Author + information
- Received March 12, 2018
- Revision received May 15, 2018
- Accepted June 26, 2018
- Published online August 15, 2018.
- Paco E. Bravo, MDa,e,
- Kana Fujikura, MD, PhDa,
- Marie Foley Kijewski, PhDb,
- Michael Jerosch-Herold, PhDa,
- Sophia Jacob, BSb,
- Mohamed Samir El-Sady, BSb,
- William Sticka, CNMTb,
- Shipra Dubey, PhDb,
- Anthony Belanger, PhDb,
- Mi-Ae Park, PhDb,
- Marcelo F. Di Carli, MDa,b,c,
- Raymond Y. Kwong, MD, MPHa,c,
- Rodney H. Falk, MDc,d and
- Sharmila Dorbala, MD, MPHa,b,c,d,∗ ()
- aNoninvasive Cardiovascular Imaging Program, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts
- bDivision of Nuclear Medicine, Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
- cCardiovascular Division, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts
- dAmyloidosis Program, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
- eDivisions of Nuclear Medicine and Cardiology, Departments of Radiology and Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
- ↵∗Address for correspondence:
Dr. Sharmila Dorbala, Cardiac Amyloidosis Program, Cardiovascular Division, Department of Medicine, Heart & Vascular Center, Division of Nuclear Medicine, Department of Radiology, Brigham and Women’s Hospital and Harvard Medical School, 70 Francis Street, Shapiro 5th Floor, Room 128, Boston, Massachusetts 02115.
Objectives This study sought to test whether relative apical sparing (RELAPS) of left ventricular (LV) longitudinal strain (LS) in cardiac amyloidosis (CA) is explained by regional differences in markers of amyloid burden (18F-florbetapir uptake by positron emission tomography [PET] and/or extracellular volume fraction [ECV] by cardiac magnetic resonance (CMR)].
Background Further knowledge of the pathophysiological basis for RELAPS can help understand the adverse outcomes associated with apical LS impairment.
Methods This was a prospective study of 32 subjects (age 62 ± 7 years; 50% males) with light chain CA. All subjects underwent two-dimensional echocardiography for LS estimation and 18F-florbetapir PET for quantification of LV florbetapir retention index (RI). A subset also underwent CMR (n = 22) for ECV quantification. Extracellular LV mass (LV mass*ECV) and total florbetapir binding (extracellular LV mass*florbetapir RI) were also calculated. All parameters were measured globally and regionally (base, mid, and apex).
Results There was a significant base-to-apex gradient in LS (−7.4 ± 3.2% vs. −8.6 ± 4.0% vs. −20.8 ± 6.6%; p < 0.0001), maximal LV wall thickness (15.7 ± 1.9 cm vs. 15.4 ± 2.9 cm vs. 10.1 ± 2.4 cm; p < 0.0001), and LV mass (74.8 ± 21.2 g vs. 60.8 ± 17.3 g vs. 23.4 ± 6.2 g; p < 0.0001). In contrast, florbetapir RI (0.089 ± 0.03 μmol/min/g vs. 0.097 ± 0.03 μmol/min/g vs. 0.085 ± 0.03 μmol/min/g; p = 0.45) and ECV (0.53 ± 0.08 vs. 0.49 ± 0.08 vs. 0.49 ± 0.07; p = 0.15) showed no significant base-to-apex gradient in the tissue concentration or proportion of amyloid infiltration, whereas markers of total amyloid load, such as total florbetapir binding (3.4 ± 1.7 μmol/min vs. 2.8 ± 1.5 μmol/min vs. 0.93 ± 0.49 μmol/min; p < 0.0001) and extracellular LV mass (40.0 ± 15.6 g vs. 30.2 ± 10.9 g vs. 11.6 ± 3.9 g; p < 0.0001), did show a marked base-to-apex gradient.
Conclusion Segmental differences in the distribution of the total amyloid mass, rather than the proportion of amyloid deposits, appear to explain the marked regional differences in LS in CA. Although these 2 matrices are clearly related concepts, they should not be used interchangeably.
Left ventricular (LV) longitudinal strain (LS) by speckle-tracking echocardiography is an independent predictor of mortality in amyloidosis (1). Although global LS is universally impaired in subjects with clinically manifest cardiac amyloidosis (CA), the apical segments of the LV are typically less affected or even spared in some cases, a finding consistently observed in both light chain (AL) and transthyretin (ATTR) amyloidosis (1,2). Importantly, emerging data also suggest that apical LS is probably a better predictor of events than LS at the basal or mid LV segments (2).
The basis for the relative apical sparing of LS (RELAPS) in amyloidosis remains incompletely understood. Prior data suggest that it may be related to lower amyloid burden at the LV apical segments. This is suggested by the preferential thickening of the basal and mid LV segments (3) relative to the apical LV segments, coupled with an increase in either late gadolinium enhancement (LGE) on cardiac magnetic resonance (CMR) (2) or regional uptake of technetium-99m pyrophosphate (PYP) on single photon emission computed tomography (SPECT) (4). However, both LGE-CMR and PYP-SPECT are limited techniques for regional quantitative analyses because of the following: 1) they are not fully quantitative; 2) LGE may not be detectable in cases of early or diffuse myocardial infiltration (5); and 3) measurement of relative apical uptake on SPECT is subject to significant partial volume error. Furthermore, PYP is insensitive to AL.
In contrast, contrast-enhanced (CE)-CMR imaging with T1 mapping and molecular targeted imaging with 18fluorine (18F)florbetapir positron emission tomography (PET) are advanced noninvasive techniques that can, potentially, provide more accurate estimation of amyloid burden in the heart compared with echocardiography, standard LGE, or SPECT. Native post-contrast T1 mapping permits measurement of global and regional LV extracellular volume fraction (ECV), which is significantly expanded in amyloid hearts as a result of amyloid infiltration (6,7). Florbetapir is a stilbene derivative that specifically binds to AL and ATTR deposits in human hearts ex vivo and in vivo (8,9), and molecular amyloid PET imaging provides a unique opportunity to specifically quantify global and regional amyloid burden in the heart. However, whether the RELAPS phenomenon is explained by regional differences in amyloid burden has not been investigated.
The aims of the present study were as follows: 1) to measure the regional distributions of myocardial 18F-florbetapir uptake and ECV expansion in subjects with AL cardiac amyloidosis and 2) to determine whether regional differences of these non-invasive markers of amyloid burden could help explain the relative sparing of apical LS phenomenon in this population.
Study design and subjects
A total of 32 subjects with AL CA were prospectively recruited, including 10 patients who had previously participated in a pilot study (10) and 22 patients who are part of an ongoing clinical trial. All 32 subjects underwent echocardiography and myocardial 18F-florbetapir PET, and 22 patients were additionally investigated with CE-CMR. Immunoglobulin AL amyloidosis was diagnosed in all subjects using standard criteria (11). Cardiac involvement was diagnosed by the following: 1) positive endomyocardial biopsy with typing of amyloid by immunohistochemistry or mass spectroscopy (n = 18; 43%); or 2) typical echocardiographic findings of amyloid heart disease (LV wall thickness >11 mm, bright echogenic myocardium, evidence of restrictive physiology) and elevated troponin T or age-adjusted N-terminal pro-B-type natriuretic peptide (NT pro BNP). Subjects with ATTR CA and patients with AL amyloidosis but without cardiac involvement were not included in this study. This study was approved by the Partners Human Research Committee. Each study subject provided written informed consent.
Two-dimensional echocardiography was performed in all subjects using standard clinical protocol per American Society of Echocardiography recommendations (12). Conventional analysis of the LV included measurements of wall thickness, LV ejection fraction (LVEF), and diastolic parameters. Deformation analysis of the left ventricle based on two-dimensional speckle-tracking imaging was performed offline using dedicated software (Image Arena v. 4.6, TomTec, Unterschleißheim, Germany).
18F-florbetapir PET/computed tomography
Individuals were positioned with the help of a computed tomography (CT) topogram and a low-dose CT scan of the heart was acquired for attenuation correction of PET emission data. 18F-florbetapir (∼222 to 370 MBq [6.0 to 10.0 mCi]) was injected through an intravenous catheter 1 min after the start of a 60-min list-mode three-dimensional PET cardiac acquisition. Myocardial retention of 18F-florbetapir, defined as the activity concentration in myocardial tissue between 10 and 30 min, normalized to the integral of the image-derived arterial input function (Online Figure 1), was determined using Carimas v. 2.9 (Turku PET Centre, Turku, Finland). Myocardial florbetapir retention units are per gram of tissue (μmol/min/g). Therefore, total florbetapir binding was also calculated by multiplying florbetapir retention index (RI) by the extracellular LV mass component (see below).
Contrast-enhanced CMR (n = 22)
All CMR images were acquired on a 3.0-T system (Tim Trio, Siemens, Erlangen, Germany), with electrocardiographic gating and breath-holding. The protocol consisted of steady-state free-precession cine imaging for assessing ventricular function and morphology, and native and post-contrast T1 mapping for quantification of myocardial ECV. Cine imaging was obtained from a standard stack of short-axis slices (8 mm thick, no gap) and 3 long-axis planes. Measurement of myocardial T1 was performed during a single breath-hold at end-diastole in 3 short-axis slice positions, at the basal, mid, and apical LV levels, by using the modified Look-Locker inversion recovery method (Online Figure 2) (13). T1 mapping images were acquired in the same 3 LV short-axis slices, once before (native T1) and 2 times, at 10 and 20 min (post-contrast T1), after the injection of 0.1 mmol/kg of gadoterate meglumine (Dotarem, Guerbet LLC, Bloomington, Indiana). ECV was calculated from the formula: (1−hematocrit) = (1/T1 Myo Post − T1 Myo Pre) / (1/ T1 Blood Post − 1/ T1 Blood Pre). Global myocardial ECV for an individual was calculated by averaging the myocardial segmental ECV values from the short-axis slices at the base, mid, and apical LV levels (Online Figures 2E and 2F). Subsequently, the extracellular and cellular LV mass components were calculated from the following formulas: 1) extracellular LV mass = LV mass·ECV; and 2) cellular LV mass = LV mass·(1−ECV) (7,14).
Commercially available software (MedisSuite 3.0 Medical Imaging Systems, Leiden, the Netherlands) was used to post-process and quantify LV volumes, ejection fraction, mass, wall thickness, dimensions, fractional shortening, and wall stress (WS = 1.35 x LVSP x Ac/Aw, where 1.35 is a conversion factor from mm Hg to 1,000 dyne/cm2; LVSP = noninvasive systolic blood pressure; Ac = LV end-diastolic cavity area; and Aw = LV end-diastolic wall area) (15). Measurements were obtained globally (when applicable) and segmentally at the basal, mid, and apical levels (Online Figures 3 to 6). LV wall thickness was measured following a modified American Heart Association 17-segment model.
Comparison with reference CMR values
We analyzed the data using STATA (version 13.1). Continuous variables are shown as either mean ± SD for normally distributed data or median (interquartile range) for non-normal data. p value was used for testing for a significant base-to-apex gradient. Repeated-measures analysis of variance was performed to compare mean values of LV walls (heterogeneity), combined with Bonferroni test for post hoc analysis and correction for multiple comparisons. Simple correlations were assessed using longitudinal regression analysis and Pearson correlation coefficient. Except for myocardial strain (derived from echocardiography) and 18F-florbetapir uptake, all regional (basal-mid-apical LV) parameters for this analysis were derived from CE-CMR. Finally, quartile groups were created based on maximal LV wall thickness. All statistical tests were 2-tailed, and p < 0.05 was considered statistically significant.
Baseline characteristics for all 32 subjects included in the study are illustrated in Table 1. One-half of AL subjects were male, and about half were treatment-naive at time of enrollment. Lambda was the predominant monoclonal light chain, N-terminal pro-BNP was almost invariably elevated, and renal function was generally preserved or mildly impaired in most subjects.
Echocardiography revealed the typical features of amyloidosis including small LV cavity, thickened LV walls, preserved LVEF, enlarged atria, and moderate to severe LV diastolic dysfunction (Table 1).
Global (when applicable) and regional echocardiographic, PET, and CE-CMR imaging characteristics are summarized in Table 2. Mean global LS (−12.3% ± 7.7; minimum −5.03; maximum −19.4), but not CS (−21.3% ± 6.9; minimum −6.7; maximum −34.8) were significantly impaired, despite relatively preserved LVEF (53 ± 10%; minimum 35; maximum 70) in most subjects. Regionally, LS, CS (Figures 1A and 1B), relative LVEF, and fractional shortening showed a significant base-to-apex gradient, with the apical segments showing less dysfunction compared with the mid (p < 0.001) and basal LV segments (p < 0.001). LV mass, wall thickness (Figures 1C and 1D), volumes, and wall stress also showed a clear base-to-apex gradient, with the apical segments demonstrating significantly lower LV mass, wall thickness, and volumes (thus higher wall stress) compared with the basal (p < 0.001) and mid wall segments (p < 0.001).
18F-florbetapir retention (0.089 ± 0.03 μmol/min/g; minimum 0.05; maximum 0.17) and ECV (0.51 ± 0.07; minimum 0.38; maximum 0.63) were globally abnormal in all patients. However, regionally myocardial florbetapir retention, native T1, and ECV showed no statistically significant base-to-apex gradient. In contrast, total florbetapir binding and extracellular LV mass did show a marked base-to-apex gradient as expected (Table 2, Figure 2). Moderately strong regional correlations were observed between LS and markers of total amyloid load, but only weak correlations were seen with florbetapir retention index or ECV (Figure 3).
Comparison with reference CMR cohort
Globally, indexed LV mass was 65% (86 ± 18 g/m2 vs. 52 ± 13 g/m2; p < 0.0001), ECV 96% (0.51 ± 0.07 vs. 0.26 ± 0.05; p < 0.0001), and extracellular LV mass 152% (44 ± 14 g/m2 vs. 13.6 ± 1.7 g/m2; p < 0.0001) greater, whereas cellular LV mass (42 ± 8 vs. 44 ± 10 g/m2; p = 0.40) was comparable, compared with reference adults (Online Table 1). Regionally, average LV wall thickening at the base (12.3 ± 1.5 mm), mid (10.7 ± 2.0 mm), and apex (7.6 ± 1.9 mm) corresponded to a 65%, 67%, and 21% increment of the reported average LV wall thickness at the basal (7.5 ± 1.3 mm), mid (6.4 ± 1.2 mm), and apical (6.3 ± 1.1 mm) segments of normal individuals (Online Table 2).
Regional mass contribution to global LV mass was 47 ± 5% for the basal-, 38 ± 4% for the mid-, and 15 ± 3% for the apical-LV segments (Table 2). This mass relation was maintained across quartile groups of maximal LV wall thickness, and regional differences in LS were observed across all groups (Online Figure 7).
We found that RELAPS in CA is most likely explained by regional differences in the distribution of total amyloid mass, and not the proportion of amyloid deposits as estimated by advanced PET and CE-CMR techniques.
One of our study aims was to first investigate the regional distribution of specific (florbetapir) and non-specific (ECV) probes of amyloid infiltration. Florbetapir is a stilbene derivative used in clinical practice as a potent marker of amyloid deposits in tissue (18–21). In human hearts, florbetapir uptake co-localized remarkably well to histological extracellular AL amyloid deposits in autopsy-derived myocardial tissue ex vivo (9) and also proved to be both sensitive and specific for detection of CA in vivo (8). CE-CMR, on the other hand, provides an estimate of the myocardial ECV derived from measurements of the myocardial tissue longitudinal relaxation time constant (T1) before and after the administration of an extra-cellular gadolinium-based contrast agent. In CA, ECV appears to reflect the amount of amyloid deposition in the extracellular space (22). Prior studies have shown that ECV is universally expanded in CA, even in cases where focal LGE is absent (up to 24% in one series) (5).
Both imaging modalities yielded abnormal global results in all subjects, and yet we failed to observe a significant base-to-apex gradient. The most feasible explanation is that imaging results from these modalities reflect tissue concentration or proportion, rather than an estimate of the whole tissue mass of amyloid deposition. Specifically, myocardial florbetapir (retention index) yields quantitative PET values per unit mass of tissue (concentration), whereas ECV is measured as the percent of tissue comprised of extracellular space (proportion). This is in contrast to LV wall thickening, a crude marker of total amyloid mass, which exhibited marked regional variation, wherein LV wall thickening at the basal and mid wall segments was at least 3-fold greater than at the apex, consistent with the fact that 85% of the total LV mass was concentrated between the basal and mid wall segments. Our study also confirmed prior observations that LV wall thickening in AL occurs almost exclusively from expansion of the extracellular matrix (7), as the cellular myocardial mass remained preserved compared with published reference values (7,14) and also showed that this extra-cellular LV mass is florbetapir-avid, thus allowing for quantitative methods capable of estimating the total amyloid load in the heart.
Taken together, these data indicate that although the proportion of amyloid infiltration is not remarkably different across the LV myocardium, the total amyloid mass is disproportionately greater toward the basal and mid wall segments compared with the apex. These findings also illustrate the importance of differentiating total mass from the proportion of amyloid deposits in tissue, 2 related concepts that should not be used interchangeably, but also highlight the importance of integrating morphological findings with tissue characterization to better estimate the total amyloid load in the myocardium.
LS is universally impaired in patients with symptomatic amyloid heart disease, and, although not pathognomonic, RELAPS is consistently seen in all patients with this condition and is certainly supports its diagnosis. The leading hypothesis to explain the RELAPS phenomenon has been lesser amyloid infiltration at the apical, relative to the basal and mid wall segments. Our study not only confirms prior observation but provides new insights into the pathophysiology of RELAPS in CA by revealing that total amyloid mass, and not the proportion of amyloid infiltration, is what most likely leads to the marked regional differences in LS seen in CA.
There are several limitations in our study that require discussion. First, not every patient underwent PET and CE-CMR; consequently, most of our analyses are limited to 69% (n = 22) of the study population. A control group is lacking; however, comparison with a well-established reference cohort was performed in an attempt to overcome this limitation, and in all likelihood, repeating this study using our own controls would not have changed the results or the conclusions. We evaluated a relatively small sample size, which may have affected the statistical power, so it is possible that reported trends may become significant in a larger study sample. We did not include patients with ATTR; it remains to be investigated whether our findings are applicable to those patients as well. Last, although our study reveals a strong association between regional changes in myocardial strain and LV mass, it does not allow for a direct mechanistic explanation, nor does it provide any insight as to why the basal and mid LV wall segments are disproportionately more involved relative to the apex. This will remain a topic of future investigation.
In summary, we observed that differences in the segmental distribution of total amyloid mass, and not the tissue concentration or proportion of amyloid deposits as estimated using advanced PET and CE-CMR techniques, is what determines the marked base-to-apex gradient of longitudinal strain (RELAPS) in CA. Our findings call for further methods that that can integrate both morphological and tissue characterization findings to better estimate the total amyloid load in the heart.
COMPETENCY IN MEDICAL KNOWLEDGE: Studying the mechanisms of infiltration of light chain amyloidosis and their clinical consequences on the heart are of utmost importance. In the present study, using advanced multimodality imaging techniques aimed at quantifying total amyloid load in the heart, we found that the increment in LV wall thickness occurs exclusively as a result of diffuse expansion of the extracellular matrix from amyloid infiltration. Interestingly, the density of amyloid deposits and the expansion of the extracellular volume are proportionately comparable from base to apex. However, only 15% of the total amyloid load is concentrated at the apical segments, likely explaining the relative apical sparing of longitudinal strain observed in CA.
TRANSLATIONAL OUTLOOK: Using targeted PET probes and advanced tissue characterization techniques, it is now feasible to noninvasively estimate the total amyloid mass in the heart. This may become an important clinical endpoint in future studies assessing treatment response in patients with light chain amyloidosis.
Drs. Bravo and Fujikura were supported by the National Institutes of Health T32 training grant (1T32HL094301). Drs. Dorbala, Falk, Kijewski, Di Carli, Jerosch-Herold, and Kwong are supported by National Institutes of Health RO1 grant (RO1 HL 130563); Dr. Dorbala is supported by an American Heart Association grant (AHA 16 CSA 2888 0004). Dr. Di Carli has received consulting fees from Sanofi and General Electric. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- light chain amyloidosis
- transthyretin amyloidosis
- cardiac amyloidosis
- contrast enhanced
- cardiac magnetic resonance
- computed tomography
- extracellular volume fraction
- late gadolinium enhancement
- longitudinal strain
- left ventricular
- left ventricular ejection fraction
- NT pro BNP
- N-terminal pro-B-type natriuretic peptide
- positron emission tomography
- relative apical sparing of LS
- single photon emission computed tomography
- Received March 12, 2018.
- Revision received May 15, 2018.
- Accepted June 26, 2018.
- 2018 American College of Cardiology Foundation
- Quarta C.C.,
- Solomon S.D.,
- Uraizee I.,
- et al.
- Ternacle J.,
- Bodez D.,
- Guellich A.,
- et al.
- Phelan D.,
- Collier P.,
- Thavendiranathan P.,
- et al.
- Sperry B.W.,
- Vranian M.N.,
- Tower-Rader A.,
- et al.
- Banypersad S.M.,
- Sado D.M.,
- Flett A.S.,
- et al.
- Fontana M.,
- Pica S.,
- Reant P.,
- et al.
- Dorbala S.,
- Vangala D.,
- Semer J.,
- et al.
- Park M.A.,
- Padera R.F.,
- Belanger A.,
- et al.
- Dorbala S.,
- Vangala D.,
- Bruyere J. Jr..,
- et al.
- Gertz M.A.,
- Comenzo R.,
- Falk R.H.,
- et al.
- Lang R.M.,
- Badano L.P.,
- Mor-Avi V.,
- et al.
- McDiarmid A.K.,
- Swoboda P.P.,
- Erhayiem B.,
- et al.
- Kawel-Boehm N.,
- Maceira A.,
- Valsangiacomo-Buechel E.R.,
- et al.
- Campbell-Washburn A.E.,
- Price A.N.,
- Ellmerich S.,
- et al.