UNIVERSITY OF GENOA
School of Medical and Pharmaceutical Sciences
Ph.D. program in Biotechnology in Translational Medicine
Curriculum: Translational Medicine
XXXII cycle
“DOXORUBICIN EFFECT ON MYOCARDIAL METABOLISM:
A TRANSLATIONAL 18F-FDG PET/CT APPROACH.”
Course coordinator: Prof. Rodolfo Quarto
Supervisor: Prof. Gianmario Sambuceti
Ph.D. candidate: Dr. Matteo Bauckneht
2
Alla piccola Alice,
con l’augurio di rimanere sempre così curiosa.
3 INDEX
1. OVERVIEW ON AIMS AND RATIONALE OF THE PROJECT 2. INTRODUCTION
2.1 Aims and rationale of the project 2.2 Epidemiology
2.3 Definition of Cardiotoxicity
2.4 Imaging Techniques for assessment and surveillance
2.4.1 Echocardiography (2D, 3D, strain, stress echo) 2.4.2 Cardiac Magnetic Resonance Imaging
2.4.3 Multi-gated acquisition with blood pool imaging (MUGA) 2.4.4 Single Photon Emission Computed Tomography (SPECT) 2.4.5 Positron Emission Tomography (PET)
2.4.6 Cardiac catheterization and coronary angiography
3. EXPERIMENTAL SECTION: FIRST STUDY
3.1 Materials and Methods
3.1.1 Animal Experiments 3.1.2 Clinical study 3.1.3 Statistical analysis
3.2 Results
3.2.1 In vivo animal experiments
3.2.2 Overall doxorubicin effect on cardiac metabolism in humans 3.2.3 Myocardial function assessment
3.2.4 Score-based images evaluation
3.3 Discussion
3.3.1 Doxorubicin effect on cardiac metabolism 3.3.2 Doxorubicin effect in HD patients
3.3.3 Score-based vs SUV-based images evaluation
4. EXPERIMENTAL SECTION: SECOND STUDY
4 4.2 Methods 4.2.1 Patient selection 4.2.2 Nuclear imaging 4.2.3 Echocardiography 4.2.4 Statistical analysis 4.3 Results 4.4 Discussion 4.5 Conclusions
5. EXPERIMENTAL SECTION: THIRD STUDY
5.1 Background
5.2 Materials and Methods
5.2.1 Experimental design 5.2.2 Micro-PET imaging
5.2.3 Serum biomarkers and biochemical analyses on myocardial homogenate 5.2.4 Immunohistochemical analysis
5.2.5 FDG kinetics in H9c2 cultured cells exposed to Free-DXR 5.2.6 Statistical analysis
5.3 Results
5.3.1 FDG-PET images analysis
5.3.2 Metabolic analyses and correlation between SUV and H6PD enzymatic
activity in the myocardium.
5.3.3 Immunohistochemical analyses and correlation between SUV and oxidative
load in the myocardium
5.3.4 The divergent effect of Free-DXR on FDG uptake and glucose consumption 5.4 Discussion
5.5 Limitations 5.6 Conclusion
6. CONCLUSIONS AND FUTURE DIRECTIONS
5
6.2 From bedside to bench
6.3 Conclusions and future perspectives (back to bedside)
7. REFERENCES
6
1. OVERVIEW ON AIMS AND RATIONALE OF THE PROJECT
Since the 1960s, striking advancements in antitumor agents have made a significant impact on the survival rates of cancer patients. Chemotherapy-induced cardiotoxicity is an emerging side effect of a number of novel oncologic therapies, spawning the necessity to establish not only prophylactic measures but also vigilant surveillance and treatment options in the management of cancer patients. A multidisciplinary approach between cardiologists and oncologists is vital to spearhead these critical issues. With the introduction of anthracycline chemotherapy (e.g., doxorubicin -DXR-, epirubicin, etc.), now a cornerstone in oncologic treatment, as well as cyclophosphamide, monoclonal antibody therapy (e.g., trastuzumab), and other tyrosine kinase inhibitors, the potential side-effect of antineoplastic regimens adversely affecting cardiac function with the risk of heart failure (HF) has prompted single centers and clinicians to develop various surveillance strategies. However, a standardized approach shared by the different centers is still lacking.
Beginning in the 1970s, endomyocardial biopsy with the assessment of histologic grade based on the Billingham scoring system became the ‘‘gold standard’’ for further evaluating and diagnosing cardiotoxicity related to anthracycline administration [1-4]. However, given its invasive nature and the associated potential complications, endomyocardial biopsy has been progressively replaced by non-invasive cardiac imaging, permitting newer approaches and diagnostic algorithms for monitoring and screening, mainly based on the assessment of left ventricular (LV) ejection fraction (EF), reduction of which during chemotherapy has been directly linked to higher mortality rates [5]. Since LV dysfunction induced by cardiotoxic drugs can occur as
7 early as during the infusion, or as late as decades after the exposure, prompt diagnosis and management of cardiotoxicity require a tailored and integrated approach.
Although underlying mechanisms of the cardiotoxic cascade have not been fully elucidated, in the anthracycline-related cardiac toxicity, the drug interference on the respiratory chain and consequent oxidative stress seem to play a significant role [2,6]. This effect is eventually followed by an enhanced glucose consumption [2,7] paralleled by enhanced myocardial uptake of FDG [8, 9]. On the other hand, in vitro studies documented that neonatal rat ventricular cells treated with 1 μM doxorubicin for 1 hour showed approximately a 2-fold increase in sarcolemma GLUT-1 expression [10]. This effect may be related to a reactive oxygen species induced early lipid peroxidation on GLUT-1 structure. Similarly, it has been in vitro demonstrated that endothelin-1, which is a well-known DXR cardiotoxic mediator, enhances cardiomyocite glucose uptake in a dose-dependent manner stimulating the translocation of insulin-responsive aminopeptidase in GLUT-4 [11]. However, the potential association between this “metabolic impact” of doxorubicin on myocardium and the subsequent development of cardiotoxicity has been only described in one case report [9] and its potential clinical drawback remains uncertain. Based on these considerations, the experimental section of the present thesis is a translational study aimed to verify whether response of cardiac FDG uptake to DXR might actually predict a late cardiotoxic effect. To this purpose, we first verified the dose-dependent nature of doxorubicin action on myocardial metabolism analyzing a series of cancer mouse models previously studied in our lab by microPET scanning [12]. Concurrently, we evaluated the serial PET/CT scans obtained in a cohort of patients with Hodgkin’s Disease (HD) to define the time sequence of DXR metabolic effect and to verify its
8 possible clinical correlates. Obtained results were reported in the former two clinical studies. In the third experimental study, we aimed to verify the correlation between the “myocardial metabolic impact” of DXR and the amount of oxidative stress, which is considered one of the main determinants of the anthracycline cardiotoxic cascade. Moreover, we verified whether this early biomarker identifies the potential benefit of drug delivery systems such as Liposomal DXR able to reduce drug accumulation in the myocardium.
9
2. INTRODUCTION 2.1 Epidemiology
One of the earliest observations of cancer therapy-related cardiotoxicity occurred in 1967 in two pediatric patients treated with daunomycin at St. Jude Children’s Research Hospital in Tennessee. Both patients developed acute HF and subsequently died [13]. A similar cardiotoxic effect was observed in the 1970s in adult cancer patients treated with adriamycin, with the development of HF [14]. In an analysis of 5-year survivors from the Childhood Cancer Survivor Study (involving patients diagnosed with cancer at an age <21 years, who were followed for a median of 24.5 years after diagnosis, along with 4,301 of their siblings), cancer survivors had a significantly greater incidence of a severe, disabling, life-threatening, or fatal health condition, with risk increasing proportionally to age beyond 35 years [15]. Importantly, in addition to the risk of developing a subsequent malignancy, the heart was the most commonly affected organ among childhood cancer survivors≥35 years old, with a hazard ratio of 7.9 [95% CI 5.4-11.6] for risk of cardiac involvement. Furthermore, HF [HR 11.4, 95% CI 4.7-27.3], stroke [HR 7.0, 95% CI3.3-14.8], and myocardial infarction [HR 5.0, 95% CI 3.0-8.3] were the most common conditions, thereby highlighting the magnitude of cardiotoxicity among cancer survivors. Among adult cancer patients, determination of the incidence and prevalence of CV disease attributable to cancer therapy is challenging, as pre-existing CV risk factors or CV disease are confounding factors. One of the most-studied adult cancer groups is the breast cancer survivor group, in which CV disease is recognized as a competing cause of morbidity and mortality. In this population, the cumulative incidence of cardiotoxicity is estimated to be as high as 33% after certain adjuvant breast cancer therapies [16]. Apart from
10 chemotherapy, radiation therapy contributes independently to the development of cardiotoxicity in cancer patients, with an estimated 10-30% incidence of radiation-induced cardiac disease for 5-10 years post-treatment [17]. Among cancer survivors, cardiotoxicity is reported to be the second leading cause of morbidity and mortality [18] with secondary malignancies cited as the primary cause of mortality.
2.2 Definition of Cardiotoxicity
The definition of cardiotoxicity lacks consensus across the professional societies, with a generic description by various cancer groups, such as the National Comprehensive Cancer Network’s [NCCN] limited definition: ‘‘Cardiac toxicity is damage to the heart by harmful chemicals’’15 and the National Cancer Institute’s [NCI] definition as ‘‘toxicity that affects the heart’’ [19]. While cancer therapy may cause adverse CV effects other than HF, cardiotoxicity is frequently defined in the limited context of chemotherapy-related cardiac dysfunction. Cancer therapeutics-related cardiac dysfunction, as defined by the American Society of Echocardiography (ASE) and European Association of Cardiovascular Imaging (EACVI), is a decrease in the LVEF of >10%, to a value <53% (normal reference value for 2D echocardiography), with subsequent re-confirmation on a repeat study within 2-3 weeks. In addition to being symptomatic or asymptomatic, such LV dysfunction may or may not be reversible. Another characterization of cardiotoxicity (also focusing on a decline in LVEF) is the definition by the Cardiac Review and Evaluation Committee (CREC), which supervises clinical trials of trastuzumab [20]. Based on the CREC definition, trastuzumab-induced cardiac dysfunction is characterized by a global reduction in
11 LVEF, of at least 5%, to <55%, accompanied by signs or symptoms of HF; or an asymptomatic decline in LVEF by at least 10%, to <55% [21].
While various research papers have been published on cardiac monitoring during chemotherapy [22, 23] it remains problematic to accurately compare collective evidence-based data due to different cancer therapeutics-related cardiac dysfunction definitions and underlying research methodology. Limited data are available, in this regard, to devise and articulate consensus recommendation guidelines on the most appropriate screening and surveillance strategy [17, 24].
Cancer therapeutics-related cardiac dysfunction is classified into two types: Type I and Type II. The characteristic agent implicated in Type I is doxorubicin, with cumulative dose-related effects observed. Type I cancer therapeutics-related cardiac dysfunction results in permanent, irreversible damage, with ultrastructural changes (such as myofibrillar disarray), and high risk of recurrent or worsening cardiac dysfunction with re-introduction of the offending agent. In contrast, Type II cancer therapeutics-related cardiac dysfunction is typically observed with trastuzumab, and the damage is usually reversible and unrelated to dose, without any apparent ultrastructural changes
12 [25]. In addition to type, cancer therapeutics-related cardiac dysfunction may be further categorized based on reversibility [26]. The main characteristics of the two types of cancer therapeutics-related cardiac dysfunction are summarized in Table 1. The risk of developing cancer-therapy related CV toxicity is influenced by several factors, including the patient’s characteristics, cancer type, as well as anatomic location of the cancer, and cancer therapy [the specific drug(s) or radiation]. Patient-related factors include age, preexisting CV disease, CV risk factors (obesity, diabetes, tobacco or alcohol use, other substance abuse, inactivity or sedentary lifestyle, diet, body weight), metabolic abnormalities, hypersensitivity to the drugs; and prior chemotherapy or radiation therapy [27-29]. Cancer-related factors comprise the anatomic location (lungs, breast), which may entail radiation therapy to an area in close vicinity to the heart, and the type and stage of cancer (breast, lung, colorectal, non-Hodgkin
Table 1: Characteristics of Type I and Type II cancer therapeutics-related cardiac dysfunction. Type I cancer therapeutics-related
cardiac dysfunction
Type II cancer therapeutics-related cardiac dysfunction
Characteristic agent Doxorubicin Trastuzumab
Mechanism of cardiac dysfunction
Free radical formation, oxidative
stress/damage Blocked erbB2 signaling Dose-dependency Cumulative, dose-related No dose-related Ultrastructural changes Vacuoles, myofibrillar disarray and dropout
necrosis (changes resolve overtime) No apparent ultrastructural abnormalities
Clinical course, response to therapy
Underlying damage is thought to be permanent and irreversible. Later recurrence may be related to sequential
cardiac stress
High likelihood of recovery (to or near baseline) generally in 2-4 months. Reversible with discontinuation of
Trastuzumab End-result Myocardial damage Myocardial dysfunction
Effect of rechallenge
High risk of recurrence of cardiac dysfunction which may deteriorate into
end-stage HF.
Relative safety of rechallenge. Further studies are needed in this area.
13 lymphoma), which dictate or guide the classes of agents used to treat those types of cancer. Drug-related factors include the agent, the dose administered (including the session dose and the cumulative dose), schedule of delivery, route of administration, the combination of drugs administered, the sequence of administration and the timing between these drugs, as well as the association with radiotherapy [27, 29].
Although HF or cardiomyopathy is the most commonly recognized and discussed cardiac complication of cancer therapy, CV toxicity may manifest in other ways, including myocardial depression, myocardial ischemia, hypertension, hypotension, arrhythmias (including bradycardia, tachyarrhythmias, and AV blocks), QT prolongation or torsades de pointes, hemorrhagic myocarditis, pericarditis, pericardial effusion, and thromboembolism [27]. A summarized description of the different types of cardiac damages according to specific antineoplastic agents is provided in Table 2. On the other hand, radiation-induced HF may occur months to later than a decade after the completion of radiotherapy and comprises structural abnormalities (such as valvular heart disease), circulatory problems (such as coronary artery disease [CAD], myocardial infarction, and carotid artery disease), electrical abnormalities (including rhythm and conduction abnormalities), pericarditis, and pericardial effusion [30]. The location of irradiation influences the site of peripheral artery disease development. Stroke, transient ischemic attack (TIA), and carotid artery disease are associated with head and neck radiation, as well as supraclavicular and mediastinal radiation, which also increased the risk of subclavian artery disease. Renal artery and lower extremity arterial disease may occur due to vessel injury related to abdominal and pelvic irradiation [31].
14 Risk factors for radiation-induced HD include daily radiation doses >2 Gy, cumulative radiation dose >30 Gy, patient age <50 years, anterior or left chest irradiation, and concomitant chemotherapy, such as anthracyclines [31]. Radiation induces CV damage via injury to the endothelium, micro- and macrovascular systems, with resultant myocardial ischemia and fibrosis, pericardial effusion and constriction, and valvular damage.
15 Table 2: Common chemotherapy agents associated with cardiovascular toxicity.
Class of agent Antineoplastic indications Cardiotoxicity Anthracyclines
(Doxorubicin, Daunorubicin, Epirubicin, Idrarubicin)
Breast, liver, kidney, ovarian cancer, and hematological malignancies
(Leukemia, HD, Non-Hodgkin lymphoma)
Heart failure, left ventricular dysfunction
Alkylating agents
(Cyclophosphamide,Ifosphamide)
Leukemia, Lymphoma, Testicular cancer
Heart failure, asymptomatic pericardial effusions, myo-pericarditis, myocardial
ischemia, angina
Platinum based agents
(Cisplatin)
Testicular, bladder, lung, esophageal, stomach and ovarian cancers
Angina, Myocardial Infarction, Raynaud’s, Raynaud’s stroke, PAD,
Angina, vasospasm, coronary thrombosis, CAD progression, thromboembolism (DVT/pulmonary embolism), late CV manifestation of
hypertension, LV hypertrophy, myocardial ischemia
Fluoropyrimidines
(5-fluorouracil, Capecitabine, Gemcitabine)
Solid tumors such as lung, esophageal, stomach, pancreatic, colorectal, ovarian and breast cancer
Chest pain, myocardial infarction (MI), arrhythmia, HF, cardiogenic shock and sudden death, vasospasm,
Takotsubo cardiomyopathy
Taxanes
(Paclitaxel, Docetaxel) Ovarian, breast and lung cancer
HTN, DVT/pulmonary embolism, myocardial ischemia, angina, vasospasm, myocardial infarction,
arrhythmias (including asymptomatic sinus bradycardia),
LV dysfunction, heart failure
Monoclonal antibodies
(Trastuzumab, Pertuzumab, Bevacizumab)
Breast cancer, stomach, esophageal, colorectal, lung, glioblastoma, kidney, cervical, and ovarian cancer
Left ventricular dysfunction; heart failure, angina, myocardial
infarction, Takotsubo cardiomyopathy, Raynaud’s stroke
Tyrosine kinase inhibitors
(Lapatinib, Sunitinib, Sorafenib)
Breast, kidney, liver, thyroid, gastrointestinal stromal tumors
(GIST) and pancreatic neuroendocrine tumors
QT prolongation, LV dysfunction, Angina, myocardial infarction,
Takotsubo cardiomyopathy, vasospasm
Proteasome inhibitors
(Bortezomib, Carfilzomib)
Multiple myeloma and mantle cell lymphoma
Vasospasm, angina, Left ventricular dysfunction, heart failure, hypertension, acute myocardial
16
2.3 Imaging Techniques for assessment and surveillance.
With the advent of an integrated approach for the assessment and monitoring of cardiac function in patients receiving chemotherapeutic agents, multimodality imaging is quickly becoming the standard of care in diagnosis, staging, treatment planning, and response assessment. Primary structural and morphologic imaging techniques include echocardiography, cardiac computed tomography (CT), and cardiac magnetic resonance imaging (CMRI), with functional assessment performed utilizing positron emission tomography (PET), functional MRI, and single-photon emission computed tomography (SPECT) imaging. Essentially, multimodality imaging seeks to use two or more imaging types to provide spatial co-localization of complementary information, like cardiac structure and function, with serial examinations to assess for acute and chronic temporal changes. This approach provides a synergy of valuable information, each with its own merits and disadvantages.
LVEF is highly validated as the baseline index for screening and longitudinal follow-up care for the ascertainment of cancer therapeutics-related cardiac dysfunction [32, 33]. Although the original gold standard for diagnosing and following cancer therapeutics-related cardiac dysfunction was serial invasive endomyocardial biopsies, 2D-echocardiography (2DE) is now widely accepted as the modern mainstay modality. Due to its superior tissue characterization and spatial resolution, CMR is the imaging standard for the measurement of LVEF and ventricular volumes. 2DE, however, with its complementary modalities, has modernized clinicians’ modus operandi in regard to cost-effectiveness and clinical prognostication. It is more applicable for serial evaluation of LV structure and function with newer supplemental techniques, including tissue-Doppler imaging (TDI), contrast echocardiography,
3D-17 echocardiography (3DE), and speckle-tracking echocardiography (STE). Another imaging modality used for further evaluation of cancer therapeutics-related cardiac dysfunction is multi-gated acquisition with blood pool imaging (MUGA). In addition to providing sensitive, accurate, and highly reproducible LVEF measures, non-invasive nuclear imaging techniques, such as SPECT and PET, are able to use specific radiotracers in the detection of subclinical biological and physiological cancer therapeutics-related cardiac anomalies, which precede the occurrence of clinical, anatomic abnormalities [34].
Since clinical decisions for continuation or cessation of certain life-saving oncologic therapy are currently based on accurate changes in LVEF, it is exceedingly important for the chosen imaging modality to be as accurate as possible, while maintaining minimal temporal variability, such that changes reflect true cardiotoxicity.
2.3.1 Echocardiography (2D, 3D, strain, stress echo).
Echocardiography is considered the principal imaging modality in the monitoring and evaluation of cancer therapeutics-related cardiac dysfunction. Beyond the advantages of widespread accessibility/availability, lack of radiation exposure/radioactive agents, and lack of potentially nephrotoxic agents, echocardiography facilitates assessment of both systolic and diastolic function of the LV, along with the assessment of other chambers, valves, pericardium, and aortic root [26]. Supplemental measurements and techniques such as 3DE and strain further enhance the capability of echocardiography to aid in early detection and surveillance of cardiotoxicity, especially LV dysfunction.
18 Despite the recognized status of EF as the primary measurement and parameter defining cancer therapeutics-related cardiac dysfunction, there is a lack of consensus on the threshold value, which defines cardiotoxicity. Furthermore, subtle changes in LV contractility are often not detected by 2DE. In 2014, the ASE and EACVI jointly issued an expert consensus document on the evaluation of patients during and after cancer therapy. The document identified echocardiography as the method of choice for evaluation of patients before, during, and after cancer therapy, with LVEF calculation ideally measured by 3DE along with global longitudinal strain (GLS) by STE. If 3DE is not available, then 2DE with modified biplane Simpson’s technique is the preferred method for LVEF assessment. The 2015 recommendations for chamber quantification from the ASE and the EACVI identified the biplane method of disks (modified Simpson’s rule) as the currently recommended 2D method to assess LV EF [52 JNC]. Unlike LV systolic function, LV diastolic function and right ventricular function have not been found to be prognostic of cardiac dysfunction secondary to cancer therapy. Nonetheless, conventional assessment of these parameters is recommended in the current expert consensus document by the ASE and the EACVI [26]. Global longitudinal strain (GLS) and global circumferential strain (GCS) were found to closely correlate with EF by CMR, which is considered to be the gold standard for assessment of cardiac function. In anthracycline-treated pediatric cancer survivors, 3DE had the highest sensitivity in the detection of subclinical myocardial dysfunction (EF <55%) identified by CMR [35]. Compared to 2DE, non-contrast 3DE exhibited lower temporal variability in sequential EF assessments, with better intra- and inter-observer, as well as test-retest variability, in an analysis of 56 patients with breast cancer followed with sequential echocardiograms at baseline, then every 3 months for
19 1 year. EF assessment varied up to 10-13% by 2DE approach versus 5-6% with non-contrast 3DE techniques (with no improvement in EF variability with non-contrast administration) [36].
Echocardiographic assessment of cardiac mechanics incorporates myocardial deformation, i.e., the fractional change in the length of a myocardial segment reported as strain. Usually expressed as a percentage, strain is unitless and may reflect lengthening (positive value) or shortening (negative value). Strain rate is the rate of change in strain with respect to time and is usually expressed as 1/s or s-1. GLS, which is calculated by averaging the individual myocardial wall segmental strain values, refers to the average longitudinal strain in the entire myocardium [37]. Altered myocardial deformation, as assessed by strain, has been documented to detect cardiac dysfunction prior to any noticeable reduction in LV systolic function. Strain may be approximated by TDI or STE. Although interventricular septal longitudinal strain rate (using TDI) appears to be most consistently reduced during chemotherapy, the most clinically relevant data on predicting cardiotoxicity have been based on STE-based strain [38]. Using TDI to assess strain, peak systolic longitudinal strain rate is reportedly the most consistent measurement for early detection of myocardial dysfunction during cancer therapy, whereas peak systolic GLS is thought to be the best measure when STE is utilized. Thavendiranathan et al. identified a 10% to 15% early reduction in GLS by STE (during cancer therapy), as the most useful parameter for the early prediction of subsequent cardiotoxicity, manifesting as symptomatic or asymptomatic LV systolic dysfunction [38]. While the majority of studies have focused on chemotherapy-induced cardiac dysfunction, it is known that radiation therapy to the chest also affects cardiac function, with the highest-risk patients being
20 those who received radiation therapy to the left chest. Myocardial deformation changes have been observed in those myocardial segments exposed to the highest radiation doses. Thus, strain assessment is recognized as a valuable technique for the early detection and surveillance of cancer therapeutics-related cardiac dysfunction, since a drop in LVEF is often a late sign of cardiotoxicity [38].
2.3.2 Cardiac Magnetic Resonance Imaging.
Due to its excellent temporal and spatial resolutions for non-invasive evaluation of LV volumes and LVEF, CMR is considered the gold standard for evaluation of LV systolic function. Compared with nuclear imaging, CMR is at least equivalent, if not superior, in the detection of myocardial ischemia [26]. In addition to the evaluation of specific cardiomyopathies, such as dilated cardiomyopathy (including that secondary to cardiotoxic therapy), CMR is helpful in the characterization of intracardiac and extracardiac masses, along with functional and structural assessments of the pericardium and constrictive physiology [39]. As such, CMR is recognized by the American College of Cardiology/American Heart Association with a well-documented role in screening for antineoplastic-related toxicity and adverse cardiac side-effects leading to cardiomyopathy [40]. Utilizing its superior tissue characterization capacity, it has the distinctive ability to accurately define cardiac anatomy and ventricular function (without radiation or iodinated contrast exposure), even in difficult patients, such as those morbidly obese or with pulmonary disease, in whom acoustic windows may be suboptimal using echocardiography [41]. Another area in which CMR excels is in the detection of myocyte loss as a result of cardiotoxicity, which can be indicated
21 by a decrease in LV mass index. CMR remains the gold standard for this measurement, as it provides more accurate quantification than echocardiography [42]. In breast cancer patients receiving trastuzumab, cardiac toxicity, as assessed by CMR, manifested as late gadolinium enhancement in the mid myocardium associated with a decline in LVEF, as well as increased LV end-systolic and end-diastolic volumes [43]. Since underlying CAD and prior myocardial infarction increase the likelihood of cardiotoxicity, CMR can be efficiently used to improve risk stratification by allowing the detection of subendocardial infarcts, which may be missed on SPECT imaging and may be too small to cause wall motion abnormalities noticeable by echocardiography [44].
Despite its superiority in spatial resolution and accuracy in the measurement of ventricular volumes and EF, the relatively higher cost of serial CMR, as compared to echocardiography, must be taken into consideration. Other significant disadvantages include limited center availability (in comparison to echocardiography and SPECT), as well as the relative contraindication in patients with metallic devices and renal insufficiency. Although newer lower-risk gadolinium-based contrast agents have been developed, multiple consensus guidelines, including those from the European Society for Urological Radiology, the American College of Radiology, and the UK Royal College of Radiologists, recommend against the use of gadolinium-based contrast-enhanced MRI in acute or chronic renal insufficiency with Glomerular Filtration Rate (GFR) <30 mL/(min*1.73 m2) and caution in those with stage 3 chronic kidney disease (GFR 30-59) [45, 46]. Historically, CMR has been contraindicated in patients with an implantable cardiac device because of the potential interference with device function. Nonetheless, a retrospective review of MRI scanning in 109 patients with cardiac
22 devices documented no device or lead failures [47]. A subsequent multicenter prospective study, the MagnaSafe Registry, to determine the risks of non-thoracic 1.5 Tesla MRI scan in patients with implanted pacemakers and Implantable Cardioverter- Defibrillators (ICDs), reported no device failure, generator/lead replacement, induced arrhythmia, or loss of capture in the first 829 cases [48], nor in the first 1189 cases [49]. A study of 68 MRI scans in pacemaker patients (with conditional vs. MR-unsafe pacing systems) documented clinical safety of MRI in both groups, with no significant differences between those undergoing thoracic MRI compared to other scanned areas [50]. The subsequent availability of MRI-safe implantable cardiac devices has facilitated the utilization of MRI (including CMR) in the clinical setting.
2.3.3 Multi-gated acquisition with blood pool imaging (MUGA).
Whilst the diagnostic approach to detection of cardiotoxicity has historically involved serial measurements of LVEF by MUGA for surveillance of reduction in LVEF, such decline is considered a late manifestation of cardiac damage. Other forms of nuclear imaging may serve as non-invasive modalities with the potential to detect myocardial damage on a subclinical level.
Since 1980, MUGA (also known as gated-equilibrium radionuclide ventriculography [RVG, RNV, or radionuclide angiography]) has been extensively validated in multiple clinical trials studying anthracycline chemotherapy in the treatment of a wide variety of pediatric and adult tumors. Advantages include extensive long-term follow-up, wide availability, consistently reproducible measurements, and comparable cost in relation to alternative modalities, making it one of the preferred modus operandi for the
23 detection of cardiotoxicity [51, 52]. While MUGA has a high correlation with CMR and 3DE, with demonstrable superiority over 2DE in measuring LVEF [53-55]. Van Royen et al. demonstrated in a group of 73 patients that while echocardiography and MUGA are both accurate in their ability to assess LVEF, MUGA had more consistent Intra- and inter-observer interpretation. Conversely, EF variability assessed visually by echocardiography ranged from 13 to 17%, suggesting that MUGA, as compared to 2DE, was the imaging modality of choice when a more accurate estimation of LVEF was needed, in order to direct clinical decision [56]. Of note, single head gamma cameras were used in the initial LVEF reproducibility studies comparing MUGA to echocardiography; given significant differences between currently used dual-head gamma cameras versus the previously utilized single head gamma cameras, comparability of current MUGA to 2DE and 3DE is unknown [26]. Although a significant advantage of MUGA is the reliability independent of body habitus or acoustic windows, this is counterbalanced by several drawbacks, including the inability to evaluate for pericarditis or valvular conditions, frequent soft-tissue attenuation artifacts, and the risk of significant cumulative ionizing radiation exposure (approximately 10mSv per study) related to serial imaging [57].
2.3.4 Single Photon Emission Computed Tomography (SPECT).
SPECT has been used with several radiotracers such as 99mTc-erythrocyte, 111 In-antimyosin, 123I-MIBG, 111In-Tz, 99mTc-annexin V, and 123I-BMIPP, each of which offers different functional potential applications [34].
24 Gated blood-pool SPECT allows for determination of LVEF, right ventricular EF, and wall motion abnormalities [58], with excellent concordance with planar MUGA-derived LVEF [59]. Nonetheless, a caveat when interpreting LVEF assessment by SPECT is that clinical validation studies have shown variations in LVEF based on the processing software in use. More specifically, an underestimation of LVEF (33 ± 13%) was documented when using the Quantitative Gated SPECT (QBS) processing software [60], compared to the MUGA-derived (41 ± 14%) and echocardiography-derived LVEF assessments (37 ± 15%), potentially leading to early inappropriate discontinuation of life-saving cancer treatment. By contrast, the use of the 4D-MSPECT processing software resulted in an overestimation of LVEF, compared to MUGA [61]. When compared to CMR-derived LVEF, QBS significantly underestimated LVEF in several studies [62, 63]. However, in patients with dilated cardiomyopathy, LVEF was overestimated by QBS, Emory Cardiac Toolbox (ECTB), and 4D-MSPECT, compared with CMR [64]. Thus, limited interchangeability between software programs, with variations between algorithms for LVEF estimation, needs to be considered when interpreting studies using SPECT.
In patients treated with anthracycline chemotherapy, 111In-antimyosin (which binds to
intracellular myosin exposed after myocardial cell damage and necrosis have occurred) imaging have documented antimyosin uptake, signifying the presence of myocyte cell damage, and suggesting a potential role for 111In-antimyosin in the subclinical evaluation of anthracycline-induced LV dysfunction [65, 66]. Indeed, increased uptake of 111In-antimyosin in anthracycline treated breast cancer patients
allowed early identification of those patients at higher risk of developing a decline in LVEF and symptoms of HF [67]. Similarly, in addition to the early detection of
25 cardiotoxicity in adults [68, 69] myocardial uptake of 111In-antimyosin was documented to precede LV systolic dysfunction in children treated with anthracycline therapy [69].
123I-metaiodobenzylguanidine (MIBG), a norepinephrine analog [70], is another
radiotracer shown to predict the occurrence of chemotherapy-induced HF. Since cardiotoxicity induces a compensatory adrenergic response with an upregulation of the sympathetic nervous system and renin-angiotensin system [71] the use of 123IMIBG, which mimics uptake, storage, and release mechanisms of endogenous norepinephrine, can successfully identify areas of myocardial adrenergic derangement and consequently, patients at risk for cardiotoxicity [72].
Radiolabeled trastuzumab scintigraphy, using 111In-trastuzumab, was tested experimentally and clinically, to verify whether it could provide complementary information to support HER2-targeted therapy in metastatic breast cancer. A preliminary study conducted in 20 patients suggested that a pretreatment scan with radiolabeled trastuzumab was helpful in predicting both cardiotoxicity (absent in patients without myocardial uptake of 111In-trastuzumab) and therapeutic response (positive response in patients with tumor uptake of 111In-trastuzumab) [73]. However, a subsequent study in 15 patients failed to support the previous findings, as none of the patients who developed severe LV dysfunction showed myocardial uptake of 111
In-trastuzumab on initial scans [74].
26 The focus of PET in cancer patients is to provide insight into the diagnosis of metastatic lesions and the response to oncologic therapy. Employing fluorine-18-fluorodeoxyglucose (FDG), PET imaging can be used to monitor treatment response for cardiac lymphoma, as well as for metastatic pericardial involvement. Given the high diagnostic potential in the field of HF, in the last years, hybrid imaging became appealing in cardiac dysfunction monitoring in cancer patients receiving chemo or radiotherapy, including PET/CT and PET/MR technologies [75]. However, there is currently a paucity of data with no randomized-controlled trials regarding approaches to detect cardiotoxicity. In this regard, a few studies recently pointed out that Doxorubicin can specifically affect myocardial metabolic processes altering glucose uptake patterns.
As described above, among the several mechanisms proposed to explain the doxorubicin cardiotoxic cascade, oxidative stress seems to play a significant role. On the other hand, endothelin-1 production has been reported to contribute to myocardial damage. Both these pathways have been reported to impact on myocardial glucose metabolism profoundly. Accordingly, in vitro studies documented that neonatal rat ventricular cells treated with 1 μM doxorubicin for 1 hour showed approximately a two-fold increase in sarcolemma GLUT-1 expression [10]. This effect might be related to reactive oxygen species, which can induce early lipid peroxidation on GLUT-1 structure. Similarly, it has been demonstrated that endothelin-1 enhances cardiomyocyte glucose uptake in a dose-dependent manner stimulating the translocation of insulin-responsive aminopeptidase in GLUT-4 [11]. Moreover, varied expression of neuregulin (NRG), an endogenous cardioprotective protein-ligand to the erbB2 receptor, is considered the cause for this variable susceptibility of patients [76,
27 77]. NRG-erbB2 binding is believed to regulate doxorubicin uptake into cardiac myocytes via multidrug-resistance protein by activating PI3k/Akt signaling pathway [77]. NRG has also been demonstrated to increase the glucose uptake, GLUT translocation, and expression on muscle cells [78, 79]. Possibly, the combined effect of glycolytic adaptation to mitochondrial damage and NRG stimulation may result in the increase in myocardial FDG uptake, although no specific data are currently available in the literature.
Figure 1: Case report from Gorla et al. [9] showing a substantial increase in myocardial FDG uptake on interim PET/CT as a possible early sign of doxorubicin-Induced cardiotoxicity.
28 In this line, FDG PET/CT might offer a viable window on metabolic changes within the myocardium, possibly identifying ongoing myocardial damage. In a recent paper by Gorla and colleagues [9], the case of a 30-year-old treated with anthracyclines for Hodgkin’s lymphoma is reported (Figure 1). Baseline and interim FDG PET/CT were performed before the start of chemotherapy and after 4 cycles of chemotherapy with ABVD regimen, respectively. At the second scan, FDG avid left cervical, supraclavicular, and mediastinal lymphadenopathy noted at baseline showed complete resolution. However, a significant increase in myocardial FDG uptake was pointed out between the 2 studies. Patient preparation was not significantly different between the two studies regarding diet, duration of fasting (~6 hours), or blood glucose levels (73 and 82 mg/dL, respectively) that otherwise could have influenced the myocardial uptake. The patient complained of mild dyspnea at the time of the interim PET, which progressed over subsequent weeks to NYHA Grade III. Evaluation with echocardiography revealed global hypokinesia with dilated ventricular cavities. A comparison with baseline echocardiography revealed significant deterioration in cardiac function (LVEF ~38% vs. 58% at the baseline), which was ascribed to doxorubicin-induced cardiotoxicity. Subsequently, treatment was changed to an alternate regimen with additional supportive therapy.
Similarly, Borde and colleagues [8] highlighted a change in myocardial Standardized Uptake Value (SUV) in 12/18 patients who underwent FDG PET/CT twice for staging and re-staging after ABVD for HD, respectively.
Similar findings have also been proposed in patients after radiation-induced cardiac damage. In the study by Yan et al. [80], the anterior myocardium of 12 beagles received radiotherapy locally with a single X-ray dose of 20 Grey. 18F-FDG cardiac
29 PET/CT was performed at baseline and 3 months after radiation. Two subgroups of the animals underwent two protocols before PET/CT: 12 hours of fasting, followed by a high-fat diet. Regions of interest were drawn on the irradiation and the non-irradiation fields to obtain their maximal SUV. Finally, histopathological changes were identified by light and electron microscopy. The authors showed that the ratio between the SUV of the irradiation to the non-irradiation fields significantly improved after irradiation, particularly in animals who underwent dietary preparation. The pathology of the irradiated myocardium showed obvious perivascular fibrosis and changes in mitochondrial vacuoles. An emblematic example from their experimental case series is reported in Figure 2. Even if radiation- and doxorubicin-induced myocardial damage are supported by different pathophysiological events, this paper reported that a mitochondrial injury might affect myocardial glucose metabolism and, subsequently, FDG accumulation. This topic is particularly challenging because a change in myocardial FDG uptake between subsequent PET/CT scans is often overlooked, considering its physiological variability. Moreover, without a specific patient preparation, myocardial FDG uptake might be influenced by several confounding factors. Accordingly, physiological FDG uptake by the left ventricular myocardium can hamper image interpretation due to the blurring effect caused by cardiac cycle and respiratory activity. Similarly, the marked and often unpredictable heterogeneity of its distribution can both mask or potentially simulate the presence of a “metabolic change”. For these reasons, minimizing glucose metabolism of myocardial cells as much as possible is thus mandatory when myocardial FDG PET/CT is performed to investigate infective or inflammatory myocardial diseases. To achieve this aim, a prolonged reduction in sugar intake decreases serum glucose and
30 insulin levels markedly lowering glucose consumption of myocardium by lowering GLUT4 docking to the sarcolemma can be theoretically proposed. Cardiomyocyte intracellular nutrient asset thus shifts towards a reduction in glucose availability facing an increased concentration in free fatty acid derivatives. This modification switches cardiac metabolism towards beta-oxidation and activates the peroxisome proliferator-activated receptor-α that transcriptionally promotes the expression of enzymes governing fatty acid metabolism and inhibits that of glycolytic enzymes causing a further reduction in cardiomyocyte utilization of glucose. The relevance of this pathway in allowing an adequate “target-to-background” ratio has been recently reported [81]: preceding tracer injection by 12 hours of carbohydrate-restricted, fat and protein allowed diet followed by 12 hours fasting period resulted in an “absent” cardiac uptake in 54% of patients with respect to 28% of subjects studied with the standard 6 hours fasting protocol. Adding a single bolus of heparin (50 IU/Kg) – whose lipolytic activity can induce a 5-fold increase in blood free fatty acid levels – prior to FDG administration further increased this response up to 88%. Although still suboptimal, this feature actually approaches the clinical applicability. Nevertheless, the repeatability of this performance has not been thoroughly tested. Accordingly, further studies aiming to optimize dietary preparation for this purpose should be encouraged to define better which patient preparation procedure can optimize the diagnostic potential of myocardial FDG PET/CT imaging. Based on these considerations, it is essential to specify that the currently available data about the potential use of myocardial FDG uptake as an early predictor of doxorubicin cardiotoxicity are mainly retrospective and obtained from whole-body FDG PET/CT scans, performed in the frame of staging or re-staging FDG-avid cancer lesions. Accordingly, these data
31 should be carefully evaluated and discussed. On the other hand, the advent of molecular imaging recently resulted in an increasing variety of innovative specific PET tracers able to explore sympathetic neuronal function or lipid metabolic changes, which might potentially play a role in the early detection of the Doxorubicin-induced cardiotoxic cascade [82, 83]. In this regard, a recent experimental study in a rat model of doxorubicin cardiotoxicity, utilizing PET imaging of noradrenaline signaling for assessment of cardiotoxicity, demonstrated that after 3 weeks of adriamycin administration, there was a decrease in myocardial uptake of the beta-adrenergic antagonist, [3H] CGP12177 [84]. However, it remains unclear whether beta-receptor density, as assessed by PET, can serve as a marker for cardiotoxicity in humans. On the other hand, in a mouse model of anthracycline-induced cardiotoxicity, increased uptake of [(18)F]-CP18, a PET tracer for imaging apoptosis, was observed in the doxorubicin-treated group, thereby suggesting a potential role for detection of anthracycline-induced myocardial apoptosis. However, despite these encouraging results, the complexity of most of the radiolabelling ligands, the high cost and the low availability currently limit the clinical use of molecular hybrid imaging in this type of patients. A major challenge for future studies will be to find a ligand that can easily be radiolabelled, with high selectivity and affinity, high metabolic stability, and low lipophilicity.
32
Figure 2:Images of male beagle at baseline and 3 months after radiotherapy using F-HFD protocol in the study by Yan R and colleagues [80]. A. Dose-distribution axial image. B. Cardiac FDG-PET/CT axial images before RT. C. Cardiac FDG-PET/CT axial images 3 months after RT. D. Myovation images of FDG at 3 months after RT. B shows suppression of myocardial FDG uptake before RT. C and D show high FDG uptake (arrows) corresponding to irradiated field 3 months after RT.
33
2.3.6 Cardiac catheterization and coronary angiography.
It is well-recognized that patients, even in the absence of traditional cardiac risk factors, are at higher
risk of ischemic cardiomyopathy as a result of radiation and chemotherapy, with higher doses increasing the risk for this complication. In addition, CV risk is escalated by the type of chemotherapy agent, particularly with targeted therapies, such as tyrosine kinase inhibitors, which inhibit several cardiomyocytes signaling pathways, including the vascular endothelial growth factor (VEGF) signaling pathway, which seems to play an essential role in maintaining standard endothelial control of vasomotor tone [85]. As outlined by the Society for Cardiovascular Angiography and Interventions (SCAI), [31] right and left heart catheterizations are reasonable for preoperative planning for patients with radiation-induced heart disease and evaluation of pericardial constriction and restrictive cardiomyopathy, when noninvasive imaging is non-diagnostic. Right heart catheterization is indicated in the assessment of HF, assessment of constrictive or restrictive cardiomyopathy, pulmonary hypertension, pericardial disease, and valvular disease. Indications for coronary angiography include symptomatic patients with risk factors and highly suggestive noninvasive testing. Coronary angiography is reasonable for the evaluation of LV systolic dysfunction after radiation therapy. In general, radial access is the preferred vascular access route; however, femoral access is preferred in patients on hemodialysis and those with bilateral mastectomy. Fractional flow reserve is recommended to determine the functional or hemodynamic significance of coronary stenosis before proceeding with nonurgent percutaneous coronary intervention/ revascularization.
34
3. EXPERIMENTAL SECTION: FIRST STUDY 3.1 Materials and Methods
3.1.1 Animal Experiments
Experiments were conducted under the Guide for the Care and Use of Laboratory Animals [86] and approved by the local ethical committee. Animals were kept under the same dietary regimen and divided into three groups to be treated once a week for three weeks with intravenous saline (n=5), doxorubicin 5 mg/Kg (standard-dose, n=5) or doxorubicin 7.5 mg/Kg (high-dose, n=5). Dynamic microPET imaging (Albira, Bruker US) was performed soon before and six days after chemotherapy. Patlak graphical approach [87] was adopted to estimate LV-MRGlu and corresponding index in skeletal muscle (SM-MRGlu) according to the standard procedure previously validated in our lab [12]. SUV were also estimated in LV myocardium (LV-SUV) and limb skeletal muscle (SM-SUV).
3.1.2 Clinical study
Searching the keyword HD in the database of all patients submitted to FDG-PET/CT in our lab between January 2007 and December 2015, we identified 587 patients. The study population was thus selected according to the following inclusion criteria: 1) no cardiovascular disease; 2) no diabetes; 3) normal baseline electrocardiogram and echocardiogram; 4) available staging FDG-PET/CT scan (PET1); 5) negative interim PET (PET2); 6) completion of ABVD chemotherapy scheme (doxorubicin dose: 40-50 mg/m2 per cycle); 7) negative FDG-PET/CT evaluation both 4-6 weeks
post-therapy (PET3) and at six-month follow-up (PET4); 8) no subsequent HD relapse at late clinical follow-up. This process narrowed the final population down to 69 patients.
35 The myocardial metabolic pattern in these patients was compared with the corresponding findings in 69 sex- and age-matched subjects selected from our database [88]. The institutional review board approved this study, and all patients signed a written informed consent related to the imaging procedure, as part of our routine clinical care.
FDG intravenous injection was preceded by a minimum of six hours of fasting and serum glucose level control. All FDG-PET/CT scans were acquired according to the conventional procedure, using a Hirez-16 PET/CT hybrid system (Siemens Medical Solutions). Two volumes of interest were manually drawn on LV myocardium and on SM (Longissimus thoracis) to estimate the LV-SUV and SM-SUV, respectively. CT images were used to identify the myocardium in case of absent cardiac uptake (Figure 3).
All 69 patients were interviewed by telephone; 36 of them accepted to undergo a clinical re-evaluation (14 women, mean age 39±14, age range 21-68). An experienced cardiologist, unaware of PET findings, confirmed the absence of interval development of palpitations, syncope, chest pain or dyspnea and complemented the physical examination with electrocardiogram and echocardiographic evaluation of wall thicknesses, LV diameters, ejection fraction and diastolic function by E/A wave ratio and E wave deceleration time [89].
A significant limitation of this method is that glucose metabolism activity is evaluated by means of a standardized uptake value (SUV)-based approach, which might be influenced by several factors not related to tissue characteristics, including plasma glucose concentration, length of uptake period, partial volume effects and recovery coefficient, as well as PET/CT scanner sensitivity [90]. The error margin due to these
36 factors can actually exceed 50% [90]. This issue has been tackled in clinical nuclear medicine through the introduction of scales able to standardize metabolic qualitative estimation. An emblematic example is the five-point scale criteria (also called Deauville criteria), initially proposed for the evaluation of FDG PET/CT of patients affected by Hodgkin Disease (HD) [91]. This approach is based on the visual comparison of relative metabolic activity in the evaluated tissue with a reference area (mediastinal blood pool and liver) and showed excellent inter-observer reproducibility. Following the success in HD, the five-point scale has been tentatively proposed to other malignancies, including Non-Hodgkin Lymphoma [92] and head and neck cancer [93], as well as to inflammatory diseases such as vasculitis [94, 95]. Based on these considerations we performed a post-hoc analysis on PET1 images aiming: 1) to identify the capability of baseline FDG to identify patients at higher risk of developing DXR induced cardiotoxicity using a five-point scale approach 2) to evaluate whether a score-based image evaluation is superior in predicting myocardial dysfunction with respect to SUV-based approach.
Accordingly, PET1 images were visually evaluated, focusing on the myocardium, following five-point scale criteria [91]. In the case of heterogeneous myocardial FDG uptake, the highest uptake was scored and compared to reference tissues. In patients showing mediastinum or liver disease, only healthy tissue uptake was considered as
37 reference. Examples of corresponding PET images to different score classes are shown in Figure 4.
38
Figure 4: Score-based approach to FDG PET images.
Examples of score-based evaluation of myocardial FDG uptake among the 36 enrolled HD patients. None of PET 1 myocardial images was classified as score 1.
39
3.1.3 Statistical Analysis
All data are presented as mean ± standard deviation or proportions. Differences between paired and unpaired continuous data were analyzed by Student t-test, as appropriate. Categorical variables were analyzed using the chi-square test. A probability value p<0.05 was considered statistically significant. In instances of skewed data distribution, values were transformed using a natural logarithmic transform. Finally, the ability of LV-SUV1 to predict the occurrence of cardiotoxicity, while adjusting for various potential confounders, was tested by multivariate logistic regression analysis. The presence/absence of cardiac abnormalities was tested with respect to the following baseline covariates: age, gender, Ann-Arbor staging, mediastinal irradiation, cumulative administered doxorubicin dose, baseline myocardial FDG uptake. Since clinical assessment for the presence of cardiac abnormalities occurred at various times (from 8 to 26 months) after the administration of ABVD, also follow-up duration was included in the model as a covariate. Due to collinearity between Ann-Arbor staging and mediastinal irradiation, these two variables could not be included in the same model. Their role as confounders were tested separately with all the other covariates included. Since the results of the two analyses were similar, only those of the first one is reported. The multivariate analyses proceeded by means of a backward stepwise procedure, based on the likelihood ratio test, with a p-value for removal ≥0.1. The estimated coefficients with their standard errors were used to compute the Odds Ratios with 95% confidence intervals. The same model was also fitted separately for each covariate to estimate the univariate Odds Ratios with 95% CI.
40 The capability of the score-based image investigation to predict cardiotoxicity was tested by means of univariate analysis. Furthermore, receiver-operating characteristic (ROC) curve analysis was performed to evaluate the capability of both a five-point scale and LV-SUV to discriminate between individuals who developed or not cardiac toxicity and to determine the corresponding cut-off values. As post-hoc analysis, this value was tested to identify the best predictor of cardiotoxicity. Statistical analyses were performed by a biomedical statistician, using a dedicated software application (SPSS, version 21.0; IBM).
3.2 Results
3.2.1 In vivo animal experiments
In animal models, doxorubicin increased cardiac glucose disposal (Fig. 5A) without affecting either body weight or serum glucose level (data not shown). At the compartmental analysis of dynamic PET scans (Fig. 5B), LV-MRGlu remained stable in control mice (from 17.9±4.4 to 18.9±4.8 nMol x min-1 x g-1; p=ns). Standard dose increased LV-MRGlu from 17.5±3.7 to 27.9±9.1 nMol x min-1 x g-1 (p<0.05 vs. controls; p<0.05 vs corresponding baseline). This effect was significantly more evident at the high drug dose that augmented LV-MRGlu from 16.7±5.1 to 37.2±7.8 nMol x min-1 x g-1 (p<0.01 vs. controls and corresponding baseline; p<0.05 vs. standard-dose). As expected, the analysis of LV-SUV strictly reproduced all these findings.
Cardiac selectivity of the doxorubicin metabolic effect was confirmed by the absent response of SM in all groups. In fact, both SM-MRGlu (Fig. 5C) and SM-SUV remained remarkably stable before and after treatment. Consequently, the ratio
LV-41 SUV/SM-SUV remained stable at baseline values in sham mice (3.28±0.9), while it increased in a dose-dependent fashion after chemotherapy (to 4.42±1.02 and to 6.43±1.96 in standard and high doxorubicin dose, p<0.05 and p<0.01, respectively).
Figure 5: Dose-dependent doxorubicin effect on LV-MRGlu.
Panel A displays axial and sagittal planes of microPET studies in mice before and after treatment with saline, doxorubicin 5mg/kg and 7.5 mg/kg, respectively. As displayed in Panel B, doxorubicin administration was followed by a significant increase in LV-MRGlu, as opposed to stable values in untreated mice. Moreover, the dose-dependent nature of doxorubicin metabolic effect was confirmed by the significant difference between post-therapy scans in animals treated with standard or high doses. Doxorubicin administration did not affect SM-MRGlu (Panel C). *=p<0.05 vs after treatment; **=p<0.01 vs after treatment.
42
3.2.2 Overall doxorubicin effect on cardiac metabolism in humans
Clinical data and time intervals between the four PET/CT studies of patient population and controls are reported in Table 3. Baseline LV-SUV was similarly distributed in controls, in the 69 studied subjects and in the 35 excluded patients because of positive PET2 (Fig. 6A). Similarly, no difference in age, gender, and Ann-Arbor staging could be observed between the 69 recruited patients and the 138 excluded ones either because of positive PET2 (n=35) or other reasons (n=103), suggesting a low likelihood of selection bias. In HD population, LV-SUV was 2.37±1.6 at baseline, showed a progressive increase during doxorubicin treatment up to PET3, and remained persistently elevated at PET4 (Fig. 6B). By contrast, overall cardiac uptake remained unchanged in control subjects throughout the study period. Accordingly, HD patients showed significantly higher LV-SUV values with respect to controls both at PET3 and PET4 (Fig. 6B). Again, the selective nature of the cardiac response was confirmed by the divergent behavior of SM metabolism that was only scarcely and transiently affected by doxorubicin (Fig. 6C). Therefore, the ratio between LV- and SM-SUV remained stable during treatment, while it significantly increased during follow-up (Fig. 6D) to values significantly higher than those observed in control subjects. Despite this trend in average values, patients’ response to chemotherapy only partially reproduced the repeatability of doxorubicin action on myocardial metabolism observed in animal experiments. Although no patient showed a progressive reduction in myocardial tracer retention during or after therapy, doxorubicin effect on FDG myocardial uptake was heterogeneous and largely independent from the cumulative drug dose.
43 Table 3: Baseline clinical characteristics of HD and control enrolled patients.
HD Controls p
Age 39 ± 13 (range 19-58) 41 ± 8 (range 20-72) ns
Male Sex 37/69 (53%) 35/69 (50%) ns
Weight (Kg) 67.1 ± 12 76.5 ± 7 <0.05
Gycemia at FDG injection (mg/dl) 79 ± 7 (range 61-101) 83 ± 11 (range 62-94) ns
Cardiovascular Risk Profile
Hypertension 6/69 (8%) 15/69 (21%) <0.01 Tobacco Use 19/69 (27%) 30/69 (43%) <0.05 Total Cholesterol 183.7 ± 30 188 ± 53 ns LDL 114.5 ± 32 120 ± 25 ns Triglycerides 121.3 ± 49 129.7 ± 57 ns Creatinine 0.8 ± 0.1 0.85 ± 0.2 ns
Family hystory of CAD 7/69 (10%) 5/69 (7%) ns
Time intervals between PET studies
PET1 - PET2 (days) 73.7 ± 21 99 ± 90 <0.05
PET2 - PET3 (days) 148 ± 70 167 ± 98 ns
PET3 - PET4 (days) 195 ± 92 229 ± 100 ns
Overall PET1 - PET4 (days) 427±198 448±141 ns
Baseline Ann-Arbor Staging
I Stage 7/69 (10%) - - II Stage 42/69 (60%) - - III Stage 8/69 (12%) - - IV Stage 12/69 (17%) - - B symptoms 10/69 (14%) - - Mediatinic Radiotherapy 35/69 (55%) - -
44
Figure 6Myocardial and skeletal muscle divergent metabolic pattern in HD population.
The distribution of baseline LV-SUV was similar between the 69 enrolled subjects, excluded patients and controls (Panel A). In HD patients LV-SUV progressively increased from PET1 to PET3, remaining relatively stable at PET4 (Panel B). By contrast, it remained unchanged in controls. The divergent nature of SM doxorubicin effect can be appreciable at PET4 (Panel C and D). *=p<0.05 vs control.
45
3.2.3 Myocardial Function Assessment
During the interval between PET4 and cardiological interview (median time from PET4: 30 months; range: 3-96 months), none of the 69 patients reported any hospitalization potentially related to cardiac disorders. In the 36 patients who accepted to undergo a clinical evaluation, the median time elapsed from treatment start to visit was 27 months (range: 8-96 months). Electrocardiogram or echocardiogram documented new-onset abnormalities in 11 patients (31%, 4 females, mean age 44±17, age range 21-66, Table 4). Signs of possible cardiotoxicity were: dyspnea associated with a decrease in LV ejection fraction (n=2); atrial fibrillation (n=1); the appearance of negative T waves in the anterior leads (n=2); alteration in diastolic mitral flow profile (inversion of E/A wave ratio) at Doppler examination (n=6).
Figure 7:LV ejection fraction in patients with normal and abnormal cardiac follow-up. Individual and average values of LVEF in the two subgroups of patients.
46 According to inclusion criteria, pre-therapy LV dimensions and function were normal and were remarkably similar between the two subgroups. By contrast, average LV ejection fraction significantly decreased after doxorubicin in the 11 patients with the cardiotoxic response and became markedly lower with respect to the remaining 25 ones (Table 4, Fig. 7).
47 Table 4: Demographic and clinical data of normal with respect to abnormal follow-up HD patients. Normal FU (n=25) Abnormal FU (n=11) p Age 36.8 ± 12 44.5 ± 17 ns Male Sex 14 (56%) 7 (63%) ns Weight (Kg) 68.9 ± 13 70.7 ± 12 ns Gycaemia at PET1 (mg/dl) 79.2 ± 5 82.4 ± 4 ns
Cardiovascular Risk Profile
Hypertension 2 (8%) 1 (9%) ns Tobacco Use 5 (20%) 2 (18%) ns Total Cholesterol 186 ± 28 181 ± 33 ns LDL 119.6 ± 28 113.3 ± 31 ns Triglycerides 117.2 ± 56 122.9 ± 49 ns Creatinine 0.8 ± 0.1 0.7 ± 0.1 ns
Family hystory of CAD 2 (8%) 1 (9%) ns
Baseline Ann-Arbor Staging
I Stage 3 (12%) 2 (18%) ns II Stage 15 (60%) 5 (45%) ns III Stage 3 (12%) 2 (18%) ns IV Stage 4 (16%) 2 (18%) ns B symptoms 5 (20%) 2 (18%) ns Mediastinic Radiotherapy 13(52%) 5 (45%) ns
Total administered doxorubicin dose (mg) 430.9 ± 109 421.3 ± 107 ns Clinical follow-up duration (days) 1121 ± 874 860 ± 665 ns
Follow up clinical data
Chest pain - - -
Dyspnea - 2 (18%) -
Syncope - - -
Palpitations - 2 (18%) -
Follow-up ECG abnormalities - 3 (27%) * -
Baseline echocardiography
End-diastolic diameter (EDD) (mm) 49.1 ± 2 48 ± 3 ns
End-systolic diameter (ESD) (mm) 29 ± 3 31 ± 2 ns
Fractional shortening (%) 44% ± 4% 46% ± 3% ns
LV ejection fraction 59.8 ± 2.1 59.3 ± 1.7 ns
Distolic dysfunction 0 (0%) 0 (0%) ns
Follow up echocardiography
End-diastolic diameter (EDD) (mm) 49.2 ± 1 49.3 ± 5 ns
End-systolic diameter (ESD) (mm) 27.2 ± 3 33.6 ±11 ns
Fractional shortening (%) 44% ± 4% 38% ± 5% † <0.05
LV ejection fraction 60.3 ± 2 55 ± 7 † <0.05
Distolic dysfunction 0 (0%) 5 (45%) <0.01
* 1 patient developed atrial fibrillation, 2 patients developed negativization of T waves in the anterior leads.
48 The appearance of cardiac abnormality was not related to differences in gender, age, mediastinal radiotherapy, total drug dose, or follow-up duration (Table 4). By contrast, it was associated with markedly lower LV-SUV values at baseline to both the remaining 25 ones (1.53±0.9 vs. 3.34±2.54, respectively p<0.01, Fig. 8A and 8B) and control subjects (3.2±1.7, p<0.01 vs. abnormal and p=ns vs. negative clinical follow-up). This difference tended to disappear over time progressively (Fig. 8C). In fact, FDG uptake significantly and progressively increased in the 11 patients with late cardiac abnormalities (Fig. 8D). This trend persisted even after doxorubicin discontinuation as opposed to the remaining 25 subjects with negative follow-up in whom cardiac metabolic pattern remained relatively stable during and after ABVD (Fig. 8E). Multivariate analysis confirmed that baseline FDG uptake was strongly associated with the subsequent development of cardiac abnormalities, providing an additive predictive power with respect to conventional risk stratification (Table 5).
Table 5: Prediction of cardiac abnormalities: uni- and multivariate analysis Univariate analysis Multivariate analysis
df p O.R. 95% CI p O.R. 95% CI
Sex 1 0.760 0.8 0.19 - 3.35 0.142 *
Age (years) 1 0.140 1.04 0.99 - 1.09 0.339 *
Anna Arbor Stage 1 0.882 0.94 0.41 - 2.12 0.694 * Mediastinal Radiotherapy 1 0.718 1.30 0.31 - 5.39 0.711 * Cumulative doxorubicin dose
(mg) 1 0.963 1.00 0.99 - 1.01 0.486 *
Baseline LV-SUV 1 0.030 0.18 0.03 - 0.85 <0.001 0.065 0.006 - 0.74 Follow up time (days) 1 0.364 1.00 0.99 - 1.001 0.408 * * = excluded from the final model
49
Figure 8: Divergent myocardial metabolic pattern between patients with normal and abnormal cardiac follow-up.
Example cases of abnormal and normal cardiac follow-up HD patients (Panel A and B, respectively). Panel C displays FDG uptake at the four PET/CT scans in the two subgroups, baseline LV-SUV was markedly lower in patients with late cardiac abnormalities with respect to the remaining 25 ones (* = p<0.01, Panel C). FDG uptake significantly and progressively increased in the 11 patients with late cardiac abnormalities (Panel D). This trend persisted even after doxorubicin discontinuation, as opposed to the remaining 25 subjects (Panel E).
