Global Longitudinal Strain for Prognostic Staging in Wild-Type Transthyretin Cardiac Amyloidosis

Abstract

BACKGROUND:

A formal prognostic staging system in wild-type transthyretin amyloid cardiomyopathy (ATTRwt-CM), based on echocardiographic imaging, is lacking. We evaluated the prognostic performance of global longitudinal strain (GLS) staging in a large cohort of patients with ATTRwt-CM, including under tafamidis treatment and relative to National Amyloidosis Center (NAC) biomarker staging.

METHODS:

A multicentric, international ATTRwt-CM patient cohort with baseline GLS (distribution quartiles), evaluated by echocardiography, was studied, related to all-cause mortality.

RESULTS:

The study comprised 816 patients with ATTRwt-CM, median age of 81.5 years, 83% males, and 72% tafamidis initiated. During a 2.2-year median follow-up, 29.7% of patients died. GLS worsened with increasing NAC disease stage (I: −14.3%, II: −11.6%, III: −11.4%; P<0.001). Median survival per baseline GLS quartile stage 1 (<−15.8%), 2 (−15.8 to −12.9%), 3 (−12.8 to −10.0%), and 4 (GLS >−10.0%) was not met, 6.7, 4.6, and 3.4 years, respectively (P<0.001). The median GLS −12.8% cutoff predicted 1-year mortality with 74% sensitivity, 52% specificity (area under the curve, 0.73 [95% CI, 0.66–0.80]; P<0.001). GLS was the only independent echocardiographic and strong mortality predictor, independent of other predictors, including age, New York Heart Association class symptoms, NAC stage, and tafamidis treatment (hazard ratio, 1.08 [95% CI, 1.04–1.12]; P<0.001), also when restricted to 591 tafamidis-treated subjects (hazard ratio, 1.15 [95% CI, 1.08–1.22]; P<0.001). Baseline GLS −12.8% cutoff value provided further prognostic discriminative ability for mortality within each NAC disease stage stratum (all P<0.050). Likelihood ratio test indicated incremental prognostic value of GLS (staging) over baseline NAC staging (P<0.001).

CONCLUSIONS:

GLS is a strong, independent mortality predictor in ATTRwt-CM, irrespective of tafamidis treatment, that may be an adjunct or complementary to biomarker staging.

CLINICAL PERSPECTIVE.

Current report is the first to validate global longitudinal strain (GLS) by echocardiography in wild-type transthyretin amyloid cardiomyopathy as an independent predictor of all-cause mortality in a multicentric setting and including contemporary tafamidis-treated patients. GLS assessment by echocardiography represents a robust and reproducible systolic function metric, integrating many prognostic variables in wild-type transthyretin amyloid cardiomyopathy. Moreover, our analysis implies a novel imaging-based prognostic staging system, relying on baseline GLS assessment at time of wild-type transthyretin amyloid cardiomyopathy diagnosis. Baseline GLS, easily assessable at first-line echocardiographic imaging, offers prognostic staging as an adjunct or complementary to National Amyloidosis Center serum biomarker staging, demonstrating incremental value by further refining risk stratification within National Amyloidosis Center disease stage categories. Prospective studies should determine whether GLS-guided decision-making can improve patient outcomes.

See Editorial by Chetrit and Masri

Wild-type transthyretin amyloid cardiomyopathy (ATTRwt-CM) comprises age-related extracellular myocardial deposition of insoluble TTR (transthyretin) amyloid fibrils, formed by dissociation of the TTR transporter protein, in patients without TTR gene mutation.^1,2 Consequently, heart failure due to progressive myocardial wall thickening and premature death characterizes the natural disease course.^1,3 The worldwide tremendous diagnostic uptake reported over the past decade confirms its significant prevalence, particularly in elderly patients above 65 years who present with increased wall thickness and heart failure symptoms.^3,4 Although incurable to date, several disease-modifying therapies, including the TTR stabilizer tafamidis, slow down disease progression, conveying significant mortality benefit.^3,5 Estimating disease prognosis is fundamental to patient management, with various clinical, biochemical and cardiac imaging markers being related to ATTRwt-CM mortality.⁶ The National Amyloidosis Center (NAC) serum biomarker staging, applying NT-proBNP (N-terminal pro-brain natriuretic peptide) and estimated glomerular filtration rate cutoffs, provides clinically widely adopted baseline prognostication, validated in patients with ATTRwt-CM and variant-type transthyretin amyloid cardiomyopathy (ATTR-CM) without disease-modifying treatment.^4,7

Echocardiography remains undisputably the first-line imaging technique to raise suspicion for ATTR-CM in patients presenting with increased wall thickness.⁸ Moreover, 2-dimensional speckle-tracking left ventricular (LV) global longitudinal strain (GLS) is nowadays implemented in many centers and recommended as a reproducible and accurate metric to assess LV systolic function, providing superior sensitivity compared with conventional LV ejection fraction (LVEF) analysis in most myocardial disorders.^2,8 In ATTR-CM, fibril deposition starts at the basal and subendocardial myocardium, but progresses to patchy infiltrations encompassing the entire myocardium, causing progressive ventricular pseudohypertrophy and significant impairment of longitudinal ventricular function.⁸ As longitudinal fibers dominate subendocardial myocardial architecture, impaired GLS is highly prevalent among patients with ATTRwt-CM, often presenting with relative apical sparing of longitudinal deformation, and occurring early in the disease process.^8,9 Few studies, however, evaluated the prognostic value of GLS to predict mortality specifically in ATTRwt-CM, limited by small sample size, single-center origin and predominantly relying on limited GLS acquisition techniques, often restricted to single-plane analysis.^10,11 Moreover, these studies lacked comparison to NAC biomarker staging or validation in contemporary tafamidis-treated populations. Hence, a formal staging system in ATTRwt-CM based on echocardiographic imaging is currently lacking.²

Therefore, the aim of current study was 3-fold: (1) to explore the role of baseline GLS (staging) to predict mortality in a large, multicentric cohort of patients with ATTRwt-CM, (2) to evaluate its prognostic value associated with tafamidis treatment, and (3) in relation to NAC stages. We hypothesized GLS to be a significant mortality predictor, irrespective of tafamidis treatment and NAC stage, that may be used as an adjunct or complementary to biomarker staging.

Methods

Data can be shared upon reasonable request and after approval of all participating centers.

Study Population

Patients from an ongoing international cohort of consecutively diagnosed ATTRwt-CM subjects until December 2024 at 14 tertiary heart centers (10 Belgian, 1 French, 1 Austrian, 1 Dutch, and 1 Romanian) were included, if baseline LV echocardiographic GLS was assessed at cardiac diagnosis (baseline). ATTR-CM was diagnosed as recommended: (1) ATTR deposition on endomyocardial biopsy; (2) 99-technetium bone-scintigraphy with intramyocardial grade 2 or 3 radiotracer uptake, without monoclonality; or (3) extracardiac ATTR tissue deposition with suggestive noninvasive cardiac imaging findings.⁴ All patients with pathogenic TTR gene variants were excluded. Baseline demographics, cardiovascular risk factors, cardiovascular and tenosynovial history, heart failure treatments, laboratory markers, and symptomatic status by New York Heart Association (NYHA) class, were all retrieved from electronic patient chart review. In addition, tafamidis initiation was inventoried at baseline and during follow-up. Prognostic cardiac disease stage was evaluated by the NAC serum biomarker staging system, applying prespecified cutoff criteria: NT-proBNP (>3000 pg/mL) and estimated glomerular filtration rate (<45 mL/min per 1.73 m²), defining NAC stage I (0 criterion), II (1 criterion), and III (2 criteria), with higher stages reflecting more advanced disease stage.⁷ Due to the retrospective study nature, local ethical committees waived the need for written informed consent for the cardiac amyloidosis data registry, which was, however, obtained in most patients. The study complies with the Declaration of Helsinki and standards of Good Clinical Practice.

Echocardiography

All patients underwent a comprehensive baseline echocardiographic evaluation with commercially available high-end cardiac ultrasound machines (GE Vingmed, Philips, Siemens) with ECG-gated image acquisition. Conventional myocardial and valvular structure and function parameters were assessed, as recommended.^12,13 Significant LV valvular disease was defined as ≥moderate regurgitation or stenosis of the mitral or aortic valve. In addition, LV 2-dimensional speckle-tracking echocardiographic GLS (further mentioned as GLS) was assessed by (semi) automated online or offline analysis, averaging evaluation from apical 4-, 3-, and 2-chamber acquisitions, applying a 17-segment LV model, conforming to current recommendations.¹² Of note, more negative GLS values represent better longitudinal myocardial deformation and function, with a value of ≤−18.0% considered to be normal.^12,14

Follow-Up and Study End Point

Outpatient follow-up was provided every 6 to 12 months, with tafamidis initiation at the discretion of the treating physician. Of note, at data collection closure in December 2024, tafamidis was the single reimbursed disease-modifying drug therapy available in Europe for ATTR-CM indication. The single study end point comprised all-cause mortality, available for all subjects as electronic patient files are automatically coupled to the National Death Registers.

Statistics

Continuous variables were given as mean with SD, or median and interquartile range (if highly skewed distribution), with comparisons made by Student t or Mann-Whitney U test, as appropriate. Categorical data were summarized as frequencies and percentages, and compared by χ² or Fisher exact test, as applicable. Comparisons of >2 groups were performed by 1-way Anova or Kruskal-Wallis test, as indicated.

First, from the overall patient cohort, baseline characteristics between patients with versus without baseline GLS assessment were compared, retaining only patients with baseline GLS assessment for the current study. Second, the median GLS value was determined, and the association between LVEF and GLS was explored by Spearman rank correlation coefficient (r). Third, the study population was then stratified into stages based on baseline GLS distribution quartiles to facilitate risk stratification and assess potential threshold effects on outcomes, similar to previous reports on light-chain cardiac amyloidosis.¹⁵ Survival outcomes were assessed by the Kaplan-Meier method, with comparisons made by Log-rank testing; when >50% of patients within the stratification were still alive at time of analysis, median survival was reported as not met. Fourth, a receiver operating curve with area under the curve for GLS to predict 1-year mortality was created. Fifth, the association of GLS (and other variables) with mortality was investigated using Cox regression analysis. The significance of individual covariates in the Cox proportional hazards model was assessed using the Wald χ² test, calculated from the estimated regression coefficient and its SE, with higher Wald scores representing a stronger association with all-cause mortality. To avoid immortal time bias, tafamidis treatment was evaluated as a time-dependent covariate. Variables showing P<0.050 significance at the univariable level were selected for multivariable analysis. A first model evaluated baseline echocardiographic parameters of interest, including GLS. A second model then tested significant echocardiographic parameters with P<0.050 from model 1 with established prognostic variables of interest. Sixth, interaction between GLS and tafamidis treatment was verified by the supremum test for the proportional hazards assumption. Finally, a likelihood ratio test was performed, evaluating the potential incremental prognostic value of GLS (stages) over baseline NAC disease stage and age, established prognosticators in ATTR-CM. Statistical analysis was performed on IBM SPSS Statistics version 25 and SAS software version 9.4; all tests were 2-sided, with P<0.050 regarded as statistically significant.

Results

Study Population

In total, 816 patients (62% Belgian) from the initial international patient cohort of 1454 ATTRwt-CM subjects had baseline GLS evaluation and were eligible for the current study (Table S1; Figure S1). Lack of baseline GLS data is mainly related to GLS not being routinely measured in some centers, based on contemporary local practices. Diagnostic period, tafamidis initiation and LVEF were comparable between patients with versus without GLS assessment (all P=nonsignificant; Table S2). Subjects without baseline GLS evaluation were slightly older, more female patients with a higher proportion of atrial fibrillation (all P<0.05). Importantly, subjects with versus without baseline GLS availability had similar survival, without GLS availability being a mortality predictor (hazard ratio [HR], 0.89 [95% CI, 0.73–1.09]; P=0.247; Figure S1). The final study population comprised a typical cohort of patients with ATTRwt-CM of elderly (median age, 81.5 years [76.5–85.2]), predominantly male (83%), hypertrophic (mean LV septal wall thickness 16.9±3.5 mm), and symptomatic (81% NYHA class ≥2) patients with heart failure with multiple comorbidities (Table 1). It reflects a contemporary cohort as 96% (779/816) of patients were diagnosed since 2017 onwards.

Table 1.

Baseline Characteristics Overall and Per Left Ventricular GLS Stage

GLS Characteristics and NAC Stages

GLS acquisitions were made by GE Vingmed, Philips, and Siemens cardiac ultrasound machines in 77.7% (634), 11.6% (95), and 10.7% (87), respectively. GLS analysis was performed online in 40% (296 GE Vingmed, 34 Philips) and offline in 60% (286 by GE EchoPAC and 200 by TomTec Software), indicating 71% GE vendor-based evaluations. Overall, the median baseline GLS value was −12.8% (−15.8 to −10.0%) with normal GLS at ATTRwt-CM diagnosis being rare, noted in 12% only (100/816). A significant, although moderate, inverse linear correlation between GLS and LVEF was noted (r=−0.472, P<0.001; Figure 1A). GLS was abnormal in 88% (716) of patients overall. Although 69% presented with heart failure with preserved LVEF ≥50% phenotype, only 17% (93/549) within this subgroup demonstrated a normal baseline GLS value and a wide range of GLS values was noted, from −28.0% to −2.5%. Remarkably, 31% presented with LVEF <50%, of which only 3% presented with a normal GLS value. Median GLS value showed less impaired longitudinal function in patients with LVEF ≥50% versus <50% (−14.1% versus −10.0%; P<0.001, respectively). At baseline, patients presented with NAC disease stages I, II, and III in 57%, 28%, and 15%, respectively. Although higher NAC stages corresponded to more impaired GLS (median GLS in NAC I −14.3%, NAC II −11.6%, and NAC III −11.4%; P<0.001), significant GLS overlap between NAC stages existed (Figure 1B).

Figure 1.

Baseline left ventricular global longitudinal strain (GLS) distribution. A, Correlation with left ventricular ejection fraction (LVEF). B, Distribution between National Amyloidosis Center (NAC) disease stages. Orange bars indicate median GLS value; see text for details.

The overall study population was stratified into GLS stages based on GLS distribution quartiles: stage 1 (<−15.8%), 2 (−15.8 to −12.9%), 3 (−12.8 to −10.0%), and 4 (>−10.0%), with higher stages (quartiles) reflecting progressively more impaired LV longitudinal function and more advanced heart failure phenotypes, despite similar cardiovascular risk factors (Table 1). Clinically, progressive higher GLS stages represented elder subjects with more cardiac comorbidities (including atrial fibrillation, pacemaker, more tricuspid regurgitation, and left-sided valvular heart disease) and more advanced cardiac disease stage (NAC), translating into higher NYHA class symptoms, higher cardiac biomarkers, more renal function impairment, and more prior heart failure hospitalizations. On echocardiography, progressively higher GLS stages revealed more advanced structural and functional cardiac remodeling, including progressively higher LV wall thickness and left atrial diameter, worse diastolic function parameters, and more impaired LV and right ventricular function.

GLS to Predict Mortality Overall

Tafamidis treatment was initiated in 591 of 816 (72%) of subjects, at a median of 67 (19–260) days after cardiac diagnosis. More specific, tafamidis was initiated in 76%, 80%, 71% and 63% of patients in baseline GLS stages 1, 2, 3, and 4, respectively (overall difference between stages P=0.001), albeit time from diagnosis to initiation was similar between GLS stages (P=0.208). During a median follow-up of 2.2 years (1.0–3.6), 29.7% (242/816) of patients died, translating into a median survival of 5.2 years (95% CI, 4.8–5.5 years).

Significant mortality differences were noted between every GLS stage (all P<0.050), with median survival not met, 6.7 (9.1–3.6) years, 4.6 (6.6–2.5) years, and 3.4 (6.1–1.4) years for GLS stage 1, 2, 3, and 4, respectively (Figure 2A). Cumulative 1-year mortality for GLS stage 1, 2, 3, and 4 was 2%, 2%, 8%, and 17%, respectively, with 5-year mortality of 22%, 36%, 52% and 66%, respectively (P<0.001). The median −12.8% GLS cutoff predicted mortality with a sensitivity and specificity of 84% and 52%, respectively, with an area under the receiver operating curve of 0.73 (95% CI, 0.66–0.80; P<0.001; Figure 3A). Median survival for patients with baseline GLS <−12.8% versus ≥−12.8% was 9.1 years (95% CI, 5.8–12.4) compared with 3.9 years (95% CI, 3.3–4.4), respectively (P<0.001; Figure 3B). GLS was a strong univariate predictor of all-cause mortality (HR, 1.14 [95% CI, 1.10–1.18]; P<0.001). Multivariable analysis (model 1), evaluating baseline echocardiographic parameters including LV wall thickness, LVEF, LV stroke volume, diastolic E/e′, left atrial diameter, right ventricular function, systolic arterial pulmonic pressure, and significant left-sided valvular heart disease, identified GLS as the single echocardiographic predictor for all-cause mortality (HR, 1.08 [95% CI, 1.01–1.15]; P=0.031), with highest Wald score (60) of all echocardiographic parameters (Table 2). Subsequent multivariable analysis (model 2), indicated that GLS remained strongly predictive of mortality, independent of other validated prognosticators in ATTRwt-CM, including age, symptomatic status (NYHA class), NAC disease stage, mineralocorticoid receptor antagonist, and sodium-glucose cotransporter-2 inhibitor use and, importantly, tafamidis therapy initiation (HR, 1.08 [95% CI, 1.04–1.12]; P<0.001; Table 2). Similar results were obtained for noncontinuous GLS stages at univariable (Wald score 65, HR, 1.67 [95% CI, 1.47–1.89]; P<0.001) and multivariable analysis (Wald score 23, HR, 1.45 [95% CI, 1.25–1.68]; P<0.001). Compared with GLS stage 1, mortality risk increased 2-fold (HR, 1.8 [95% CI, 1.12–3.1]; P=0.016), 3-fold (HR, 3.3 [95% CI, 2.1–5.2]; P<0.001) and 5-fold (HR, 4.9 [95% CI, 3.1–7.6]; P<0.001) for GLS stages 2, 3, and 4, respectively.

Figure 2.

Cumulative survival according to baseline left ventricular global longitudinal strain (GLS) stages. A, In overall group, since cardiac diagnosis. B, In the tafamidis-treated group, since treatment. ATTRwt-CM indicates wild-type transthyretin amyloid cardiomyopathy.

Figure 3.

Association of left ventricular global longitudinal strain (GLS) with mortality. A, Receiver operating curve (ROC) with area under the curve (AUC) to predict 1-year mortality. B, Cumulative survival, stratified by GLS median value. C, Likelihood ratio test. ATTRwt-CM indicates wild-type transthyretin amyloid cardiomyopathy; and NAC, National Amyloidosis Center.

Table 2.

Univariable and Multivariable Cox Regression Analysis for Mortality Prediction Since Cardiac Diagnosis (n=816)

GLS to Predict Mortality in Tafamidis-Treated Patients

At diagnosis, patients who were subsequently initiated on tafamidis versus noninitiated subjects were on average 4.4 years younger, more likely to be male patients with less advanced heart failure phenotype, despite similar baseline cardiovascular risk factors. Indeed, tafamidis-initiated patients were less symptomatic (lower NYHA class), with lower NAC disease stage coinciding with less pronounced structural and functional remodeling, related to a 50% lower prior heart failure hospitalization history (P<0.001; Table S3). Nevertheless, baseline heart failure drug and device treatment were similar, apart from higher sodium-glucose cotransporter-2 inhibitor, sodium-glucose cotransporter-2 inhibitor use in tafamidis-initiated subjects (19.8% versus 9.0%, P<0.001, respectively). Median baseline GLS was −13.0% (−15.9% to −10.4%) in tafamidis-initiated versus −11.5% (−15.0% to −9.0%) in noninitiated subjects (P=0.001).

Median follow-up since tafamidis treatment in the 591 initiated subjects was 1.9 years (0.9–2.9), during which 19% (112/591) died. Patients with baseline GLS <−12.8% versus ≥−12.8% had lower cumulative 1-year mortality since treatment of 3% versus 14% and 5-year mortality of 26% versus 67%, respectively (P<0.001). Moreover, higher GLS stages at diagnosis related to higher mortality since treatment, with median survival not met for GLS stage 1 and 2, and 4.5 (95% CI, 3.2–5.8) and 3.3 (95% CI, 2.5–4.2) years for GLS stage 3 and 4, respectively (P<0.001; Figure 2B). One-year and 3-year cumulative mortality since tafamidis treatment per GLS stage were 1% and 7% (stage 1), 5% and 14% (stage 2), 8% and 36% (stage 3), and 20% and 47% (stage 4), respectively (P<0.001). In this patient cohort restricted to tafamidis-initiated subjects, baseline GLS was confirmed to be a strong all-cause mortality predictor (HR, 1.13 [95% CI, 1.06–1.20]; P<0.001), independent of other predictors, including age, NYHA class symptoms, and NAC disease stage at start of tafamidis treatment (Table S4). A significant interaction of GLS with tafamidis treatment effect was demonstrated (P=0.014), indicating slightly higher mortality benefit under tafamidis treatment when initiated at lower baseline GLS (better LV longitudinal function; Figure S2). Of note, in the 225 subjects that were not tafamidis initiated, GLS was also predictive for mortality (HR, 1.28 [95% CI, 1.09–1.50]; P=0.003).

GLS Versus NAC Stages to Predict Mortality

Overall, median survival for NAC stage I, II and III patients was 6.6, 4.6, and 2.6 years, respectively (P<0.001; Table 3). Progressively higher NAC disease stage was noted within increasing GLS stages (P<0.001), with NAC stage ≥II in 24% (GLS stage 1), 32% (GLS stage 2), 50% (GLS stage 3), and 71% (GLS stage 4). Although baseline GLS and NAC disease stage both predicted mortality, GLS remained significantly predictive, independent of both NAC stage at baseline and at tafamidis initiation (Table 2; Table S4). Moreover, the median baseline GLS −12.8% cutoff value provided further prognostic discriminative ability for all-cause mortality within each NAC disease stage stratum (all P<0.050; Table 3; Figure 4). Finally, the likelihood ratio test indicated GLS (both as a continuous variable and as GLS stages) provides significant incremental prognostic value over baseline NAC disease stage and age (both P<0.001; Figure 3C).

Table 3.

Median Survival Since Diagnosis by NAC Disease Stage, Stratified by Median −12.8% GLS Cutoff

Figure 4.

Cumulative survival per National Amyloidosis Center (NAC) biomarker stage, stratified by left ventricular global longitudinal strain (GLS) cutoff. ATTRwt-CM indicates wild-type transthyretin amyloid cardiomyopathy.

Discussion

The principal findings of the current report, the largest multicentric study evaluating baseline echocardiographic GLS in patients with ATTRwt-CM to date, are as follows (1) GLS is abnormal in 88% at cardiac diagnosis, irrespective of LVEF, (2) GLS (quartile staging) is a strong mortality predictor, including in tafamidis-treated subjects, (3) the prognostic value of GLS is independent of tafamidis initiation and baseline NAC stage, and (4) GLS provides incremental discriminative prognostic value within each NAC stage. With GLS being a strong and the sole echocardiographic predictor of all-cause mortality, this is the first study, to the best of our knowledge, to lend support for its role as an imaging-based prognostic staging system in ATTRwt-CM, including under disease-modifying treatment, independent of and incremental to serum biomarker staging.

GLS as Systolic Function Marker

Previous data in cardiac amyloidosis patients demonstrated an inverse linear correlation between LVEF and GLS (r=−055 to −0.66; P<0.001), in line with our results (r=−0.472; P<0.001) in 816 patients with ATTRwt-CM.^10,16 Nevertheless, in ATTR-CM, GLS assessed by echocardiography has consistently been reported to be a more sensitive marker of systolic function because LVEF significantly underestimates the degree of systolic functional impairment present in most patients, confirmed by our data.^9,14 Indeed, 69% of our study cohort presented with preserved LVEF ≥50% (heart failure with preserved ejection fraction phenotype), with 83% within this subgroup displaying abnormal baseline GLS >−18.0%, covering a variety of mild to severely reduced longitudinal function. Abnormal GLS values were noted in up to 88% of the overall study population, throughout the LVEF spectrum. A recent elegant study in cardiac amyloidosis subjects showed that LVEF decline occurs only at later disease stages with higher amyloid burden, whereas GLS impairment was among the first markers from early disease onset onwards.⁹ This phenomenon arises from myocardial TTR amyloid fibril infiltration affecting initially subendocardial and basal layers to progressively deposit throughout the entire myocardium, causing increasing pseudohypertrophy by extracellular space expansion. Early longitudinal dysfunction is initially compensated for by increased radial and circumferential myocardial function to maintain normal LVEF. Only at later disease stages with high amyloid burden, radial function declines, translating into LVEF impairment.^8,14 Intriguingly, few patients displayed a slight LVEF decline despite preserved GLS value. We speculate this finding might reflect early ATTRwt-CM stages with low amyloid burden and nonamyloid reasons that affect LVEF, such as altered loading conditions or small LV cavity volumes, where small absolute stroke volume changes markedly alter LVEF as end-diastolic volume (the denominator) is small, but less affects GLS, although LVEF technical measurement variability cannot be excluded. Brief, previous and current data support GLS to be the preferred systolic function imaging metric in patients with ATTRwt-CM to evaluate prognosis, able to capture the variety and complexity of baseline systolic function impairment, present in virtually all affected patients.^10,11 Moreover, echocardiographic GLS assessment is nowadays an easily available, robust, standardized, (semi)automated and reproducible technique with low reported interobserver and intraobserver variability (both <0.2%) and intraclass correlations of 0.96 to 0.99 and 0.96 to 0.97, respectively, in cardiac amyloidosis patients.^{8,10,14,15,17}

GLS as Mortality Predictor in Contemporary ATTRwt-CM

Contrary to light-chain cardiac amyloidosis, data on the prognostic role of GLS in ATTRwt-CM, specifically assessed by echocardiography, are scarce.¹⁵ In a study of 172 patients, Quarta et al¹⁰ showed that GLS was a mortality predictor, independent of NYHA class and estimated glomerular filtration rate (HR, 1.1 [95% CI, 1.01–1.19]; P=0.026). This report, however, applied GLS by averaging 4- and 2-chamber strain and included 56 untreated ATTRwt-CM subjects only. A larger study including 766 untreated patients with ATTRwt-CM demonstrated that in wild-type forms, LV longitudinal strain was the strongest echocardiographic mortality predictor, overall independent of NAC disease staging and NYHA class (HR, 1.05 [95% CI, 1.01–1.09]; P=0.006).¹¹ This report was a single-center observation that applied only 4-chamber strain analysis and provided final prognostication for combined wild-type and variant-type ATTR-CM, not correcting for baseline age. Our multicentric data in a large group of 816 ATTRwt-CM subjects, applying GLS assessment averaging 4-, 3- and 2-chamber strain as recommended, confirmed GLS (including quartile-derived stages) to be the sole echocardiographic and strong independent predictor of all-cause mortality, independent of established predictors, including age, NAC biomarker stage, and NYHA class (HR, 1.08 [95% 1.04–1.12]; P<0.001), an absolute impairment of GLS of 1% translated into an 8% increased death risk.

In addition, the current report is the first indicating GLS remains highly predictive for mortality in a contemporary context of tafamidis-initiated subjects, showing independent prognostic ability after correcting for disease-modifying treatment initiation, also when restricting to the 591 (72%) tafamidis-treated patients (HR, 1.15 [95% CI, 1.08–1.22]; P<0.001). That GLS remains highly prognostic under tafamidis treatment is a relevant and novel finding, as tafamidis treatment has shown, both in randomized controlled as in a real-world setting, to significantly attenuate longitudinal impairment, assessed by GLS.^18,19 Although prone to selection bias, overall mortality within the 4 GLS quartiles under tafamidis treatment was lower compared with the overall group, including untreated patients (Figure 2). In addition, our data suggest a significant tafamidis treatment interaction with baseline GLS (P=0.014) favoring early treatment at less impaired GLS, contrary to a recent post hoc analysis from the landmark ATTRACT-trial (Tafamidis in Transthyretin Amyloidosis Cardiomyopathy Clinical Trial) indicating no such interaction (P=0.66), but real-world treatment selection bias cannot be excluded.¹⁹

Imaging Versus Biomarker Prognostic Staging

NAC biomarker staging, based on serum NT-proBNP and estimated glomerular filtration rate analysis, has become the mainstay to assess prognosis in patients with ATTRwt-CM.^4,7 Interestingly, 2 large series, including the current report implementing tafamidis-treated subjects, indicate GLS remains strongly predictive for mortality, even when correcting for baseline NAC stage.¹¹ We demonstrated that GLS assessment may refine the current NAC biomarker staging system, providing additional discriminative ability within each NAC stage stratum, applying a −12.8% median GLS cutoff value (Table 3; Figure 4). GLS, therefore, could be incorporated as an adjunct to NAC biomarker staging. Similarly, the addition of GLS to conventional and validated prognostic algorithms has recently been advocated also for light-chain cardiac amyloidosis.^15,17 Furthermore, we indicated that GLS (staging) provides incremental predictive value over NAC biomarker staging (likelihood ratio test P<0.001). Indeed, GLS staging may be an easy-to-obtain alternative metric to biomarker staging, potentially outperforming biomarker-based prognostication. An approximate absolute GLS value of >16%, 13% to 16%, 10% to 13%, and <10%, respectively, reflects low (stage 1), moderate (stage 2), moderate-to-high (stage 3), and high (stage 4) future risk for mortality, irrespective of disease-modifying tafamidis treatment. These GLS stages may form the basis of a formal and novel imaging-based prognostication system. Such imaging-based prognostication system may additionally prove to be helpful for future trial design and patient selection or guide overall patient management.

Although GLS and NAC stage are both associated with disease severity, our results indicate that GLS captures additional prognostic information beyond that provided by NAC stage and age, underscoring its integrative value in risk stratification of ATTRwt-CM. The rationale for GLS use as an imaging-based and potentially superior prognostication system may be explained by several reasons. The cardiac amyloid burden has been identified as the main driver of ATTRwt-CM outcome, and might be better reflected by GLS than NT-proBNP, which rather represents a final pathway of several mechanisms that integrate heart failure pathophysiology (fluid status, renal dysfunction, neurohormonal activation, myocardial wall stress, and function).⁸ Major determinants of GLS, reflecting intrinsic myocardial contractility, however, include tissue characteristics, geometry and loading conditions, which are all affected by cardiac disease severity in cardiac amyloidosis subjects.¹⁴ Indeed, longitudinal strain has been directly and linearly correlated with histological cardiac amyloid load in several studies (r=0.45–0.72, all P<0.050).¹⁶ Further tissue characteristics found in patients with ATTRwt-CM include myocardial inflammation, mainly macrophages, in ≈30% of patients and, at later stages, fibrosis and collagen deposition, both related to poor survival, which may all affect GLS.^20–22 In addition, a prognostic linear correlation of GLS with ventricular wall thickness in cardiac amyloidosis exists.¹⁰ Finally, significant left-sided valvular disease, including aortic stenosis or worsening of mitral regurgitation are mortality predictors in ATTR-CM and may further impair GLS.^11,23 Brief, GLS is a systolic function marker that integrates a wide variety of factors that have been shown to be (independent) mortality predictors in ATTRwt-CM.

Echocardiography is the first-line imaging technique for the diagnosis and follow-up of patients with ATTRwt-CM. Nowadays, semiautomated GLS assessment ensures reproducible, nontime-consuming systolic function assessment, feasible in the vast majority of patients, without incremental cost, which should become standard practice during routine echocardiography. This versatility at minimal extra time and cost expense positions GLS assessment next to other prognosticators, including serum biomarkers, 6-minute walking test and quality of life scores, adding granularity by providing insights into intrinsic myocardial damage, rather than merely reflecting symptoms or hemodynamic impact. Moreover, GLS provides incremental discriminative ability within the established NAC biomarker staging, underscoring its complementary role. This is clinically relevant, as 2 patients with similar NAC stage could have a significantly different disease trajectory, with worse prognosis for the patient with more impaired GLS, affecting further decision-making or mandating closer follow-up.

Limitations

Several limitations apply to the current report. First, subjects with baseline GLS available were selected from a larger patient cohort, although no differences in diagnostic period, tafamidis initiation nor LVEF were noted. Moreover, equal outcome was noted, irrespective of baseline GLS availability, which was not predictive of mortality. Therefore, this selection is unlikely to have introduced relevant outcome bias. Second, various vendor platforms were used, with up to 71% of GLS analyses being performed on GE Vingmed-based software, so findings may not be extrapolated to other vendors or platforms for GLS analysis. Nevertheless, confirming the prognostic value of GLS using multiple vendors in a multicentric setting is an important strength of the current study. Moreover, recent evidence showed that vendors providing software tracking solutions for clinical use, result in similar GLS values.²⁴ Third, the feasibility of echocardiographic GLS assessment versus other echocardiographic variables or serum biomarkers may be lower, due to potential technical issues, including body habitus, arrhythmia presence or poor image or speckle-tracking quality. Fourth, current findings apply to the European, mostly White population, and therefore cannot be extrapolated to patients with hereditary ATTR-CM, non-White populations or other ethnicities.

Conclusions

LV GLS is a powerful predictor of mortality in patients with ATTRwt-CM, including in tafamidis-treated subjects. Its prognostic value is independent of tafamidis TTR stabilizer therapy and biomarker NAC staging. GLS assessment represents an adjunct to the current NAC biomarker staging system, which it refines by further increasing discriminative ability within each NAC stage stratum. Moreover, GLS assessment (including stages) provides incremental predictive value over NAC biomarker staging and may form the basis of a complementary, formal and novel imaging-based prognostication system in ATTRwt-CM.

ARTICLE INFORMATION

Acknowledgements

Special thanks to Noa Kozmine, Katrien Derycker, Dixiana Mora Lobo, Sabine Creyf, and Steffie Vanslambrouck for their valuable administrative assistance and dedicated patient care.

Sources of Funding

None.

Disclosures

Drs Debonnaire, Dujardin, Donal, Verheyen, Dupont, Pouleur, Droogmans, Jurcut, Bondue, Zach, Ajmone Marsan, and Trenson received consultancy or speaking fees from Pfizer on the topic of cardiac amyloidosis. The other authors report no conflicts.

Supplemental Material

Tables S1–S4

Figures S1–S2

Supplementary Material

hci-19-e018862-s001.pdf ^{(305.6KB, pdf)}