Modeling of Survival and Frequency of Cardiovascular-Related Hospitalization in Patients with Transthyretin Amyloid Cardiomyopathy Treated with Tafamidis

Abstract

Introduction: ATTR-ACT (Tafamidis in Transthyretin Cardiomyopathy Clinical Trial) demonstrated the efficacy and safety of tafamidis in transthyretin amyloid cardiomyopathy (ATTR-CM). Model-based analyses from ATTR-ACT can examine predictor effects on dose-response/exposure-response relationships.

Methods: Parametric hazard distributions were developed for all-cause mortality and frequency of cardiovascular-related hospitalization. Time-to-event models were fitted to survival data, and repeated time-to-event models were fitted to hospitalization data. Disease-specific characteristics were assessed as baseline predictors of event hazards.

Results: There were 441 patients in this analysis. At month 30, 70.5% (tafamidis) and 57.1% (placebo) of patients were alive, with 154/441 deaths reported; 495 cardiovascular-related hospitalizations occurred. The cumulative risk of death was 42.1% (95% confidence interval [CI] 24.2-58.0) lower with tafamidis than with placebo, regardless of New York Heart Association (NYHA) class; significant predictors of decreased risk were genotype (wild-type), greater 6-Minute Walk Test (6MWT) distance, higher left ventricular ejection fraction (LVEF), and lower blood urea nitrogen (BUN) and N-terminal pro-B-type natriuretic peptide concentrations. The average cumulative risk of cardiovascular-related hospitalization up to 30 months was 40.8% (95% CI 31.0-49.7) lower with tafamidis in NYHA class I/II patients. Significant predictors of reduced risk were greater 6MWT distance, higher LVEF, and lower BUN and troponin I concentrations.

Conclusions: Tafamidis reduced cumulative mortality and hospitalization risk versus placebo in patients with ATTR-CM. Baseline predictors of outcome were consistent with the cardiovascular nature of the disease and suggested that earlier treatment may improve outcomes. CLINICAL TRIALS.

Gov identifier: NCT01994889 (date of registration: November 26, 2013).

Conflict of interest statement

S. Riley, M. Boucher, and L.O. Harnisch are full-time employees of, and hold stock and/or stock options with, Pfizer. C. Vong was a full-time employee of Pfizer at the time of the analysis; she is now a full-time employee of Novartis AG and holds stock options with Novartis AG.

© 2021. The Author(s).

Figures

Fig. 1

Kaplan–Meier visual predictive check for the base time-to-event Gompertz survival model in a placebo, b tafamidis 20-mg, and c tafamidis 80-mg cohorts, comparing the observed data (line) to the 90% prediction interval of the simulated data (green area) with the time-to-event model

Fig. 2

Final model effects of identified covariates and the baseline hazards for a TTE (survival) and b RTTE (cardiovascular-related hospitalization). Figure shows the percentage change in risk of death with values for each covariate in the 25th percentile or 75th percentile, compared with the risk at their median values. Baseline incidence/ranges for covariates (total population): genotype, 24.0% ATTRv; 6MWT, 24–822 m; BUN, 8.7–129.0 mg/dL; LVEF, 11.0–85.4%; NT-proBNP, 35.2–2598.0 pmol/L; troponin I, 0.03–12.22 ng/mL. 6MWT 6-Minute Walk Test, ATTRv variant transthyretin amyloidosis, ATTRwt wild-type transthyretin amyloidosis, BUN blood urea nitrogen, LVEF left ventricular ejection fraction, NT-proBNP N-terminal pro-B-type natriuretic peptide, NYHA New York Heart Association, RTTE repeated time-to-event, TTE time-to-event

Fig. 3

Cumulative survival probability at month 30 given baseline characteristics from the final time-to-event model. The blue line represents the overall cumulative survival probability assuming all patients received placebo. Tertile distributions are represented for each covariate, with their respective range. 6MWT 6-Minute Walk Test, BUN blood urea nitrogen, COVBASE composite set of baseline covariates, LVEF left ventricular ejection fraction, NT-proBNP N-terminal pro-B-type natriuretic peptide

Fig. 4

Cumulative risk reduction over 30 months for each hospitalization event to occur with tafamidis. The dashed horizontal intercept is the overall reduction in the weighted average cumulative risk in New York Heart Association class I/II patients over 30 months treated with tafamidis. An individual hypothetical risk reduction was calculated based on 1000 simulated datasets using the model, treatment effect, and the design matrix of the trial with all patients on placebo then subsequently on tafamidis treatment. Since every simulation gives a different number of hospitalizations per patient, each trial simulation was summarized according to the frequency of events, and a weighted mean in the reduction was calculated based on the distribution of cumulative events. Each point on the graph corresponds to the risk reduction for the specific number of events, whereas the horizontal line takes the weighed mean approach