Myocardial remodelling after withdrawing therapy for heart failure in patients with recovered dilated cardiomyopathy: insights from TRED‐HF

To characterize adverse ventricular remodelling after withdrawing therapy in recovered dilated cardiomyopathy (DCM).


Conclusion
In TRED-HF, withdrawing therapy caused rapid remodelling, with early tissue and functional changes, even amongst patients who did not relapse.

Introduction
Dilated cardiomyopathy (DCM) is characterized by eccentric hypertrophy associated with an increase in myocyte size and extracellular matrix expansion due to interstitial and focal replacement fibrosis. 1,2 Left ventricular (LV) reverse remodelling is characterized by reduction in LV size, regression of hypertrophy and fibrosis and an improvement in systolic function. It may be observed in as many as 40-60% of cases and is associated with resolution of symptoms and an excellent outcome. 3,4 Recent work from our group has demonstrated that many asymptomatic patients with DCM and improved LV function relapse after withdrawing heart failure therapy. 5 This confirms that these patients have remission of heart failure rather than sustained recovery or cure. 5 Amongst these patients, relapse is characterized by LV dilatation and deterioration in systolic function.
Knowledge of the features that accompany early adverse remodelling should lead to improved understanding of disease pathophysiology and may guide the use of treatments that target cellular and interstitial components of the disease. Previous work has demonstrated important sex and genotype differences in remodelling amongst patients with DCM. 6,7 Knowledge of disease characteristics that influence the type and degree of remodelling might enable personalized treatment. 2,8 Cardiovascular magnetic resonance (CMR) enables comprehensive characterization of ventricular remodelling. This includes the assessment of ventricular function and myocardial deformation as well the quantification of LV mass and its cellular and extracellular components, using parametric mapping. 9 In this study, serial CMR assessment was used to characterize changes in myocardial tissue composition and myocardial mechanics after withdrawing therapy, with or without relapse, amongst patients taking part in TRED-HF (Therapy withdrawal in REcovered DCM-Heart Failure). 5

Methods
TRED-HF was an open-label, randomized trial with a follow-on single-arm cross-over phase examining the safety and feasibility of withdrawing treatments for heart failure in patients with recovered DCM. A full description of the methods is provided elsewhere. 5 The trial was registered on ClinicalTrials.gov (NCT02859311).
The study was approved by the National Research Ethics Committee and authorized by the Medicine and Healthcare Products Regulatory Agency. All participants provided written, informed consent. At inclusion, all participants were asymptomatic and had a diagnosis of recovered DCM, with a previous LV ejection fraction (LVEF) <40% that subsequently improved to ≥50%, with normal LV end-diastolic volume, a N-terminal pro-B-type natriuretic peptide (NT-proBNP) level <250 ng/L and who were still taking at least one heart failure therapy [loop diuretic, beta-blocker, angiotensin-converting enzyme (ACE) inhibitor, angiotensin receptor blocker (ARB) or mineralocorticoid receptor antagonist (MRA)]. Patients were randomized 1:1 to phased withdrawal of pharmacological heart failure therapy or to continue therapy, over 6 months. Patients had CMR assessment at baseline, 16 weeks and 6 months.
Therapy was withdrawn in a supervised, step-wise fashion over a maximum of 16 clinic or telephone review. Loop diuretics, if prescribed, were withdrawn first, followed by MRAs, beta-blockers and ACE inhibitors or ARBs. Those randomized to the control arm continued therapy and had follow-up visits at 8 weeks, 16 weeks and 6 months. After 6 months, these patients entered a single-arm cross-over phase and had therapy withdrawn, as described above, between 6-12 months. They were followed up in the same way as the randomized phase of the trial after entering the cross-over phase.
The primary endpoint was a relapse of DCM defined by any one of the following: (i) a reduction in LVEF by >10% and to <50%, or (ii) an increase in LV end-diastolic volume by >10% and to above the normal range, or (iii) a twofold rise in NT-proBNP from baseline and to >400 ng/L, or (iv) clinical evidence of heart failure. Therapy was re-introduced as soon as any of the primary endpoint criteria were fulfilled. The management of patients who did not meet the primary endpoint, but suffered adverse events was determined by the study team and the participant's usual physicians.

Cardiovascular magnetic resonance
Cardiovascular magnetic resonance was performed at baseline, 16 weeks and 6 months, in both the randomized and cross-over phases, using a standardized protocol on a single 3 Tesla scanner (Skyra, Siemens, Erlangen, Germany). Long-and short-axis cine images were acquired using breath-hold steady-state free precession images. Measurement of ventricular volumes and mass was carried out using CMR Tools (Cardiovascular Imaging Solutions, London, UK) using a thresholding technique that includes papillary muscles and trabeculae as part of the LV mass. LV and right ventricular (RV) global longitudinal strain (GLS) were measured from the horizontal long-axis view by a single expert operator (X.C.), who was blinded to trial arm and phase, using feature-tracking software (Medis Suite MR, Medis, Leiden, The Netherlands).
At baseline and 6 months in the randomized and cross-over phases, native and post-contrast T1 maps were acquired at basal and mid-ventricular level in identical short-axis planes, using a breath-hold 5-3-3 modified Look-Locker inversion recovery sequence. Two maps were acquired in each plane. Post-contrast maps were acquired, 15 min after the administration of gadobutrol (0.1 mmol/kg). A single expert operator (V.V.) who was blinded to study arm and phase, measured global myocardial and blood pool T1 on short-axis slices using dedicated software (CVI42, Circle Cardiovascular Imaging, Calgary, Alberta, Canada). Endocardial and epicardial borders were contoured and partial volume artefact from blood was minimized by using a 10% automatic offset from each border. The extracellular volume (ECV) fraction was calculated from the mean myocardial and blood pool T1 values using a published formula. 9 The haematocrit was taken from blood tests performed immediately before each scan. LV mass was calculated from the LV volume and specific gravity of myocardium (1.05 g/mL); LV cell and extracellular matrix mass were derived using the ECV fraction.

Statistical analysis
Characteristics of patients are presented at randomization. Variables are presented as mean/standard deviation (SD), or median/interquartile range if skewed and compared between men and women and carriers and non-carriers of TTNtv using Mann-Whitney U test for continuous data and Fisher's exact test for categorical data. The effect of withdrawing therapy on LV, cell and matrix mass index and LV and RV GLS was examined by comparing these variables between randomized groups using a regression model in which the value at follow-up was the response variable and the treatment indicator and value at baseline were the explanatory variables (i.e. analysis of covariance). It was estimated that a sample size of at least 28 (14 in each group) would have 80% power to detect a 6 g/m 2 increase in LV mass, with the hypothesis that this would be driven by cellular rather than interstitial changes in the early phase, assuming a standard deviation of 6 for interstudy change and an alpha of 0.05.
Since the number of patients was small, we also performed a non-randomized comparison of these values before and after therapy was withdrawn in either the randomized (baseline at 0 months) and cross-over phases (baseline at 6 months). Comparisons were made using paired t-tests.
Differences in the change in these values were also compared amongst men and women and amongst carriers and non-carriers of TTNtv using analysis of covariance.
A P-value of <0.05 was taken as significant throughout. Statistical analyses were performed using Stata version 15.1 (Stat Corp., College Station, TX, USA).

Results
Of the 51 patients randomized, two were excluded as echocardiography was performed in place of CMR due to implanted electronic cardiac devices. One patient withdrew from the study shortly after enrolment. Therefore, data from 48 patients were included ( Figure 1). One patient randomized to the control arm did not cross-over after 6 months, therefore analyses examining patients who had therapy withdrawn in either phase of the study included 47 patients. Baseline and follow-up parametric mapping data were not available, due to the sequence being unavailable, for 13 of 48 patients in the randomized phase and 11 of 47 patients who had therapy withdrawn in either phase of the study.
At enrolment, the mean age of patients was 53 years (SD 12.

Differences in remodelling by sex and genotype
Women had smaller LV mass before therapy was withdrawn compared to men [mean (SD) 53. 2 Table 4).
The effect of sex on change in other variables was also not significant (Table 4).
Similarly, carriers of TTNtv who had therapy withdrawn in either the randomized or cross-over phases of the study, had greater increases in LV matrix mass compared to patients without TTNtv Table 3 Non-randomized comparison of baseline and follow-up variables amongst patients who had therapy withdrawn in the randomized or cross-over phases     (mean effect of TTNtv: 2.6 g/m 2 ; 95% CI 0.4-4.8; P = 0.02) ( Table 4). The effect of genotype on change in other variables was not significant (Table 4).

Discussion
This is the first study to investigate the serial changes in tissue characteristics and cardiac mechanics that accompany early adverse remodelling in patients with DCM. By harnessing advanced CMR techniques including parametric mapping and feature-tracking, we demonstrate that withdrawing pharmacological therapy leads to a rapid reduction in LV and RV GLS and an increase in overall LV mass and LV cell mass. Due to the relatively small number of patients, a non-randomized comparison of baseline and follow-up values amongst all patients who had therapy withdrawn was also performed. This suggested there was also an increase in LV extracellular matrix mass after therapy was withdrawn. The absence of a change in remodelling variables over follow-up amongst patients who continued therapy supports the validity of the findings of the non-randomized analyses. These results are important for several reasons. They emphasize that early adverse remodelling is associated with diminished longitudinal deformation of both the left and right ventricle. This is in keeping with previous studies which have suggested that DCM is a global myocardial process that involves both ventricles. 12,13 The development of changes in RV function within 8 weeks of completing withdrawal of therapy, before the development of symptoms or elevated plasma concentrations of natriuretic peptides, supports the notion of intrinsic RV disease, rather than simply remodelling related to increasing afterload. Equally, we also recognize the absence of evidence supporting a beneficial effect of heart failure therapy on intrinsic RV disease.
Myocardial relapse occurred rapidly amongst patients in the TRED-HF trial. One might have expected that short-term adverse remodelling would be driven by cellular changes such as abnormal calcium handling, energetic dysfunction or sarcomeric dysfunction. Indeed, a marked increase in LV cell mass was observed after withdrawing therapy in both the randomized and non-randomized comparisons, reflecting myocyte hypertrophy, a pathological hallmark of DCM. 1 Non-randomized comparisons, however, also demonstrated an increase in LV extracellular matrix mass between baseline and follow-up after withdrawing therapy. Although this was not borne out in the randomized comparison, possibly due to small patient numbers, this suggests that there might also be rapid extracellular matrix remodelling following therapy withdrawal. Whether this is rapid accumulation of interstitial fibrosis or secondary to interstitial oedema is unclear but deserves further consideration and investigation.
Patients who had therapy withdrawn and did not meet the primary relapse endpoint also had an increase in LV cell mass and matrix mass and a reduction in GLS. This is in keeping with the reduction in LVEF reported amongst this group of patients in the primary analysis 5 and confirms evidence of early adverse remodelling even amongst patients who did not meet the trial criteria for relapse. This supports the concept that a greater Table 4 The effect of sex and genotype on myocardial remodelling amongst patients who had therapy withdrawn in the randomized or cross-over phases Men vs. women (n = 47) a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .   proportion of patients would have relapsed if therapy had been withdrawn for a greater length of time. It also demonstrates the importance of considering adverse remodelling and relapse as being on a continuous spectrum rather than an all-or-nothing binary phenomenon. Previous work has demonstrated important differences between men and women with DCM as well as carriers and non-carriers of TTNtv. 6,7,14  total LV, LV cell and LV matrix mass compared to men. After withdrawing therapy, women had a larger absolute increase in LV mass, although after adjustment for differences at baseline, the effect of sex on LV mass was non-significant. The explanation for this is unclear. It is well established that women are more likely to have reverse remodelling in response to treatment compared to men. 15 It is possible that women have more complete reverse remodelling compared to men and that following therapy withdrawal, a greater deterioration. Further investigation of sex differences in remodelling and the effects of specific therapies are required. Consistent with previous work, 7 carriers of TTNtv tended to have lower LV mass and cell mass index at baseline. 14 Interestingly, they also had greater expansion of extracellular matrix mass during therapy withdrawal. Verdonschot and colleagues previously demonstrated that patients with DCM and TTNtv had greater interstitial fibrosis compared to genotype negative patients with DCM. 7 Our data support the concept that TTNtv may lead to a more fibrotic phenotype. Sarcomeric variants have been associated with up-regulation of genes involved in extracellular matrix expansion in models of hypertrophic cardiomyopathy. 16,17 Other studies have confirmed that interstitial expansion is an early feature of disease. 16,17 Whether patients with TTNtv may be more likely to benefit from targeted anti-fibrotic agents deserves further attention. 2

Limitations
The small number of patients in this sub-study and the incomplete data on parametric mapping data are important limitations and should be borne in mind when interpreting the results. Correction for multiple testing was not performed due to the exploratory nature of the analysis. The analyses investigating differences in remodelling based on sex and genotype should be viewed as hypothesis-generating and require validation in larger studies, considering the small numbers of patients in these sub-analyses. Nevertheless, these results are consistent with previous data and suggest that important differences exist within these sub-groups. It should also be recognized that changes in LV geometry can affect measures of systolic function, including ejection fraction and strain. Previous data have confirmed that GLS is confounded to a lesser degree than ejection fraction by such changes. 18

Conclusions
In TRED-HF, withdrawing therapy for heart failure led to a deterioration in measures of LV and RV systolic function and LV hypertrophy due to an increase in both LV cell and extracellular matrix mass within 6 months. This suggests that early adverse remodelling is a biventricular process with both cellular and interstitial changes. Such changes were observed amongst patients who had therapy withdrawn even if they did not meet the trial criteria for relapse, suggesting that more patients would have relapsed if therapy had been withdrawn for longer. Sex-and genotype-specific differences in remodelling may exist; greater understanding of these may enable more personalized therapy.

Supplementary Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. .