Clinical data used for the present investigation were extracted from the database of the observational, prospective, multicenter AMIGAL registry, which includes consecutive patients with amyloid cardiomyopathy recruited in 7 hospitals of the public health care system of the Autonomous Community of Galicia (figure 1). Patient enrolment started on January 1, 2018 and is still ongoing. For the present analysis, we selected patients who had been included in the registry from the beginning until October 31, 2022.

Figure 1.

Central illustration. Summary of the most relevant characteristics and results of the study. ATTR, transthyretin amyloidosis; SPAP, systolic pulmonary arterial pressure; TAPSE, tricuspid annular plane systolic excursion.

(0.25MB).

According to the study protocol, described elsewhere,16 diagnosis of AL cardiac amyloidosis is established by the detection of light-chain deposits in endomyocardial biopsy specimens alone or in a noncardiac biopsy, together with highly suggestive cardiac imaging findings. ATTR cardiac amyloidosis is defined by a grade ≥ 2 cardiac uptake on 3,3-diphosphono-1,2-propanodicarboxylic acid scintigraphy in patients with no evidence of monoclonal gammopathy in serum and urine immunofixation and serum free light-chain assay. Histological confirmation of transthyretin deposits is required only in the presence of monoclonal gammopathy. A genetic test to reveal pathogenic variants of the transthyretin gene is recommended in patients with ATTR cardiac amyloidosis. The diagnosis of other types of cardiac amyloidosis different to AL and ATTR, eg, Apo-A IV cardiac amyloidosis, requires the demonstration of an abnormal deposition of the causal protein in endomyocardial biopsy specimens by means of mass spectrophotometry.

The study protocol was approved by the Ethics Committee for Clinical Research of Galicia, and written informed consent was obtained from all participants before enrollment. The project conforms to the principles outlined in the Declaration of Helsinki.

Transthoracic echocardiography was performed by board-certified cardiologists during routine clinical follow-up and according to practice guidelines.17 Data on echocardiographic parameters were collected from echocardiographic reports available in the clinical history of each patient.

Systolic pulmonary artery pressure was calculated as the sum of maximum systolic transtricuspid gradient, measured by means of continuous flow Doppler in the apical 4-chamber view, plus central venous pressure, estimated on the basis of the diameter and respiratory variation of the inferior vena cava in the subxiphoid view.

Specific guidelines for the echocardiographic estimation of central venous pressure17 recommend assigning a value of 3mmHg in cases where the diameter of the inferior vena cava is ≤ 21mm and collapses> 50% with a sniff and assigning a value of 15mmHg in cases where the diameter of the inferior vena cava is> 21mm and collapses <50% with a sniff. In intermediate cases that do not fit this paradigm, the guidelines recommend assigning a value of 8mmHg.

Tricuspid annulus plane systolic excursion was measured by means of the M-mode in the apical 4-chamber view.

Patients included in the registry who had missing data for calculating the TAPSE/SPAP ratio were excluded from the present analysis.

Patients were followed up from the day of their inclusion in the registry until November 30, 2022, or until the date of death or heart transplant, whichever occurred first. Major clinical outcomes evaluated in this study were overall survival and survival free of HF hospitalization. In the assessment of the combined endpoint, hospitalization for heart transplant was considered as an equivalent of HF hospitalization.

Causes of death were collected from death certificates or, when available, from autopsy reports. For statistical analysis, unknown causes of death were considered as being of a cardiovascular origin.

For statistical comparisons, patients were categorized into 3 groups according to tertiles of the baseline TAPSE/SPAP ratio. Categorical variables are reported as absolute numbers and percentages, while continuous variables are expressed as means±standard deviation or as medians [interquartile range], as appropriate. The Kolmogorov-Smirnov test was used to evaluate the conformity of continuous variables to normal distribution.

Comparison of baseline characteristics, echocardiographic parameters, and laboratory data across TAPSE/SPAP tertiles was performed by using the chi-square test for linear trends for categorical variables and by using analysis of variance for linear trends in the case of continuous ones.

Cumulative survival and cumulative survival free of HF hospitalization are summarized by Kaplan-Meier survival curves and were compared across tertiles of TAPSE/SPAP ratio by means of the log-rank test.

We used multivariable Cox proportional-hazards regression analysis to adjust the effect of confounding bias on the statistical associations observed between baseline TAPSE/SPAP ratio and the risk of all-cause mortality and the combined endpoint of all-cause mortality or HF hospitalization. We rejected the use of backward stepwise models to identify potential confounders to avoid possible loss of information due to limited statistical power. Instead, we designed 5 different Cox regression models in which we entered the clinical, echocardiographic, analytical, treatment or staging variables that we considered as potential confounders, on the basis of previous knowledge and the distribution of this variables across categories of the TAPSE/PSAP ratio. Thus, our primary goal with this analysis was to control potential confounding bias, by calculating estimated hazard ratios adjusted by different combinations of relevant clinical characteristics.

For this purpose, we designed 5 multivariable models that included different combinations of adjusting baseline covariables that were distributed asymmetrically across tertiles of the TAPSE/SPAP ratio and, therefore, were considered as potential confounders. Model 1 (clinical co-variables) included age, gender, type of cardiac amyloidosis (ATTR vs other types), atrial fibrillation, New York Heart Association (NYHA) class, and physical signs of congestion. Model 2 (laboratory covariables) included NT-proBNP, glomerular filtration rate, uric acid, urea, gamma-glutamyl transpherase, alkaline phosphatase, and potassium. Model 3 (echocardiographic variables) included maximum left ventricular wall thickness, left ventricular ejection fraction, moderate or severe mitral regurgitation, and moderate or severe tricuspid regurgitation. Model 4 (treatment variables) included the use of loop diuretics, beta-blockers, mineralocorticoid receptor antagonists, and anticoagulants. Model 5 (disease stages) included cardiac amyloidosis stages, as defined by means of the United Kingdom (UK) National Amyloid Centre (NAC) staging system,15 a validated prognostic model based on serum NT-proBNP concentrations and glomerular filtration rate.

We calculated Harrell's c-statistics to evaluate the predictive ability of the UK NAC staging system15 in our sample, as well as the incremental predictive capacity of adding the TAPSE/SPAP ratio to this biomarker-based model. In addition, we evaluated the association between the TAPSE/SPAP ratio and study outcomes stratified by UK NAC stages. Comparisons of Harrell's c-statistics between 2 different Cox regression models were performed by means of the Comparec-statistical package for R.

A 2-tailed P value <.05 was used to indicate statistical significance. Analyses were performed with SPSS 25 (IBM, United States), STATA 14 (Statacorp, United States) and R 4.3.2 (R Foundation, Austria).

A total of 315 patients with cardiac amyloidosis were included in the registry from January 1, 2018 to October 31, 2022. Among them, 233 (74.0%) had available baseline echocardiographic data for calculating the TAPSE/SPAP ratio, and constituted the population assessed in the present analysis. In 72 out of the 82 (87.8%) patients excluded from the study, lack of data were due to the presence of a trace or no tricuspid regurgitation on baseline echocardiography, and consequently the transtricuspid gradient could not be calculated to estimate SPAP.

Of the final study cohort, 209 (89.7%) patients had ATTR amyloid cardiomyopathy, 23 (9.9%) had AL amyloid cardiomyopathy and 1 (0.4%) had Apo-A IV amyloid cardiomyopathy. Among ATTR cases, 176 (84.2%) were confirmed wild-type, 6 (2.9%) were confirmed variant cases, and 27 (12.9%) were indeterminate, as these patients were not genotyped.

Figure 2 shows the distribution of the baseline TAPSE/SPAP ratio in the study population. The median value of this parameter was 0.48mm/mmHg, and the interquartile range was 0.33-0.65mm/mmHg. The first (lowest), second (intermediate) and third (highest) tertiles of the TAPSE/SPAP ratio were <0.38mm/mmHg, 0.38 to 0.58mm/mmHg and> 0.58mm/mmHg, respectively.

Figure 2.

Distribution of baseline values of the TAPSE/SPAP ratio in the study population. SPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annulus plane systolic excursion.

(0.14MB).

The patients’ baseline clinical, laboratory and echocardiographic characteristics are shown in table 1 and classified according to tertiles of the TAPSE/SPAP ratio.

There were no significant differences across tertiles of the TAPSE/SPAP ratio in demographic characteristics. However, a lower TAPSE/SPAP ratio was associated with an increased prevalence of atrial fibrillation, NYHA class III or IV, and physical signs of congestion.

The TAPSE/SPAP ratio was inversely associated with serum levels of NT-proBNP, uric acid, alkaline phosphatase, and gamma-glutamyl transpherase; it was also directly associated with glomerular filtration rate. Patients with a TAPSE/PSAP ratio within the highest tertile were more frequently in UK NAC stage I.

On echocardiography, patients with a lower TAPSE/SPAP ratio showed a higher prevalence of moderate or severe tricuspid regurgitation, as well as lower left ventricular ejection fraction.

Prescription of anticoagulants, diuretics and mineralocorticoid antagonists at baseline was significantly more frequent among patients with lower a TAPSE/SPAP ratio, with no differences among study subgroups regarding the prescription of other pharmacological agents.

The proportion of patients who received specific disease modifying therapies such as tafamidis (for ATTR), chemotherapy or autologous stem cell transplantation (for AL) was similar across tertiles of baseline TAPSE/SPAP ratio.

Patients were followed up for a median of 680 [371-1234] days. During this time, 65 (27.9%) patients died and 7 (3.0%) underwent heart transplantation. In addition, 68 (29.2%) patients required at least 1 hospital admission due to HF during follow-up. Cardiovascular causes accounted for 47 out of 65 (72.3%) deaths registered in the study. Table 2 details specific causes of death in the study population, according to tertiles of the baseline TAPSE/SPAP ratio.

Figure 3 show the Kaplan-Meier estimates for the cumulative probability of survival (A) and the cumulative probability of survival free of HF hospitalization (B) in the study population, stratified according to tertiles of baseline TAPSE/SPAP ratio. In both cases, the log-rank test for linear trends showed a statistically significant association between a lower TAPSE/SPAP ratio and an increased risk of the adverse clinical outcome (P <.001 for both comparisons). The median survival of patients with a baseline TAPSE/SPAP ratio within the first (< 0.38mm/mmHg), second (0.38-0.58mm/mmHg) and third (> 0.58mm/mmHg) tertiles were 2.8, 3.5 and 4.6 years, respectively.

Figure 3.

Kaplan-Meier estimates of the overall cumulative probability of survival (A) and the cumulative probability of survival free of heart failure hospitalization (B) in patients with cardiac amyloidosis, stratified by tertiles of baseline TAPSE/SPAP ratio. SPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annulus plane systolic excursion.

(0.24MB).

Table 3 summarizes the results of univariate and multivariate Cox regression models constructed to evaluate the association between the baseline TAPSE/SPAP ratio and study outcomes and, thereafter, to adjust potential confounding bias that might affect it. Individual univariate and multivariate coefficients of all explored covariables are presented in tables 1 and 2 of the supplementary data.

Univariate Cox regression showed a statistically significant association between the baseline TAPSE/SPAP ratio and all-cause mortality (hazard ratio [HR] per 0.1mm/mmHg,0.78; 95% confidence interval [95%CI], 0.68-0.89) and the combined endpoint of all-cause death or HF hospitalization (HR per 0.1mm/mmHg,0.79; 95%CI, 0.72–0.87). Estimated HR for all-cause death were 5.44 (95%CI, 2.55-11.59) for patients who showed a baseline TAPSE/SPAP ratio within the first vs the third tertiles and 3.72 (95%CI, 1.73-7.99) for patients who showed a baseline TAPSE/SPAP ratio within the second vs the third tertiles. Estimated HR for all-cause mortality or HF hospitalization were 3.36 (95%CI, 1.99-5.67) for patients who showed a baseline TAPSE/SPAP ratio within the first vs the third tertiles and 2.22 (95%CI, 1.30-3.77) for patients who showed a baseline TAPSE/SPAP ratio within the second vs the highest tertiles.

The association between baseline TAPSE/SPAP ratio and study outcomes remained as independent and statistically significant in all the 5 adjusted multivariable Cox regression models constructed in this study (table 3). Adjusted HR estimated per 0.1mm/mmHg increase of baseline TAPSE/SPAP ratio ranged from 0.76 to 0.84 for all-cause mortality and from 0.79 to 0.86 for the combined endpoint all-cause death or HF hospitalization. Adjusted HR estimated for patients with baseline TAPSE/SPAP ratio within the first vs the third tertiles ranged from 4.06 to 6.46 for all-cause mortality and from 2.11 to 3.22 for the combined endpoint all-cause death or HF hospitalization. Adjusted HR estimated for patients with baseline TAPSE/SPAP ratio within the second vs the third tertiles varied between 3.08 and 4.26 for all-cause mortality and between 1.57 and 2.19 for the combined endpoint all-cause death or HF hospitalization.

In the study sample, higher UK NAC stages at baseline were associated with significantly lower cumulative survival and significantly lower cumulative survival free of HF hospitalization (figure 1 of the supplementary data). Increased risk of both adverse outcomes was similar for patients in stage III vs stage I (HR for all-cause death or HF hospitalization,3.32; 95%CI, 1.92-5.75; HR for all-cause death,2.90; 95%CI, 1.41-5.99) and for patients in stage II vs stage I (HR for all-cause death or HF hospitalization,3.90; 95%CI, 2.37-6.42; HR for all-cause death,4.14; 95%CI, 2.16-7.91), with no statistically significant differences between patients at stage III vs stage II (HR for all-cause death or HF hospitalization,0.85; 95%CI, 0.54-1.35; HR for all-cause death,0.70; 95%CI, 0.39-1.27).

Multivariate Cox regression model 5 included baseline UK NAC stages together with baseline TAPSE/SPAP ratio (table 3). According to this model, baseline TAPSE/SPAP ratio was independently associated with the combined endpoint of all-cause death or HF hospitalization (HR per 0.1mm/mmHg=0.86; 95%CI, 0.78-0.96) and all-cause mortality (HR per 0.1mm/Hg=0.84; 95%CI, 0.73-0.97). However, if TAPSE was included in the multivariable model instead of the TAPSE/SPAP ratio, the associations were no longer statistically significant, either with all-cause death or HF hospitalization (HR for 1mm increase of TAPSE=0.96; 95%CI, 0.91-1.02) or for all-cause mortality (HR for 1mm increase of TAPSE,0.95; 95%CI, 0.89-1.02).

UK NAC staging system showed moderate predictive ability both for the combined event of all-cause death or HF hospitalization (Harrell's c-statistic=0.668) and for all-cause mortality (Harrell's c-statistic=0.662). The addition of TAPSE/SPAP ratio to the predictive model based on UK NAC stages resulted in a moderate increase of its predictive capacity, both for the combined event of all-cause death or HF hospitalization (Harrell's c-statistic=0.707; P for comparison=.019) and for all-cause mortality (Harrell's c-statistic=0.705; P for comparison=.065). The addition of TAPSE to the predictive model based on UK NAC stages instead of TAPSE/SPAP ratio resulted in a lower increase of predictive capacity (Harrell's c-statistic for all-cause death or HF hospitalization=0.695; P for comparison=.089; Harrell's c-statistic for all-cause death=0.693; P for comparison=.078).

Figure 4 shows the Kaplan-Meier curves of cumulative survival and cumulative survival free of HF hospitalization in patients with baseline TAPSE/SPAP ratio ≤ 0.58 (first and second tertiles) vs> 0.58 (first tertile), categorized according to baseline UK NAC stages. In the subgroup of patients in UK NAC stage I, reduced baseline TAPSE/SPAP ratio (≤ 0.58) was associated with significantly lower survival free of HF hospitalization (P log-rank=.003, figure 4A) and overall survival (P log-rank=.001, figure 4B). However, in the subgroup of patients in UK NAC stages II or III, no statistically significant difference with regard to either adverse outcome were observed according to baseline TAPSE/SPAP ratio (P log-rank for all-cause death or HF hospitalization=.484, figure 4C; P log-rank for all-cause mortality=.150, figure 4D).

Figure 4.

Kaplan-Meier estimates of the cumulative survival free of heart failure hospitalization (A,C) and the cumulative survival (B,D) according to United Kingdom National Amyloidosis Centre stages. SPAP, systolic pulmonary artery pressure; TAPSE, tricuspid annular plane systolic excursion; UK NAC, United Kingdom National Amyloidosis Centre.

(0.33MB).

In the specific subgroup of patients with ATTR cardiac amyloidosis, estimated Harrell's c-statistics of the UK NAC staging system were 0.652 for the combined endpoint of all-cause mortality or HF hospitalization and 0.640 for all-cause mortality and increased respectively to 0.692 (P for comparison=.030) and 0.686 (P for comparison=.063) when baseline TAPSE/SPAP were added to the predictive model.

The major strengths of this study are its multi-institutional nature, as well as the prospective inclusion of a substantial number of patients with close clinical follow-up and the collection of comprehensive clinical information. Its major limitations are the observational nature of the study and the absence of external monitoring, which may have led to potential information, selection, and confounding biases. Furthermore, our analysis is based on data previously collected from a general registry of patients with amyloid cardiomyopathy, which was not specifically designed for the purpose of this study. Some prognostic variables of interest, such as serum troponin, were not available for this analysis. Echocardiographic studies were performed locally during routine clinical follow-up by different operators, which may have led to some heterogeneity in the measurement of functional parameters. Finally, data from cardiac catheterization were not available and consequently we could not validate our findings invasively.