicated implanted devices (e.g. pacemaker, implanted cardiac defibrillator, cardiac resynchronisation therapy device, cochlear implant); embedded metal objects (e.g. shrapnel); and any other contraindication for cardiac MRI.
Patient Preparation
Serial cardiac imaging was performed in 17 female breast cancer participants at baseline (PET/MRI), 3 months post-doxorubicin (PET/MRI) and 12 months post-doxorubicin (cardiovascular MRI only). Patients were fasted for a minimum of 6 hours followed by 75 g of oral glucose 2 hours prior to the scan. One hour after the oral glucose load, insulin was administered as previously described.22 Blood glucose level was assessed after 15 minutes and subjects were administered 6 mCi of 18F-FDG. Subjects then rested for 1 hour to ensure adequate glucose uptake before PET/MRI image acquisition. This preparation protocol was followed to assess myocardial metabolism through standardised uptake values (SUV) on PET.
Cardiovascular MRI Acquisition
Cardiovascular MRI (CMRI) was performed using a 3.0 Tesla integrated whole-body PET/MRI scanner (Biograph mMR, Siemens Healthcare). To assess LVEF and mass, 12 consecutive 8.5-mm short-axis images and 2-, 3- and 4-chamber long-axis images of the left ventricle (LV) were acquired using a cine balanced steady-state free precession (bSSFP) sequence. Modified look locker (MOLLI) images were acquired for T1 quantification from both mid-ventricular short axis and 4-chamber slices using an 11-image, 17-heartbeat 3-(3)-3-(3)-5 bSSFP sequence.23 T2 mapping was performed using a GeneRalized Autocalibrating Partial Parallel Acquisition sequence on the same midpapillary ventricular short-axis slice as T1 mapping. Tagged images were acquired in three long-axis and three short-axis views (basal, mid, apical). Pixelwise T1 and T2 maps were generated inline on the scanner.
For assessment of contrast enhancement, a total dose of 0.2 mmol/kg gadobutrol (Gadovist, Schering) was injected and 10 minutes after contrast injection, short- and long-axis 2D inversion recovery late gadolinium enhancement images were acquired with an inversion-recovery gradient-echo imaging sequence to evaluate focal myocardial fibrosis. Finally, post-contrast MOLLI T1 mapping was repeated in identical manner as native T1 mapping, 12 minutes after gadolinium injection. This time was consistent across participants.
Cardiac MRI Analysis
CMR images were analysed by two experienced observers (JHC and JB with 3 and 5 years of CMRI experience, respectively) independently, blinded to the clinical data. Cine bSSFP CMR images were analysed using Circle Cardiovascular Imaging 42 (cvi42) user interface analysis software. The endocardium and epicardium of the LV were automatically contoured on all phases of the LV, with manual adjustments made when required. Left ventricular end diastolic volume (LVEDV) and end-systolic volume (LVESV) and LVEF were calculated using summation of discs method. Native and post-contrast blood T1 times were measured from a region of interest (ROI) manually drawn in the centre of the blood pool. Myocardial T1 and T2 values were measured in a septal region of interest per slice in the short-axis view and were reported as a value for the septum but assumed to be representative of the entire heart.
Extracellular volume fraction (ECV) was calculated as ECV=(1–Hematocrit).λ, where:
where T1m and T1b are the T1 values from myocardium and blood, respectively, in both native and post-contrast.
Intracellular mass was calculated as (1-ECV) × LV mass and expressed as g.
An average of the two independent sets of measurements for LV volumes, LV mass, LVEF, T1, T2 and ECV was used as reference standard for baseline and follow-up CMRI findings.
CMR-tagged Image Analysis
Semi-automated analysis of the tagged images was performed using CIM software (CIMTag2D v.8.1.6, Auckland MRI Research Group).24 The grid was aligned to the myocardial tagging planes at end diastolic frame using guide-point modelling.25 End-systolic frame was identified visually, and tags were manually adjusted at some key frames during the cardiac cycle including the end-systole and last frame.26 Longitudinal and circumferential myocardial strains were calculated by the software from the motion of the intersected tag lines in the three long-axis and three short-axis slices, respectively. Global longitudinal strain (GLS) and global circumferential strain (GCS) parameters were derived by averaging the peak strain values of individual segments using 17- and 16-segment models.
PET Image Analysis
PET images were analysed using the PCARD module of PMOD software (v3.6, PMOD Technologies). 18F-FDG uptake was quantified by measuring the mean SUV taken as an average of the automatically segmented entire LV by an experienced senior operator (JOD).
Statistical Analysis
Statistical analysis was performed using commercial statistical software (SPSS Software v.27, IBM). Normality was assessed using Shapiro–Wilk test. Continuous data were described as mean ± SD for parametric data or median inter-quartile ranges for non-parametric data, as appropriate. Categorical data were described as frequencies and percentages. Paired sample t-test or Wilcoxon matched-pair signed rank test was performed to compare changes in CMRI and PET parameters from baseline and 3 months and from baseline and 1 year. A p-value of <0.05 was considered statistically significant.
Results
Seventeen women with breast cancer completed the baseline and 3-month PET/MRI scan, and 14 of them went on to complete the 1-year CMRI scan. The mean age was 50 ± 10 years and mean cumulative dose of doxorubicin was 232 ± 19 mg/m2 (Table 1). A minority of these patients had concomitant risk factors (Table 1).
Baseline versus 3 Months PET-CMRI Parameters
There was a significant reduction in LVEF at 3 months when compared to baseline (62 ± 4% at baseline versus 57 ± 6% at 3 months, p<0.001) and was mainly driven by an increase in LVESV but no change in LVEDV (Table 2). There was an associated reduction in GCS and GLS. There was no evidence of focal fibrosis on the late gadolinium enhancement images. Native myocardial T1 was significantly higher at 3 months (1,260 ± 51 ms at baseline versus 1,304 ± 35 ms at 3 months, p=0.004), with an associated increase in ECV (26 ± 3% at baseline versus 29 ± 3% at 3 months, p=0.012) but there was no change in myocardial T2 (Table 2). There was a reduction in intracellular mass (47 ± 8 g at baseline versus 44 ± 7 g at 3 months, p=0.03) and associated reduction in SUV of 18F-FDG uptake (6.2 ± 2.6 at baseline versus 4.2 ± 2.3 at 3 months, p=0.012).
There was no correlation between both absolute and relative changes in LVEF at 3 months from baseline with change in T1, ECV or SUV. However, there was a positive correlation with relative reduction in SUV and relative reduction in GLS at 3 months (Pearson’s correlation coefficient, 0.54, p=0.032).
Baseline versus 1-year CMRI Parameters
Complete paired 1-year data sets were available in 14 patients (Table 3). LVEF remained lower than baseline at 1 year (62 ± 4% at baseline versus 56 ± 6% at 1 year, p<0.001) with an associated reduction in GCS and GLS. Native myocardial T1 values remained higher than baseline but with no significant difference in T2 and ECV. Intracellular mass was similar to baseline level at 1 year post-treatment. There was no correlation between both absolute and relative changes in LVEF at 1 year from baseline with change in T1 or ECV.
Discussion
In this small pilot study of breast cancer patients, we demonstrated a significant increase in LVESV and decline in LVEF at 3 months post-chemotherapy and this was associated with an increase in myocardial T1 and ECV values but no change in T2 values. There was also an associated reduction in intracellular mass and a reduction in glucose metabolism. These findings suggest the presence of diffuse interstitial fibrosis rather than oedema in patients 3 months post-chemotherapy with doxorubicin. In turn, this impacts on glucose metabolism and was also associated with a reduction in myocardial contractility and LVEF. Of note, the reduction in glucose metabolism was only correlated with a reduction in GLS. In a subset of patients with 1-year follow-up CMRI, intracellular mass recovered to baseline levels but T1 remained higher than baseline. However, GLS and LVEF at 1 year remained lower than baseline. Therefore, doxorubicin chemotherapy is associated with regression of intracellular mass (which would explain the reduction in glucose metabolism) and interstitial fibrosis at 3 months. At 1 year following doxorubicin chemotherapy, there is restoration of intracellular mass by interstitial fibrosis as evidenced by persistently high T1.
GLS is the most robust deformation parameter for speckle-tracking echocardiography and together with LVEF, change in cardiac biomarkers is recommended in current guidelines for the assessment of asymptomatic CTRCD.9 The prognostic value of early measurement of systolic deformation indices in the prediction of subsequent LV systolic function has been evaluated in several studies, both in animals and humans.27,28–31 Our study adds to the literature that doxorubicin chemotherapy leads to early architectural changes in the myocardium, likely due to interstitial fibrosis. There was a reduction in 18F-FDG uptake associated with a reduction in intracellular mass. However, the intracellular mass returned to baseline by 1 year but there was evidence of ongoing interstitial fibrosis. Targeting the early increase in T1 by CMRI with antifibrotic agents rather than the reduction in 18F-FDG uptake by PET with myocardial energetics could be future therapeutic targets against CTRCD and our findings are hypothesis-generating.
Primary and secondary strategies for preventing the onset of anthracycline-induced cardiotoxicity include limiting the anthracycline dose and administering it in divided doses, modifying its delivery using liposomal encapsulation to modify its pharmacokinetics and tissue distribution, co-administering dexrazoxane (which protects topoisomerase 2b) and other agents, such as β-blockers, angiotensin-converting enzyme inhibitors and angiotensin-receptor blockers (reviewed by Vejpongsa and Yeh).32 Mitochondrial impairment, increased apoptosis, dysregulated autophagy and increased fibrosis have also been shown to be crucial players in doxorubicin cardiotoxicity.33 These cellular processes are all linked by one highly conserved intracellular kinase: adenosine monophosphate–activated protein kinase (AMPK).33 AMPK regulates mitochondrial biogenesis via PGC1α signalling, increases oxidative mitochondrial metabolism, decreases apoptosis through inhibition of mTOR signalling, increases autophagy through ULK1 and decreases fibrosis through inhibition of TGF-β signalling.33 Thus, the ability to activate AMPK after treatment with cancer therapeutics may be a crucial factor for decreased risk of cardiomyopathy.34
Our study has shown that structural, metabolic and functional changes can occur in patients by 3 months post-treatment. Whether prophylactic pre-treatment with β-blockers, angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers and SGLT2 inhibitors can mitigate against the development of anthracycline-induced cardiotoxicity needs to be evaluated in future studies. The cardioprotective effect of statin therapy in breast cancer or lymphoma patients undergoing doxorubicin chemotherapy has recently been investigated by Hundley et al.35 There was no difference in LVEF at 2 years in patients treated with statin therapy when compared to those treated with placebo. Of note, there was a 3% reduction in LVEF at 2 years when compared to baseline. This is lower than what was seen in our study (6% reduction) and may be due to a shorter follow-up period and higher risk profile of our patients.
Limitations
There are several limitations of this proof-of-concept, small pilot study. First, due to the small overall sample size, significant longitudinal differences could not be established in SUV, GLS, GCS, ECV and myocardial characterisation between the group which developed CTRCD versus the group that did not develop CTRCD. Second, the patients recruited were at low risk for anthracycline cardiotoxicity given the low prevalence of cardiovascular risk factors, concurrent/previous trastuzumab exposure, and relatively low cumulative doxorubicin dose. Future larger studies should consider recruiting higher-risk patients with a higher projected event rate. Third, we did not perform FDG-PET at 12 months. FDG-PET is expensive and involves radiation exposure. Nevertheless, this might have provided valuable insights into the metabolic changes occurring at 1 year. Therefore, we are unable to confirm whether the reduction in GLS and LVEF at 12 months despite normalisation in ECV and ICM was associated with any changes in glucose metabolism. Last, we did not track changes in cardiac medication post-chemotherapy and therefore we could not comment on whether any treatment changes could have influenced the CMRI findings at 1 year.
Conclusion
Our pilot study has shown that subtle architectural myocardial changes assessed by CMRI occur at 3 months post-doxorubicin chemotherapy, suggesting the presence of diffuse interstitial fibrosis and a reduction in intracellular mass. This was associated with a reduction in myocardial glucose metabolism. The subtle architectural changes by CMRI appeared to recover by 12 months but T1 remained higher than baseline. Doxorubicin chemotherapy therefore likely leads to early interstitial fibrosis and may be a future therapeutic target against CTRCD.
Clinical Perspective
- At 3 months post-doxorubicin chemotherapy we observed diffuse interstitial fibrosis (elevated T1), a reduction in intracellular mass, and a reduction in myocardial glucose metabolism; changes which were associated with a reduction in global longitudinal strain and left ventricular ejection fraction.
- The reduction in left ventricular ejection fraction and global longitudinal strain persisted at 12 months and was associated with persistently elevated T1.
- Doxorubicin chemotherapy therefore likely leads to early interstitial fibrosis and may be a future therapeutic target against cancer therapy-related cardiac dysfunction.