Cardiac amyloidosis is prevalent in older patients with aortic stenosis and carries worse prognosis

A comprehensive TTE was performed according to the guidelines [16] in accredited Echocardiography Lab by the Intersocietal Commission for the Accreditation of Echocardiography. The pulsed-wave Doppler transmitral inflow velocities and tissue Doppler-derived mitral annular velocities were obtained from apical 4-chamber views for assessment of diastolic function in accordance with the societal guidelines (i.e., pulsed-wave Doppler-derived transmitral inflow velocities of the early phase (E) and late phase (A) of diastole and pulse-wave tissue Doppler-derived mitral annular velocity imaging for both septal and lateral walls) [17]. The E/e’ ratio of early diastolic filling/tissue Doppler velocity annulus was used as the surrogate of LV filling pressures. Longitudinal LV systolic function was also obtained by measurement of the peak systolic velocity from both medial and lateral annuli. Spectral Doppler of the LV outflow tract (LVOT, pulse wave) and aortic valve (AV, continuous wave) were measured using the best baseline view determined for Doppler assessment. For each Doppler measurement 3 cycles were averaged and post–premature ventricular contraction beats were discarded (5 cycles were averaged for patients with atrial fibrillation). Aortic valve area (AVA) was calculated by using the continuity equation formula: AVA = (LVOT area x LVOT VTI)/AV VTI [18] (LVOT = left ventricular outflow tract; VTI = velocity time integral). Stroke volume was calculated by Doppler using the VTI of the LVOT and its diameter in midsystole in the parasternal long-axis view. Stroke volume index was calculated by indexing stroke volume to body surface area. Low-flow, low-gradient AS was defined by stroke volume index <35 ml/m² and aortic valve mean gradient was <40 mmHg. Severe AS was defined as an indexed AVA ≤ 0.6 cm²/m² [19].

CMR was performed with a 1.5 T CMR system (Magnetom Espree Siemens Healthineers, Erlangen, Germany) using a 32-channel phased array cardiovascular coil in a CMR laboratory accredited by the Intersocietal Commission for the Accreditation of Magnetic Resonance Laboratories employing dedicated CMR technologists. Balanced steady-state free precession cine images were acquired (slice thickness 6 mm, 4 mm gap) during 5 to 10-s breath holds in the usual short and long axis orientations. CMR assessment of LV volumes, LV ejection fraction (LVEF), LV mass was performed by manual tracing of the endocardial borders at end-diastole and end-systole in each of the short-axis slices as per standard clinical evaluation.

To identify CA as part of the CMR, LGE imaging was performed 10–15 min after a 0.2 mmol/kg IV gadoteridol bolus (Prohance, Bracco Diagnostics, Princeton, New Jersey, USA) using phase-sensitive inversion recovery (PSIR) pulse sequence which recently has been shown to be an accurate method to identify CA and prognostically important [14]. We identified CA by CMR when LGE imaging demonstrated characteristic pattern of marked myocardial enhancement (either subendocardial or transmural) with associated morphologic findings such as increased LV wall thickness and abnormal myocardial and blood pool kinetics [11, 14]. Applying similar classification used by Fontana et al., a patient with basal transmural LGE but mid/apical subendocardial LGE would be classified as having transmural LGE [14].

Given the retrospective nature of this report, patient’s advanced age, frailty and multiple comorbidities, AL (light-chain) amyloidosis was not systematically excluded via invasive cardiac biopsy or 99^m–Tc-pyrophosphate nuclear scan (introduced at our center in late 2015), which has recently shown to be an excellent non-invasive method to confirm wtATTR CA [20]. Nonetheless, primary CA was attempted to be excluded based on negative bone marrow and fat-pad biopsies (1 patient); positive 99^m–Tc-pyrophosphate nuclear scan, Perugini grade 3, which is more specific for wtATTR (1 patient) and negative serum/urine immunoelectrophoresis and immunofixation for paraproteins (4 patients). All imaging studies were analyzed and interpreted by experienced CMR readers, blinded to the clinical outcomes. CMR results were reported to the clinicians caring for the patients.

A subset of patients with AS and with AS + CA had native T1 mapping and Extracellular Volume Fraction (ECV) available for measurement. As previously reported by our group [21], T1 mapping was obtained using breath-held modified Look-Locker inversion recovery (MOLLI) sequence after generation of in-line parametric mapping of the basal and mid LV short-axis slices, averaging measures from the middle third of the myocardium, to avoid partial volume effects. Regions of interest excluded myocardium with myocardial infarction and carefully avoided myocardium near infarcted myocardium. We did not exclude foci of nonischemic scar on LGE images (ie, atypical of myocardial infarction) from ECV measures acquired in noninfarcted myocardium. Quartiles of native T1 mapping and ECV values were tested for their association with all-cause mortality and stratified according to the treatment received (AVR vs. medical therapy).

Surgical Thoracic Society Predicted Risk of Mortality (STS-PROM) was calculated for each patient according to the planned treatment (i.e.: AVR +/− coronary artery bypass grafting) using Society of Thoracic Surgery online calculator (version 2.73) which is a well-validated composite score comprised of over 40 clinical parameters and risk-factors.

Baseline demographic data and clinical variables were summarized with continuous variables expressed as mean ± SD and categorical data presented as frequency (percentage). Differences between the groups were compared with the Student’s t test for continuous variables and the chi-square test for categorical variables. AVR performed either via surgical or TAVR method was considered as a time-dependent covariate. The primary end-point was all-cause mortality after CMR study. Univariable models tested the association of clinical risk factors and imaging findings to all-cause mortality. Multivariable Cox regression models were used to assess the relationship between wtATTR and all-cause mortality. Statistical analysis was performed using SAS software (version 9.4, SAS Institute, Cary, North Carolina, USA).

A total of 113 consecutive patients with moderate and severe AS (59 males, median age 74 years, interquartile range: 62–82 years) were studied (96/113, 85% with severe AS). AS combined with CA was present in 9 patients (8%, all >80 years; 8/9 males). The median time interval between the clinical CMR study and TTE study was 6 days (interquartile range of 0–15 days). Baseline clinical and imaging characteristics are summarized in Table 1. The average age for patients with CA was higher than those with isolated AS (88 ± 6 vs. 70 ± 14, P < 0.0001, Fig. 1). Among those over the median age of 74 years, the prevalence of CA was 9/57 (16%), and after excluding women, the prevalence was 8/25 (32%). In our cohort, 1 out of 4 male octogenarians presenting with symptomatic AS were found to have concomitant CA detected by CMR.

Patients with CA also had on baseline TTE larger left atrial volumes, smaller indexed AVA, and poor longitudinal function by tissue Doppler. The prevalence of severe AS was not statistically different between groups (77% vs 89%, P = 0.43, Table 1). Consistently, the CMR showed pronounced concentric LV remodeling in CA with higher LV mass index, higher mass/volume ratio and lower stroke volume index. The average LVEF trended lower in CA patients (43 ± 17% vs. 52 ± 18%, P = 0.18).

Native T1 and ECV values were available in 72% and 63% of the cohort, respectively, and significantly higher in AS + CA patients, when compared to patients with only AS. (Table 1). At our center, normal values for native T1 mapping and ECV for healthy controls are 1016 ± 28 msec and 23.7 ± 2%, respectively [22]. STS-PROM was also higher in the AS and CA patients (6.9 ± 4.2% vs. 3.8 ± 3.7%, P = 0.02), consistent with higher surgical risk and greater burden of comorbidities. There was a high burden of atrial fibrillation in CA patients compared to isolated AS (67% vs 20%, P = 0.006). Of note, none of the CA patients met electrocardiographic criteria for low voltage. On the other hand, low-flow, low-gradient AS physiology was very common in CA (7/9 patients or 78% vs. 47/104 or 45% with AS, P = 0.06) (Table 2). On LGE imaging, all 9 patients with CA demonstrated typical transmural myocardial involvement as shown in Fig. 2.

Table 2 provides a detailed characterization of these 9 patients identified with AS + CA.

Over the follow-up period (median 18 months, interquartile range [IQR]: 11–30 months), 59 patients received an AVR (42 surgical AVR and 17 a TAVRs), and 40 patients (35%) died. Patients with CA had significantly higher 1-year all-cause mortality than patients with isolated AS (56% vs. 20%, P < 0.001, Fig. 3). Among patients with isolated AS, 55/104 (53%) received AVR and among patients with CA, 4/9 (44%) received AVR, all of them transcatheter AVR.

Univariate analysis of all-cause mortality showed several associations between mortality and patient characteristics (Table 3). Limited events constrained the number of variables that could be included in the multivariable Cox regression model (10 events per predictor variable). Hence, multiple models were created to determine whether CA was a risk factor associated with mortality when adjusting for potential confounders. After adjustment for AVR, STS PROM, LVEF, NYHA Class ≥ III at presentation, the presence of CA remained a predictor of all-cause mortality with similar hazard ratio (Table 4, Models 1 through 3). Since CA was only present in relatively older AS patients, a subgroup analysis was performed in the subgroup aged greater than the total population’s median age of 74 years. Within this older subgroup, the presence of CA, remained a predictor of all-cause mortality with nearly 3-fold increased risk (HR = 2.87, 95% CI 1.02–8.05, P = 0.04).

Exploratory analysis of patients with native T1 and ECV values available for analysis demonstrated that although native T1 was not associated with the primary outcome (Fig. 4), ECV was associated with increased risk for all-cause mortality in a progressive dose-response manner (Fig. 5). Similar findings were observed when analysis was stratified based on the performance of AV replacement (Fig. 6, Panel A and Fig. 7, Panel A) vs. medical therapy (Fig. 6, Panel B and Fig. 7, Panel B). Of interest, for those patients with moderately-severe AS and managed conservatively without AVR, being in the lowest ECV quartile (ie: < 24.8%) was associated with a “protective” effect. Specifically, there were no deaths noted up to 12 months of follow-up for those patients in the lowest ECV quartile (Fig. 5 and Fig. 7, Panel A).

Our study has limitations. First, relying predominantly on LGE imaging pattern to detect CA, we may have misclassified patients’ amyloidosis status. A novel algorithm recently proposed by Gillmore and colleagues using upfront nuclear scintigraphy scan for patients with heart failure and clinically suspected findings on CMR [26] is currently what has been used in our center starting in late 2015 when Tc-PYP scan became available. We acknowledge that the absence of endomyocardial biopsy data in these patients prevented definitive diagnosis of CA and subtyping. However, endomyocardial biopsy is not only invasive for these patients who have a high burden of comorbidities and frailty but can also be associated with sampling error. We believe that the totality of the clinical evidence reasonably supported its diagnosis which was accepted by clinicians caring for the patient. Whether the combined use of LGE pattern and ECV measurements could improve the sensitivity to detect CA is unknown needs to be tested prospectively. Nonetheless, regardless of the potential misclassification we obtained significant findings and results.

Second, this single-center study of the referred CMR patients may involve selection bias that could influence disease prevalence estimates. Our CMR eligibility criteria exclude patients with cardiac devices (defibrillators, pacemakers) or renal dysfunction (i.e., estimated glomerular filtration rate < 30 ml/min/1.73 m² given the need for gadolinium contrast. Yet, these exclusions may lower prevalence estimates, since both chronic kidney insufficiency and of arrhythmia/conduction disease requiring implantable cardiac devices increase with aging, the potential initial overestimation of a “selected CMR cohort” might be counter-balanced by the relative contra-indications for CMR study eligibility and performance. Supporting that observation is the recent publication by Castaño who systematically obtained Tc-PYP scanning in patients who received TAVR (mean age = 84 ± 6 years) [33]. The authors identified the prevalence of 16% for ATTR-CA, which is similar prevalence noted to our cohort when we considered those above median cohort age of 74 years. Nonetheless, our cohort represents a selected group of AS patients who received CMR for certain clinical indications.

Third, data on other clinical outcomes such as cardiac-specific mortality, heart failure hospitalizations, were not always available. Our limited sample size did not permit subgroup analyses yet it was sufficiently large to permit multivariable survival analysis for the first time. Lastly, CMR techniques using gadolinium are not applicable to some of these patients with advanced renal dysfunction. Evolving CMR quantitative capabilities using native (noncontrast) T1 mapping [34, 35] and ECV may prove useful but require further validation in larger cohorts.

It is important to note that not all providers embraced CMR to characterize myocardial fibrosis in the myocardium of AS patients (as articulated by Dweck et al. [36]), which underscores the potential for referral bias.