Abstract
Myocardial infiltration by amyloid fibrils causes a severe and progressive form of heart failure. Until recently, this was not treatable. Several novel therapies have recently become available, increasing the urgency to make an accurate diagnosis, evaluate risk, and determine treatment response. Molecular imaging with positron-emitting amyloid tracers has a key emerging role in the evaluation and management of cardiac amyloidosis. In this review, we discuss molecular imaging of cardiac amyloidosis using amyloid PET tracers, including recent advances with a focus on the future.
Myocardial infiltration by amyloid fibrils formed by misfolded proteins leads to cardiac amyloidosis (CA)—one of the most deadly forms of heart failure. In this review, we discuss molecular imaging of CA using amyloid PET tracers, including the current state of the art, recent advances, and future directions. To provide context for the imaging techniques, we begin with a brief overview of CA.
OVERVIEW OF CA
Systemic amyloidoses are a group of disorders characterized by protein misfolding (1). Misfolded proteins can circulate as monomers, dimers, or oligomers that aggregate into protofibrils and amyloid fibrils. Amyloid fibrils are composed of protofilaments (protein layers with a generic cross-β structure) and additional molecules such as glycosaminoglycans and a serum amyloid P-component (2). Amyloid deposits stain with Congo red and manifest increased apple green birefringence under polarized light (2). CA typically exists as a part of systemic amyloidosis and is almost always the cause of death.
There are 2 major types of CA: light-chain (AL) amyloidosis, caused by misfolded immunoglobulin light-chain proteins, and transthyretin (ATTR) amyloidosis, caused by misfolded transthyretin proteins (1,3). AL amyloidosis often affects the kidneys, the heart, and various other organ systems (1). Without therapy, median survival is less than 12 mo. There are 2 main forms of ATTR CA: ATTR wild type (ATTRwt) and ATTR variant (ATTRv). In ATTR CA, the homotetrameric ATTR protein breaks down into monomers and dimers that readily aggregate into fibrils. ATTRwt CA is a disease of aging for which diagnosis is often delayed or missed. It is increasingly recognized as an important cause of heart failure with preserved ejection fraction in older adults. Usually, the cardiac and musculoskeletal symptoms predominate. ATTRv amyloidosis is caused by a variant TTR protein that is often inherited in an autosomal dominant manner and affects younger individuals (4,5). ATTRv symptoms can arise from primary involvement of the heart (cardiomyopathic phenotype), peripheral nerves (neuropathic phenotype), or both (mixed phenotype). Without treatment, ATTR CA has a poor prognosis, with a median survival of 4.8 y (6). In addition to these main types, certain rare forms of amyloidosis are caused by misfolding of other proteins such as leukocyte chemotactic factor 2, gelsolin, lysozyme, apolipoprotein AI, and apolipoprotein AIV (2). Patients with rare forms of amyloidosis may have cardiac, renal, facial, central nervous system, or other symptoms, based on the involved organs; therefore, there is a need to image amyloid in all potentially affected organs.
EVALUATION OF CA
The evaluation of CA typically incorporates serum biomarkers, serum free light-chain studies, and cardiac imaging and may include biopsy (7,8). Biomarkers (N-terminal pro–brain natriuretic peptide, troponin T, estimated glomerular filtration rate, serum transthyretin levels, serum free light-chain levels, and serum and urine immunofixation electrophoresis) play a crucial role in diagnosis and risk stratification, with an emerging role for assessment of treatment response (8). However, none of these biomarkers are specific for cardiac involvement in AL CA or ATTR CA. Given the lack of specific serum biomarkers, development of amyloid-specific cardiac imaging metrics is crucial. Echocardiography and cardiac MRI can provide characteristic imaging features, including increased myocardial thickness and mass, a bright and sparkly appearance of the myocardium, diastolic dysfunction, and reduced longitudinal contraction quantified by global longitudinal strain (7). Although these imaging markers raise the suspicion of CA, they are also not specific for CA.
CURRENT KNOWLEDGE AND PRACTICE GAPS
For heart failure patients in whom ATTR CA is suspected because of typical cardiac imaging features and normal serum-free light-chain studies, a definitive diagnosis can be based on bone-avid tracer (99mTc-labeled pyrophosphate [PYP], 3-diphosphono-1,2-propanodicarboxylic acid [DPD], or hydroxymethylene diphosphonate [HDP]) cardiac SPECT (9), without the need for biopsy. Although bone-avid SPECT tracers have high specificity (100% for grade 3 or 3 myocardial utpake in the absence of a monoclonal process) and high sensitivity (99%) in specific clinical contexts (grade 1, 2, or 3 myocardial uptake without exclusion of monoclonal process), their sensitivity (70%) for nonbiopsy diagnosis of ATTR CA is relatively low (9). They are also unsuitable for diagnosing AL and rare forms of CA. Semiquantitative metrics for 99mTc-PYP/DPD/HDP using SPECT/CT are emerging (10) but are not as well developed as those for PET/CT. Also, the mechanism underlying myocardial uptake of 99mTc-PYP/DPD/HDP has not been established. Microcalcifications have been proposed as a potential mechanism of bone-avid tracer uptake, but this proposal is not strongly supported by the literature. Thelander et al. (11) used 4 cardiac sections (3 with ATTR and one with AL) and found cardiac microcalcifications in ATTR hearts at regions distinct from amyloid deposits, but autoradiography images with 99mTc-DPD were limited. Furthermore, 18F-NaF PET, a tracer of microcalcification, has proven to be of limited value for detection of ATTR CA or AL CA (12). Amyloid PET tracers bind to amyloid deposits independently of precursor protein and are quantitative. Therefore, there is substantial interest in amyloid-specific PET tracers for diagnosing AL CA, ATTR CA, and rare forms of amyloidosis and for quantifying amyloid burden to identify early CA and response to therapy.
THERAPIES FOR CA
In recent years, there has been significant progress in specific therapies targeting the precursor protein. For AL CA, the current standard of care is plasma cell–directed chemotherapy with daratumumab in combination with cyclophosphamide, bortezomib, and dexamethasone (13). This regimen is highly effective and rapidly reduces light-chain levels. For ATTR CA, TTR stabilizing (tafamidis, AG10, diflunisal) or silencing (patisiran, vutrisiran, inotersen, eplontersen) therapies have emerged as the primary options (14). Tafamidis, the only approved therapy for ATTR CA, is less effective in patients with advanced heart failure (New York Heart Association heart failure class 3), underscoring the importance of early identification of ATTR CA. Despite clinical improvement, heart failure often persists after successful precursor protein therapy, because existing therapies do not directly target the amyloid fibrils, which are responsible for organ dysfunction. Consequently, reduction of amyloid deposits represents the next frontier of therapeutic advancements in amyloidosis.
AMYLOID PET RADIOTRACERS
Several amyloid-binding radiotracers, originally developed for imaging β-amyloid in the brain, have been successfully repurposed to image systemic and cardiac amyloid deposits (Table 1). 11C-Pittsburgh compound B (11C-PiB) and 18F-flutemetamol are structurally similar to thioflavin T (a dye used for histologic staining of amyloid), whereas 18F-florbetapir and 18F-florbetaben are stilbene derivatives (Fig. 1A). Thioflavin T binds to various fibrils independently of the precursor protein, suggesting that it recognizes a structural feature common among fibrils—likely the β-sheet surface of the amyloid fibril along channels formed by cross-strand ladders (15). These tracers appear to similarly bind to the amyloid fibril, and 18F-flutemetamol (Fig. 1B) has been shown to bind to 3 different amyloid binding sites (BS1, BS2, and BS3) (16). A novel radiotracer, 124I-evuzamitide (p5 + 14), has recently been developed to directly image systemic amyloid deposits, binding to glycosaminoglycans (Figs. 1C and 1D), which are abundant on amyloid fibrils (17,18). Postmortem studies have demonstrated specific binding of 18F-florbetapir (Fig. 2) (19) and cyanoflutemetamol (20) to myocardial AL and ATTR deposits. Different β-amyloid binding tracers may have differing affinities to these binding sites (21,22).
Characteristics of Available Amyloid PET Radiotracers
Molecular structures of various amyloid radiotracers and potential binding sites. (A and B) β-amyloid PET radiotracers (A) bind to β-sheets on amyloid fibrils (B, 19F nuclear MR [NMR] study of 18F-flutemetamol and β-amyloid). (C and D) Synthetic polypeptide 124I-evuzamitide (C) has 12 charged lysine side chains (blue), which bind to negatively charged side chains exposed on long axis of amyloid fibrils of hypersulfated heparan sulfate glycosaminoglycans (D). AL fibril model was developed from cryogenic electron microscopy data, and evuzamitide model was predicted using iterative threading assembly refinement. DeepView Swiss-PdbViewer (version 4.1) was used to make images. (A and B reprinted with permission of (53) and (22), respectively. C and D courtesy of Dr. Jonathan Wall, University of Tennessee.)
Autopsy-derived myocardial sections showing specific binding of 18F-florbetapir with myocardial AL and ATTR deposits. These autoradiography results confirm specific binding of 18F-florbetapir to ATTR (top row) and AL (middle row) deposits on myocardial sections and no binding to control myocardium (bottom row). Minimal to no nonspecific binding was seen (right column). (Reprinted with permission of (19).)
IMAGING PROTOCOLS FOR AMYLOID PET SCANS
Imaging protocols for CA PET tracers have not been standardized. Usually, static cardiac and partial-body scans are acquired. A 10- or 20-min static cardiac scan starting 10 min after injection of 185–370 MBq of radiotracer may be adequate for diagnosis; 10–20 min are optimal for 11C-PiB (23). Dynamic cardiac scans are acquired for research purposes. For 124I-evuzamitide, static imaging is performed 5 h after intravenous administration of 37 MBq. We perform a 30-min cardiac scan because of the low injected activity and low positron abundance (23%). The 5-h delay allows clearance of tracer through the kidneys. Thyroid blockade with potassium begins 1 d before the scan and continues for 6 d.
INTERPRETATION OF AMYLOID PET SCANS USING VISUAL AND SEMIQUANTITATIVE METRICS
PET scans for systemic and cardiac amyloidosis are interpreted through visual assessment. Semiquantitative imaging metrics of SUV and myocardium–to–blood-pool activity concentration ratios on static images are also commonly used clinically. SUV is defined as the mean or maximal tracer uptake in the region or volume of interest/(injected activity/patient weight)—SUVmean and SUVmax, respectively. Percentage injected dose is another semiquantitative measure derived from static images; it is a weight-independent measure, defined as the product of mean activity concentration in the region or volume of interest and its volume normalized to injected activity. Cardiac amyloid activity is defined as the product of SUVmean and the volume of myocardial pixels above a threshold activity concentration. Retention index (RI; the ratio of mean myocardial activity to the integral of blood-pool activity until the midpoint of the scan) is derived from dynamic image series.
SUV-based metrics are widely used in the oncology literature to differentiate benign from malignant lesions and to follow the response to therapy. Several technical factors (scanner, acquisition, reconstruction parameters), biologic factors, the imaging protocol (e.g., time between tracer administration and imaging), and the size of the region or volume of interest may affect SUVs; hence strict standardization is mandatory (24). SUVmax represents the value from a single voxel and can be contaminated by spillover from overlapping activity from bone or other organs. SUVmean represents the mean value of all voxels in the volume of interest, is dependent on the volume-of-interest definition and may be insensitive to identify early CA, which may start focally (25). As opposed to SUV metrics, cardiac amyloid activity and percentage injected dose metrics are volumetric metrics that take into account the volume of amyloid in the myocardium (25). Also, SUVmax and SUVmean (and thus cardiac amyloid activity) are metrics normalized to body weight, which may not be optimal for highly specific tracers such as amyloid tracers, and percentage injected dose (weight-independent) may be superior to SUV-based metrics with amyloid tracers (25). Beyond these semiquantitative metrics, quantitative metrics are under evaluation for assessment of amyloid burden, as described in the section on absolute quantitation of PET cardiac amyloid tracers.
DETECTION OF CA
The first report of PET imaging in CA used 11C-PiB (26) and found significantly higher myocardial retention in CA patients than in controls (mean RI of 0.054 min−1 [range, 0.033–0.134 min−1] in amyloidosis patients vs. 0.025 min−1 [range, 0.020–0.031 min−1] in healthy controls) (Fig. 3A). Similar results were confirmed with 18F-florbetapir (median RI of 0.023 [interquartile range, 0.015–0.024] in controls vs. 0.043 [interquartile range, 0.034–0.051] in amyloidosis patients, P = 0.002) (Fig. 3B) (27). In this study, a target-to-background ratio (left ventricular activity to blood-pool activity on 10- to 30-min static images) of more than 1.45 perfectly separated CA patients from controls.
Myocardial retention of 18F-florbetapir in CA and control cohorts. RI of 11C-PiB (A) and 18F-florbetapir (B) in myocardium was significantly higher in amyloidosis cohort than in control cohort, with no overlap in 95% CIs. p.i. = after injection. (A and B reprinted with permission of (26) and (27), respectively.)
Because ATTR CA is usually diagnosed at a later stage than AL CA, ATTR CA patients typically have a greater amyloid burden. If myocardial tracer uptake is proportional to amyloid burden, then greater myocardial tracer uptake would be expected in ATTR CA than in AL CA. But, notably, in the 18F-florbetapir PET study (27), myocardial uptake was highest in the AL cohort, with both the AL cohort and the ATTR cohort having significantly higher uptake than controls (AL CA > ATTR CA > controls) (Fig. 4A). Structural differences in AL and ATTR peptides, as well as differences in the binding strength of the β-amyloid radioligands to AL and ATTR peptides, may explain the lower signal in ATTR CA. Indeed, a recent 19F nuclear MR study (22) found that flutemetamol binds to the amyloid β-40 fibril polymorph with a stoichiometry of 1 ligand per 4–5 peptides and that half of the ligands are tightly bound whereas the other half are loosely bound. But this needs to be further studied. Consistent results with lower myocardial uptake in ATTR CA were reported for 11C-PIB (RI of 0.086 [interquartile range, 0.075] in AL CA and 0.045 [interquartile range, 0.014] in ATTR CA) (Fig. 4B) (23) and for 18F-florbetaben (Figs. 4C and 4B) (RI of 0.043 [range, 0.032–0.065] in AL CA and 0.035 [range, 0.022–0.042] in ATTR CA) (28,29). There are insufficient data to compare 18F-flutemetamol in AL and ATTR CA. In contrast, in a recent study using 124I-evuzamitide PET/CT in patients with AL and ATTR CA (30), we found that 124I-evuzamitide myocardial uptake (SUVmax, SUVmean, percentage injected dose, and cardiac amyloid activity) was similar in ATTR CA and AL CA. Notably, in the same cohort of patients with ATTR CA, myocardial uptake of 124I-evuzamitide was greater than that of 18F-florbetapir (Fig. 5) (30). These findings suggest that quantitative 124I-evuzamitide PET/CT may hold promise for detection of early ATTR CA.
Myocardial uptake of amyloid PET radiotracers 18F-florbetapir (A), 11C-PiB (B), and 18F-florbetaben (C and D) in AL CA, ATTR CA, and control cohorts. Small cohort studies from multiple centers showed that amyloid tracer retention in heart is highest in AL CA cohort, intermediate in ATTR CA cohort, and lowest in control cohorts. These findings suggest lower binding affinity of β-amyloid tracers to ATTR deposits than to AL deposits. (Reprinted with permission of (23,27,29,54).
Comparison of myocardial uptake of 124I-evuzamitide and 18F-florbetapir in same cohort of patients with AL CA and ATTR CA. All other measures of amyloid, left ventricular (LV) percentage injected dose (%ID), SUVmean, and cardiac amyloid activity (CAA) were higher in ATTR CA with 124I-evuzamitide, except for LV target-to-background ratio (TBR), which was higher with 18F-florbetapir. CMP = cardiomyopathy. (Reprinted with permission of (30).)
Of all the 18F-amyloid tracers, 18F-flutemetamol is the least studied for the evaluation of CA. A histologic study showed cyanoflutemetamol labeling of AL, ATTRwt, and ATTRv on myocardial sections, suggesting potential utility for imaging CA (20). One study (31) evaluated 21 patients with the V30M ATTR variant, including 18 patients with prior liver transplantation, using a dynamic 18F-flutemetamol PET/CT protocol. Reported values for specificity and sensitivity were 100% and 88%, respectively, using a septal SUV threshold of 1.46; similar results were reported for RI and for lateral wall metrics. Another study evaluated 12 patients with CA and 5 patients with nonamyloid heart failure using 18F-flutemetamol. This study showed limited sensitivity, with visually apparent uptake in only 2 of 12 amyloidosis patients (32). In most of these cases, data were acquired from 60 min after radiotracer injection, much later than in most CA studies. The optimal protocol and the diagnostic value of 18F-flutemetamol for CA warrant further study.
Amyloid PET imaging may have incremental value in certain cases of hereditary ATTR CA. Studies have shown that 99mTc-DPD uptake was negative in some ATTR CA patients, particularly those with certain TTR variants (33); it has been proposed that this may relate to the fibril type. The amyloid fibrils can exist as fragmented fibrils (type A, containing C-terminal fragments, and full-length fibrils) or as only full-length fibrils (type B) (34). ATTRwt almost always consists of type A fibrils. Patients with the variant forms of ATTR amyloidosis may contain only type B fibrils, which are associated with early onset of disease. Type A fibrils are congophilic and strongly birefringent, whereas type B fibrils are weakly congophilic (35). In one study (36), 10 patients with the V30M variant underwent 99mTc-DPD SPECT as well as 11C-PiB PET/CT. Those with type B fibrils (n = 5) had negative 99mTc-DPD findings. However, 11C-PiB myocardial uptake was significantly higher in all patients than in healthy volunteers (P < 0.0001), regardless of fibril type (type A or type B). Inconsistent results were reported in a more recent study, which found negative 11C-PiB results in ATTRv (late-onset V30M and non-V30M) and in ATTRwt (37). More experience is needed.
PET CA imaging differs from brain imaging in that nonspecific uptake is minimal in the heart, simplifying visual interpretation of these images. Furthermore, several of these tracers have shown evidence of pathologic uptake in organ systems (Fig. 6) (38–41) compared with physiologic uptake in healthy controls, particularly in AL CA. Moreover, these tracers can detect amyloid deposits from various precursor proteins (Fig. 6). Thus, a disadvantage of amyloid PET imaging is that although it can detect the presence of amyloid deposits in the myocardium, it cannot differentiate among the various types of amyloid. Identifying type of amyloidosis requires additional testing, such as blood assays, tissue biopsy, or genetic analysis.
Myocardial (top 2 rows) and systemic organ (bottom row) uptake of 18F-florbetapir and 124I-evuzamitide in patients with various forms of amyloidosis. Liver uptake is physiologic on 18F-florbetapir imaging. Amyloid PET tracers bind to various types of amyloid and can image amyloid deposits in heart as well as in certain systemic organs. Apo-A-IV = apolipoprotein AIV.
The evidence for the diagnostic accuracy of amyloid PET tracers has come mainly from small, single-center studies of select groups of patients with a definite diagnosis of CA and controls who are mostly healthy (except for certain studies in which left ventricular hypertrophy (29) and hypertrophic cardiomyopathy (23) subjects were included). This subject selection has likely led to overestimation of the sensitivity, specificity, and overall diagnostic accuracy of these tracers. In practice, patients with amyloid phenocopies and suspected amyloidosis need to be evaluated; the diagnostic accuracy of these tracers in that cohort remains to be determined. The diagnostic value of 18F-florbetaben is being evaluated in a phase 3 clinical trial. This study (NCT05184088) is evaluating patients with either suspected CA, a putative diagnosis of CA with some diagnostic uncertainty (e.g., unclear etiology), or a definitive diagnosis of amyloidosis but unclear cardiac involvement. The primary outcome measure is diagnostic accuracy for AL amyloidosis. In an exploratory analysis, the sensitivity and specificity of 18F-florbetaben PET for differential diagnosis among AL CA, ATTR CA, and non-CA will be assessed. This study is important, as currently we have no Food and Drug Administration–approved radiotracers for AL CA. The availability of a Food and Drug Administration–approved radiotracer for imaging AL CA might transform clinical practice.
Overall, 18F-amyloid PET tracers and 11C-PiB have consistently demonstrated excellent diagnostic accuracy for detecting AL CA, albeit in selected cohorts. On the basis of limited data, it appears that myocardial uptake of β-amyloid PET tracers is lower in ATTR CA than in AL CA. By contrast, pilot data suggest that 124I-evuzamitide has similar uptake in AL and ATTR CA. Currently available data suggest that 18F-florbetapir, 124I-evuzamitide, and other amyloid PET tracers are capable of imaging several types of amyloid in the heart and other organs. Further research and larger-scale studies are needed to establish the optimal diagnostic role of various amyloid PET tracers.
DETECTION OF EARLY CA
Early detection of CA is crucial for improving clinical outcomes. The existing literature has focused on CA patients with increased wall thickness, representing an advanced phenotype of the disease. The ability of amyloid PET tracers to identify early CA, before overt changes in cardiac structure, remains poorly understood. β-amyloid tracers have high sensitivity for AL CA and may be useful for detection of early AL CA. But the apparently lower myocardial uptake of β-amyloid PET tracers in ATTR CA may limit its application for detection of early ATTR CA. Two studies using 18F-florbetapir and 11C-PiB tracers investigated early AL CA and ATTR CA in patients with normal left ventricular wall thickness. In the 18F-florbetapir study (42), 40 patients with biopsy-proven systemic AL amyloidosis underwent PET/CT imaging. Among a subset of patients (n = 10) with a normal left ventricular wall thickness and normal cardiac biomarkers (N-terminal pro–brain natriuretic peptide and troponin T), approximately 50% had abnormal 18F-florbetapir RI values, similar to those observed in patients with AL CA (Fig. 7). These patients also had a slightly elevated extracellular volume on cardiac MRI. Together, these findings suggest that 18F-florbetapir PET/CT can identify an early phenotype of AL CA. Similarly, the 11C-PiB study (23) included 51 participants (15 AL, 21 ATTR, 15 controls), of whom 11 had evidence of systemic amyloidosis but normal wall thickness. Five of these 11 patients were positive by visual assessment on 11C-PiB scans, including 2 of 6 ATTRv carriers and 2 of 5 AL amyloidosis patients. These patients also had higher RI values. The results of these 2 studies suggest that amyloid PET imaging can reveal the presence of early AL CA and ATTR CA.
Diagnosis of preclinical AL CA using 18F-florbetapir PET/CT. Shown are 18F-florbetapir myocardial RI in AL CA (left dot plot, AL cardiomyopathy, untreated), AL CA with successful AL amyloidosis therapy and hematologic remission for >1 y (middle dot plot, AL remission cardiomyopathy), and systemic AL amyloidosis patients with normal left ventricular wall thickness and normal cardiac biomarkers (right dot plot, AL before cardiomyopathy). Dotted lines indicate normal limits derived from healthy controls. About half of patients in AL-before-cardiomyopathy cohort showed abnormally elevated myocardial 18F-florbetapir RI values, which were similar to those of patients with clinically documented AL cardiomyopathy. Panel on right shows images from 2 patients with AL before cardiomyopathy, one with significant myocardial uptake of 18F-florbetapir (PET-positive) and another with no uptake (PET-negative). These findings suggest that 18F-florbetapir can identify early preclinical AL amyloid deposits in myocardium. (Reprinted with permission of (42).)
RISK ASSESSMENT WITH AMYLOID PET TRACERS
The available literature on the use of CA PET imaging for risk stratification is limited. In one study of 11C-PiB imaging of 41 patients with AL amyloidosis, over a median follow-up period of 423 d (interquartile range, 93–1,222 d), 24 patients experienced a cardiac event, that is, death, cardiac transplantation, or acute decompensated heart failure (43). Patients in the highest tertile of 11C-PiB uptake (median SUV ratio for myocardium to aortic blood, 8.67; interquartile range, 7.69–12.3) had a significantly worse prognosis (log-rank P = 0.014). Both the E/e′ ratio and the SUV ratio were identified as independent predictors of events (43). Similarly, another study using 11C-PiB PET reported that a higher myocardial uptake of this radiotracer is associated with poorer outcomes (Fig. 8) (44). Sample sizes were small in both studies, precluding risk-adjusted survival analyses. Currently, no prognostic data on amyloid tracer PET imaging for ATTR amyloidosis and rare forms of amyloidosis have been reported.
Kaplan–Meier survival curves showing that survival in AL CA patients is worse with high, than with low, myocardial 11C-PiB uptake. (Reprinted with permission of (44).)
ASSESSMENT OF TREATMENT RESPONSE
Limited data exist on treatment response in CA. Echocardiography and cardiac MRI are the primary assessment tools (45–47), and most studies have focused on AL CA. Currently available precursor protein–directed therapies are intended to enhance natural amyloid resorption, improve cardiac structure and function, and, potentially, reduce amyloid burden. Amyloid-depleting therapies, at present investigational, are expected to decrease amyloid burden and further improve cardiac structure and function. A recent phase I study assessed the efficacy of an antibody-based therapy for removing ATTR fibrils and reported a reduction in myocardial uptake of bone-avid 99mTc tracer as early as 4 mo after treatment initiation, as well as with an improvement in extracellular volume (48). These preliminary findings highlight the crucial role of cardiac imaging in assessing treatment response. With the emergence of various therapeutic approaches for amyloidosis, molecular imaging has the potential to identify nonresponders who may benefit from alternative treatment strategies. Serial CA PET imaging may be a successful approach to monitoring amyloid burden and guiding management. Visual identification of changes on serial amyloid scans is extremely challenging; therefore, quantitative amyloid PET imaging is essential for accurate evaluation of treatment response.
ABSOLUTE QUANTITATION OF PET CARDIAC AMYLOID TRACERS
Absolute quantitation of PET imaging data is accomplished through mathematic modeling of tissue and blood time–activity curves to yield kinetic parameters reflecting physiologic processes. The models assume plasma delivery of tracer to one or more (functional) tissue compartments. For cardiac amyloid tracer imaging, the first tissue compartment is the cardiomyocytes; the second compartment, if present, represents amyloid fibrils in the interstitial space, to which the tracer binds. The kinetic parameters of interest include rate constants, which reflect movement of the tracer from the plasma to the myocardial cells, from the cells back to the plasma, from the myocardial cells to the amyloid fibrils, and (in some cases) from the amyloid fibrils back to the myocardial cells, as well as derived parameters based on multiple rate constants. Because the myocardium is perfused with blood, the fraction of blood within myocardial tissue is an additional fitted parameter.
Kinetic parameters have the potential to be sensitive markers of disease presence, amyloid burden, disease progression, or therapeutic response. However, this approach is not widely used in CA, because it is complex and technically demanding. Cardiac tissue time–activity curves are derived from one or more volumes of interest representing all or part of the left or right ventricular myocardium. The blood time–activity curve (input function) is either obtained directly by arterial sampling, followed by processing and well-counting of samples, or derived from blood-pool activity in the images. The arterial blood can be analyzed for plasma (rather than whole-blood) activity concentration and corrected for radioactive metabolites, which are present for many tracers, thereby yielding the metabolite-corrected plasma time–activity curve. However, this approach is invasive and requires processing of blood; furthermore, the blood time–activity curve changes as the tracer bolus travels through the circulation. Image-derived blood-pool time–activity curves have the advantage of proximity to the heart, but they represent whole blood rather than plasma and, for most tracers, contain radioactive metabolites. These curves should be corrected using time-dependent plasma–to–whole-blood activity ratios and metabolite fractions; these functions can be determined individually from venous blood samples or inferred using data from previous cohorts.
Fitting of the time–activity curves to the compartmental models is complex. The appropriate model must be determined (how many compartments, reversible or irreversible binding, etc.). Furthermore, because PET data are noisy, robust estimation of rate constants (other than the flow-related rate constant) is not always possible; indeed, in some situations, determination of individual rate constants cannot be accomplished even with noise-free data. For this reason, derived parameters, representing combinations of rate constants, are often used.
To date, 2 groups have applied compartmental analysis to CA PET imaging. Kero et al. (49) published a small study in which they investigated compartmental analysis of 11C-PIB PET data. Their methodology was rigorous; they compared a 1-tissue-compartment model with reversible and irreversible 2-compartment models. They also tested 2 variants of the reversible model in which certain parameters were fixed. The input function was obtained by arterial sampling with metabolite analysis, and the model incorporated corrections for spillover from both ventricular blood pools. They found that the 2-compartment irreversible model yielded the best fits to their data from 7 AL and ATTR CA patients based on the Akaike criterion and ability to generate robust parameter estimates. They also calculated RI and SUV metrics and reported very strong correlations between a derived kinetic parameter (net influx rate, Ki) and the simpler metrics. Classification performance was measured in a retrospective cohort of 10 amyloidosis patients and 5 healthy controls. They reported significant between-group differences in SUV, RI, and Ki, with some overlap in Ki values but perfect discrimination for the simpler metrics. Santarelli et al. (50) applied a 2-tissue-compartment irreversible model to 18F-florbetaben cardiac PET images in a 36-subject (11 AL CA, 10 ATTR CA, and 15 controls with other heart conditions) cohort, finding that although no single parameter classified all 3 groups, they were able to classify the 3 groups using multiple parameters. Interestingly, they found that the rate constant representing binding to amyloid was zero for controls and very low for ATTR, implying that a 1-tissue-compartment model would be preferable for these groups. Metabolite correction was not performed.
These 2 studies demonstrate that, although complex and technically difficult, compartmental analysis of cardiac PET data is feasible with 11C-PIB and 18F-florbetaben, and kinetic parameters are useful for separating CA from controls and AL from ATTR. It remains to be determined how their sensitivity compares with simpler metrics that can be used clinically for early detection of CA and for monitoring changes due to disease progression or therapy. Kinetic modeling methods remain challenging, and development of population metabolite curves may be necessary for accurate quantification in clinical practice.
FUTURE
The future of amyloid PET imaging in novel investigations for CA is open. Multicenter studies of unselected cohorts of patients with suspected CA are needed to establish the true performance of this imaging modality for definitive diagnosis in patients with suspected amyloidosis. The detection of preclinical amyloid deposits remains a significant challenge with current technologies; amyloid PET tracers may play a significant role. In addition, comprehensive investigations of uptake in healthy subjects are necessary to define physiologic normal limits of tracer uptake with which patient data can be compared. Whole-body amyloid PET imaging would make possible assessment of amyloid deposition in various organs, providing a comprehensive staging of disease burden, analogous to oncology metabolic imaging studies. The capability of serial imaging at multiple time points after initiation of therapy to identify nonresponders at an early stage has not yet been assessed. These studies must incorporate measurement of the repeatability and reproducibility of metrics derived from amyloid PET images. The establishment of large registries of images of multiple tracers acquired under standardized imaging protocols, similar to the IDEAS study (51) or the ADNI study (52) in Alzheimer disease, would be instrumental in acquiring more knowledge about CA and guiding management to improve clinical outcomes in this rare and underdiagnosed disease. Of the many exciting potential future applications, three are most promising for advancing clinical care: sensitive and specific detection of amyloidosis as it may minimize need for biopsy, detection of small amounts of amyloid, as it makes possible early therapy before overt cardiac dysfunction, and quantification of amyloid burden, as it may be critical to develop amyloid fibril–depleting therapies.
CONCLUSION
CA PET provides valuable diagnostic insights for patients with CA. These tracers enable the detection of various types of amyloid both in the heart and in the entire body. Data on the efficacy of amyloid PET imaging for diagnosis of ATTR CA, assessment of risk, and monitoring of response to therapy with amyloid PET tracers are very limited. Amyloid PET imaging is expected to play a crucial role in early diagnosis and in monitoring of therapy, particularly for amyloid-depleting therapies. More research is needed to optimize the clinical utility of amyloid PET imaging.
DISCLOSURE
This work was supported by NIH grant K24HL157648 to Sharmila Dorbala. Sharmila Dorbala also receives consulting fees from Pfizer, GE Healthcare, Novo Nordisk, and Astra Zeneca and has received investigator-initiated grants from Pfizer, Attralus, GE Healthcare, Phillips, and Siemens. No other potential conflict of interest relevant to this article was reported.
Footnotes
Published online Nov. 1, 2023.
- © 2023 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication June 6, 2023.
- Revision received September 28, 2023.