Background
The population of cancer survivors is rapidly growing resulting in a new medical need: the management of long-term complications of anticancer treatments in particular at the cardiovascular level [
1‐
3]. Anthracyclines remain important drugs often mandatory in the treatment of many cancers despite their known cardiotoxicity. Anthracyclines mechanisms of action are multiple and still need to be better understood. They include DNA damage, apoptosis, inflammation, calcium dysregulation and reactive oxygen species production [
4]. The recent European guidelines [
5] recommend both baseline assessment and close monitoring of cardiac function to better prevent cardiotoxicity. Cardiotoxicity induced by anticancer treatment is today defined as any reduction of ejection fraction to below 50% or a > 10% reduction from baseline falling below the lower limit of normal [
5]. There is thus no validated tool to detect cardiotoxicity before any change of cardiac function in clinical routine and that could be of interest in investigating cardioprotective strategies.
Different approaches were previously examined to detect cardiotoxicity before any drop in ejection fraction [
6]. Nuclear imaging and magnetic resonance (MR) imaging are considered as promising tools to detect early cardiac toxicity. Moreover, nuclear imaging tracers allow to visualize molecular mechanisms involved in cardiotoxicity [
2,
7]. [
123I]metaiodobenzylguanidine (MIBG) cardiac uptake has been studied to evaluate cardiac adrenergic innervation [
2,
7]. However, more studies are needed to clarify the role of this tracer. Indeed, initial steps in the progression of heart failure involve hyperadrenergic state that results in reduced neuronal reuptake and downregulation of adrenergic receptor [
8]. [
123I]MIBG imaging could therefore constitute a potential early marker for cardiac damage leading to heart failure.
More recently, [
18F]fluorodeoxyglucose (FDG) has also been proposed as a positron emission tomography (PET) predictive tracer for cardiac toxicity [
9‐
14]. PET imaging offers the advantage to allow more precise image quantification than single photon emission computed tomography/X-ray computed tomography (SPECT/CT) imaging in the clinical setting. Moreover, [
18F]FDG PET imaging is a routine exam performed in most cancer patients. Its possible use for early detection of cardiac toxicity of anticancer treatments is therefore attractive because it would not require any additional exam and radiation exposure. However, literature data are scarce and somewhat contradictory, and the potential of [
18F]FDG PET imaging to detect cardiac toxicity needs to be evaluated more precisely. It can be however hypothesized that modifications of cardiac [
18F]FDG uptake should reflect several mechanisms that could be modified by anticancer treatments: cardiac tissue perfusion, inflammation or energy metabolism changes.
The objectives of the present study were to evaluate whether [18F]FDG and [123I]MIBG were able to detect early anthracycline-induced cardiotoxicity, before any modification of cardiac function. We used an experimental model of doxorubicin-induced cardiotoxicity in rats. Cardiac function was evaluated using MRI which can provide gold standard measurements of the left ventricular end-diastolic and end-systolic volumes. Imaging experiments using PET-MR and SPECT-CT were performed longitudinally in order to monitor in parallel and in the same animals the uptake of [18F]FDG, [123I]MIBG and cardiac function. This experimental design allowed to determine if the possible modifications of [18F]FDG or [123I]MIBG uptake occurred in the same time and are predictive of cardiac function impairment due to doxorubicin.
Materials and methods
Experimental model of doxorubicin-induced cardiotoxicity
All animal studies were conducted in accordance with the legislation on the use of laboratory animals (directive 2010/63/EU) and were approved by an accredited ethical committee (C2ea Grand Campus n°105) and the French ministry of research (authorization #6191). After baseline imaging, Wistar Han male rats (175–200 g, Charles River, France) were randomized into control (n = 5) or doxorubicin group (n = 5). Saline (control group) or doxorubicin (5 mg/kg doxorubicin Accord, Accord Healthcare—France) at 2.5 mL/kg intravenous injections were administered beginning on day 1 and repeated 3 times every 10 days (cumulative dose of 15 mg/kg). Imaging experiments were performed before the first doxorubicin/saline injection and every 2 weeks until the end of the 6-week experimental period. SPECT-CT and PET/MR imaging acquisitions were performed at 24 h intervals. The rats were observed daily, and body weights were monitored during the experimental period. No animals were excluded from the present study. One MR imaging dataset from doxorubicin group at week 2 was excluded because ECG-gating was not reliable.
PET/MR imaging
Simultaneous PET/MR imaging was performed on a fully integrated system (MR Solutions, Guildford, UK) consisting of a 7 T dry magnet (Powerscan MRS-7024-PW) coupled to a SiPM-based dual ring PET system [
15]. Animals from each experimental group were distributed all along the imaging experiment duration to reduce the potential bias.
Before anesthesia for PET/MR imaging, a blood sample was collected to measure glycemia (Novapro glucosemeter, Nova Biomedical, France). PET acquisitions were performed from 15 to 45 min after 20 MBq [18F]FDG intravenous injection (Curium, Dijon, France). The animals were not fasted before [18F]FDG injection, kept under anesthesia during the 15 min uptake time, and heated at 37 °C from anesthesia induction to the end of image acquisition. MRI cardiac cine acquisitions were performed simultaneously with PET acquisitions. Images were respiratory- and cardiac-gated (PC Sam, SAII, Stony Brook, USA). Twelve temporal frames per cardiac cycle were acquired. A spoiled gradient recalled echo-fast low angle shot (FLASH GRE) cine sequence was used with: repetition time: 10 ms, echo time: 3 ms, flip angle: 40°, slice thickness: 1.5 mm, voxel size: 0.23 × 0.23 × 1.5mm3 and 4 signal averages. The ventricle was covered by contiguous short axis slices from basis to apex. MR left ventricular volumes and mass were determined by manual contouring of left ventricle (LV) endocardium and epicardium on short axis slices (AnimHeart, CASIS, Dijon, France) on end-diastolic and end-systolic frames to determine end-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF) and LV mass. MR right ventricular volumes were determined by manual contouring of endocardium on short axis slices (ImageJ, NIH, USA) to determine right ventricular ejection fraction (RVEF). The investigator who performed cardiac function analyses was unaware of the treatment. PET reconstructions were performed for the complete 30-min scan. A 3D region of interest within LV tissue was manually delineated using VivoQuant software (Invicro, USA) and the radioactivity measured and corrected to LV mass (measured on MR images) and injected dose corrected to radioactive decay and expressed as a percentage of injected dose per gram (%ID/g).
SPECT-CT imaging
SPECT-CT imaging was performed the day before PET/MR acquisitions. A Nanospect/CT plus camera (Mediso Ltd., Budapest, Hungary) was used. Rats were injected intravenously with 60 MBq [
123I]MIBG (Adreview, GE Healthcare, Velizy-Villacoublay, France) 4 h before the beginning of the SPECT acquisition. X-ray CT acquisitions (55kVp, 34mAs) were performed first, followed by helical SPECT acquisitions with 70−110 s per projection frame. Iodine-123 photopeak (159 keV) was used with a 20% wide energy window. The CT and SPECT reconstructions were performed using an image processing software provided by Mediso Ltd. The SPECT/CT fusion image was obtained using Vivoquant software (Invicro, USA). Each scan was visually interpreted, and 3D regions of interest corresponding to the heart and mediastinum were manually drawn using VivoQuant software (Invicro, USA) in order to determine their radioactivity content (Additional file
1: Fig. S1). The cardiac region of interest was delineated manually using both SPECT signal and CT cardiac contours. The whole ventricular muscle and cavity were included in the region of interest. Mediastinum was defined as a fixed-size spherical region of interest at the level of the first intercostal space in front of trachea. Heart to mediastinum (H to M) ratio was calculated by dividing the Bq/mm
3 in the myocardial region of interest by the Bq/mm
3 in the mediastinal region of interest.
Cardiac tissue processing: histomorphometry
At 6 weeks, animals were euthanized by an intraperitoneal injection of 140 mg/kg pentobarbital (Euthasol Vet®, Dechra Veterinary products). The heart tissue was collected, weighed after atrial tissue removal and paraffin-embedded for histomorphometry measurements after Hemalun-eosin staining. After slide scanning (Nanozoomer HT 2.0, Hamamatsu photonics K.K., Japan), the LV epicardial and endocardial contours were manually traced on short axis slices obtained in the central third of the heart. LV area (mm2) was determined, and the mean LV thickness (mm) was calculated as a ratio of the LV area over LV external contour length.
Statistics
All statistical analyses were performed using the GraphPad Prism software. Data are expressed as mean ± SEM. The data were compared using a Student’s t-test (histomorphometry and heart weight) or a 2-way ANOVA followed by Bonferroni’s post-hoc tests (body weight, glycemia and imaging results). Correlations analyses were performed using Pearson correlation. P values under 0.05 were considered statistically significant.
Discussion
We showed that both glucose metabolism evaluated by [18F]FDG/PET and cardiac innervation evaluated by [123I]MIBG / SPECT were impacted by a doxorubicin repeated administration in rats. Our work constitutes the first longitudinal study evaluating simultaneously in the same animals, cardiac function, glucose metabolism and sympathetic innervation following anthracycline treatment. In addition, we showed that the modification on [123I]MIBG preceded the change in [18F]FDG uptake and cardiac dysfunction. Altogether, these results suggest that cardiac innervation imaging may constitute a better early marker of cardiac dysfunction associated with anthracycline therapy than glucose metabolism imaging.
Several studies previously examined that [
123I]MIBG uptake changes after anthracycline administration in both human [
8,
16‐
18] and animals [
19‐
21]. In accordance with these data, we showed that the cardiac [
123I]MIBG H to M ratio was markedly lower in doxorubicin treated rats. Moreover, cardiac [
123I]MIBG monitoring showed a different pattern than cardiac function and [
18F]FDG cardiac uptake with a clear decrease from the 2 week timepoint when compared to control group, before any modification of cardiac function parameters. Surprisingly, cardiac [
123I]MIBG uptake regularly increased from baseline to week 6 in our control group. No clear explanation has been found in previously published studies. It can be however hypothesized that since the animals used in the present work were 6–8 weeks old at baseline, as for most of preclinical cardiovascular studies performed in rodents, they were still growing. This result emphasized the major role of carrying out a control group to be able to detect potential effects of a treatment. The mechanisms involved in the decrease of cardiac [
123I]MIBG uptake are not fully understood. It was suggested that destruction of adrenergic nerve tissue caused by oxidative stress associated with doxorubicin administration could play a role [
19]. However, further studies are needed to more thoroughly explore the mechanisms involved.
Moreover, we have shown that while [
18F]FDG cardiac uptake correlated with LV ejection fraction, the [
123I]MIBG H to M ratio did not. These results suggest that the modification of [
18F]FDG cardiac uptake follows cardiac dysfunction impairment and therefore, do not constitute a valuable early marker of cardiotoxicity induced by anthracyclines. Indeed, the variability of baseline [
18F]FDG cardiac uptake in clinical practice could also constitute an obstacle for a possible development of cardiac [
18F]FDG PET imaging to evaluate cardiotoxicity of anticancer agents. Conversely, [
123I]MIBG cardiac SPECT-CT could constitute a good alternative. Indeed, the absence of correlation between LV ejection fraction and [
123I]MIBG uptake, tended to indicate that the time course of changes for these parameters were different, i.e. [
123I]MIBG changes appeared before LV ejection fraction decrease. However, the routine use of [
123I]MIBG cardiac SPECT-CT in clinical practice needs to be discussed. Indeed, quantification of the SPECT signal in clinical practice is delicate. However, the development of new Cadmium-Zinc-Telluride cameras could overcome this limitation thanks to the improved spatial resolution and quantification capabilities [
22]. New PET tracers of cardiac innervation are also under evaluation [
23]. Among the most advanced tracers under development, [
18F]flubrobenguane may be the most promising [
24,
25] with a minimal liver uptake [
24], a good cardiac wall signal-to-noise ratio in both humans and animals [
24,
25], and a stable storage in nerve terminals [
25]. It would be therefore particularly interesting to evaluate this new tracer as an early marker of anticancer therapy cardiotoxicity, especially using PET/MR that could simultaneously evaluate cardiac function and morphology.
Similarly to numerous previously published preclinical studies, we reported impairment of cardiac function and morphology following anthracycline therapy [
26‐
31]. However, it is quite delicate to compare studies because there is no consensus about how to induce cardiac dysfunction following doxorubicin administration in rodents. Indeed, administration routes (intravenous [
9,
14,
19,
28], intraperitoneal [
16,
29,
32,
33], or subcutaneous [
26,
31]), treatment schedule and doses administered (from 5 to 50 mg/kg) varied widely between studies. In the present work, we choose to use intravenous injections of doxorubicin and to repeat the treatment 3 times at 10 days’ interval (15 mg/kg cumulative dose) to better mimic clinical practice. In our experimental conditions, doxorubicin treatment induced cardiac dysfunction with a significant reduction of ejection fraction after 6 weeks which was preceded by a decrease in end-diastolic volume from the 4 week-time point. If ejection fraction decrease is well described in the literature, data on diastolic function are scarce [
34]. Indeed, most studies focused on ejection fraction and did not report changes of systolic or diastolic parameters. To our knowledge, the decrease in end-diastolic volume observed in our study has not been reported before. A close monitoring of diastolic function and volumes should be considered to better detect possible cardiotoxic events related to anthracyclines.
We also reported a significant decrease of cardiac [
18F]FDG uptake in the doxorubicin group compared with the control group at weeks 4 and 6. While [
18F]FDG PET has been suggested as a potential marker for anthracycline-induced cardiotoxicity, few studies have been performed, and results are conflicting. Most publications reported an increased cardiac [
18F]FDG uptake after anthracycline administration [
9,
10,
12,
14,
35,
36], but some results indicated a maintained or decreased [
18F]FDG cardiac uptake [
11,
13,
31]. The discrepancies between studies, and the results observed in our work, may have different explanations. The delay between last anthracycline administration and PET acquisition are not always precised [
11,
12,
35,
36] and could rather reflect acute transient effects such as lipid peroxidation, inflammation or apoptosis [
32,
33] than long-term effects. In most of the preclinical studies, the delay between last doxorubicin administration and [
18F]FDG uptake administration is fixed between 6 days and 2 weeks [
9,
13,
14,
31] excluding acute effect observation. The results reported were, however, not always in agreement. Surprisingly, [
18F]FDG cardiac uptake levels in control animals varied considerably: when we report 9.8 ± 1.0% ID/g (Fig.
4), others observed 1.5 to 2% ID/g [
16] or quite low values [
9,
14]. In accordance with our study, a much higher cardiac uptake of [
18F]FDG was reported by Shen et al
. [
13] with a baseline SUV of 9.4 ± 2.1 in control groups compared with 0.8 in the Bauckneht et al
. study [
14]. Mechanisms of action involved in the decrease in cardiac uptake of [
18F]FDG were not evaluated here, but several hypotheses could be drawn. Firstly, experimental conditions could vary between studies [
9,
13,
14,
31] with animals that can be fasted or not, different types of anesthetics were used, and uptake period could also be variable and performed in awake or anesthetized animals. There would be therefore a need for more precise guidelines for [
18F]FDG cardiac PET imaging in preclinical studies. Secondly, [
18F]FDG cardiac uptake could also be impacted by a decreased myocardial perfusion However, previously reported results [
31] and internal data in another set of animals (Additional file
1: Table 1) showed that cardiac perfusion was not impacted by doxorubicin using [
99mTc]Sestamibi SPECT imaging. We actually showed that [
99mTc]Sestamibi cardiac uptake, which is a tracer of myocardial tissue perfusion, was similar in control and doxorubicin treated rats up to week 6 after the beginning of doxorubicin administration. A general reduction of myocardial metabolism constitutes thus a more probable explanation for our results as previously suggested by other authors [
2,
31]. Finally, one limitation of the present work results from the possible partial volume effect on PET images associated with cardiac remodeling that could not be excluded to play a role since LV thickness was reduced from 2.1 to 1.5 mm corresponding therefore to the range in which this effect may exist.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.