Predicting the Long-Term Effects of CRT on Mortality
Predicting the Long-Term Effects of CRT on Mortality
Objectives: This study was designed to investigate whether selected baseline variables and early response markers predict the effects of cardiac resynchronization therapy (CRT) on long-term mortality.
Background: Cardiac resynchronization therapy reduces long-term morbidity and mortality in patients with moderate or severe heart failure and markers of cardiac dyssynchrony, but not all patients respond to a similar extent.
Methods: In the CARE-HF (Cardiac Resynchronization in Heart Failure) study, 813 patients with heart failure and markers of cardiac dyssynchrony were randomly assigned to receive or not receive CRT in addition to pharmacological treatment and were followed for a median of 37.6 months. A model including assigned treatment, 15 pre-specified baseline variables, and 8 markers of response at 3 months was constructed to predict all-cause mortality.
Results: On multivariable analysis, plasma concentration of amino terminal pro–brain natriuretic peptide (univariate and multivariable model chi-square test: 105.0 and 48.4; both p <0.0001) and severity of mitral regurgitation (chi-square test: 44.0 and 17.9; both p <0.0001) at 3 months, regardless of assigned treatment, were the strongest predictors of mortality. Ischemic heart disease as the cause of ventricular dysfunction (chi-square test: 34.9 and 7.4; p <0.0001 and p = 0.0066), being in New York Heart Association functional class IV (chi-square test: 18.8 and 9.6; p <0.0001 and p = 0.0020), or having less interventricular mechanical delay (chi-square test: 29.8 and 8.8; p <0.0001 and p = 0.0029) at baseline all predicted a worse outcome. However, the reduction in mortality in patients assigned to CRT was similar before (hazard ratio: 0.602; 95% confidence interval: 0.468 to 0.774) and after (hazard ratio: 0.679; 95% confidence interval: 0.494 to 0.914) adjustment for variables measured at baseline and at 3 months.
Conclusions: Patients who have more severe mitral regurgitation or persistently elevated amino terminal pro–brain natriuretic peptide despite treatment for heart failure, including CRT, have a higher mortality. However, patients assigned to CRT had a lower mortality even after adjusting for variables measured before and 3 months after intervention. The effect of CRT on mortality cannot be usefully predicted using such information. (CARE-HF CArdiac Resynchronization in Heart Failure; NCT00170300)
In patients with moderate or severe heart failure due to left ventricular (LV) systolic dysfunction and QRS prolongation on the surface electrocardiogram, cardiac resynchronization therapy (CRT) improves heart function, symptoms, quality of life, and prognosis. The clinical response to CRT is variable, with a continuous spectrum ranging from dramatic to disappointing, and, so far, its prediction from baseline variables has met with limited success. A series of small observational trials has suggested that cardiac dyssynchrony measured by a variety of echocardiographic techniques might predict the response to CRT. For example, Penicka et al. studied 49 patients and concluded that echocardiographic measures of both intra- and interventricular dyssynchrony predicted improvement in ventricular function at 6 months. In a study of 54 patients, Yu et al. reported that tissue Doppler imaging predicted improvement in ventricular function at 6 months. Similarly, Bax et al. studied 80 patients who received CRT using tissue Doppler imaging and reported that the risk of death or hospitalization with heart failure at 1 year was only 6% in the 49 patients with intraventricular dyssynchrony >65 ms but 50% in those with less marked dyssynchrony. This could reflect an intrinsically better prognosis in patients with dyssynchrony or a treatment effect. It is not clear whether the use of short-term improvements in ventricular function as a surrogate measure of long-term clinical response is appropriate, and the few attempts using data from randomized controlled trials have, so far, been unable to corroborate these findings, although the techniques used might now be considered outmoded. Recently, a large prospective, observational study (PROSPECT [Predictors of Response to CRT] trial) was unable to find clinically useful predictors of the response to CRT over 6 months in terms of clinical status or improvement in LV function.
The complex and varied reasons for failure to respond to CRT may be the reason that no robust pre-implantation marker of therapeutic response to CRT has yet been found. Patient factors may include the severity and type of dyssynchrony, the extent and location of myocardial scarring, and the severity of and reasons for mitral regurgitation. It is also likely that dyssynchrony may develop, and perhaps disappear, as ventricular dysfunction progresses. The success of CRT will also depend on the technical success of lead positioning and device programming. An alternative approach to try to assess the role of dyssynchrony in determining the long-term response to CRT is to investigate whether the initial response to therapy predicts long-term outcome, because this simultaneously tests both the patient substrate and the adequacy of implementation of CRT. For example, Yu et al. noted that changes in ventricular function 3 to 6 months after implantation but not changes in symptoms predicted long-term survival in an observational study of 141 patients. Kubanek et al. noted that change in natriuretic peptides at 3 months predicted clinical improvement at 1 year in a study of 43 patients. However, responses to concomitant pharmacological treatment, the existence of comorbid conditions, intercurrent events, and "placebo" effects will be important additional determinants of the apparent clinical response to CRT in observational trials.
An observational study can address the question of whether a patient who receives CRT will do well or not, but cannot tell whether the intervention changed the course of the disease because it lacks a control group. Randomized trials are required to prove whether an intervention alters the natural history of disease. A patient who dies 1 year after receiving CRT might be considered to have had an excellent response if similar patients in the control group die within a few weeks. Similarly, some patients who appear to respond very well to CRT might do just as well without it.
Clearly, CRT reduces long-term mortality. However, if the effects of CRT on mortality can be explained by either baseline measures of dyssynchrony and/or the initial response to therapy, then being randomized to CRT should no longer be associated with reduced mortality after adjustment for these factors. Accordingly, we undertook an analysis to identify whether baseline variables and the early response to CRT could explain its effects on long-term mortality in the CARE-HF (Cardiac Resynchronization in Heart Failure) trial.
Objectives: This study was designed to investigate whether selected baseline variables and early response markers predict the effects of cardiac resynchronization therapy (CRT) on long-term mortality.
Background: Cardiac resynchronization therapy reduces long-term morbidity and mortality in patients with moderate or severe heart failure and markers of cardiac dyssynchrony, but not all patients respond to a similar extent.
Methods: In the CARE-HF (Cardiac Resynchronization in Heart Failure) study, 813 patients with heart failure and markers of cardiac dyssynchrony were randomly assigned to receive or not receive CRT in addition to pharmacological treatment and were followed for a median of 37.6 months. A model including assigned treatment, 15 pre-specified baseline variables, and 8 markers of response at 3 months was constructed to predict all-cause mortality.
Results: On multivariable analysis, plasma concentration of amino terminal pro–brain natriuretic peptide (univariate and multivariable model chi-square test: 105.0 and 48.4; both p <0.0001) and severity of mitral regurgitation (chi-square test: 44.0 and 17.9; both p <0.0001) at 3 months, regardless of assigned treatment, were the strongest predictors of mortality. Ischemic heart disease as the cause of ventricular dysfunction (chi-square test: 34.9 and 7.4; p <0.0001 and p = 0.0066), being in New York Heart Association functional class IV (chi-square test: 18.8 and 9.6; p <0.0001 and p = 0.0020), or having less interventricular mechanical delay (chi-square test: 29.8 and 8.8; p <0.0001 and p = 0.0029) at baseline all predicted a worse outcome. However, the reduction in mortality in patients assigned to CRT was similar before (hazard ratio: 0.602; 95% confidence interval: 0.468 to 0.774) and after (hazard ratio: 0.679; 95% confidence interval: 0.494 to 0.914) adjustment for variables measured at baseline and at 3 months.
Conclusions: Patients who have more severe mitral regurgitation or persistently elevated amino terminal pro–brain natriuretic peptide despite treatment for heart failure, including CRT, have a higher mortality. However, patients assigned to CRT had a lower mortality even after adjusting for variables measured before and 3 months after intervention. The effect of CRT on mortality cannot be usefully predicted using such information. (CARE-HF CArdiac Resynchronization in Heart Failure; NCT00170300)
In patients with moderate or severe heart failure due to left ventricular (LV) systolic dysfunction and QRS prolongation on the surface electrocardiogram, cardiac resynchronization therapy (CRT) improves heart function, symptoms, quality of life, and prognosis. The clinical response to CRT is variable, with a continuous spectrum ranging from dramatic to disappointing, and, so far, its prediction from baseline variables has met with limited success. A series of small observational trials has suggested that cardiac dyssynchrony measured by a variety of echocardiographic techniques might predict the response to CRT. For example, Penicka et al. studied 49 patients and concluded that echocardiographic measures of both intra- and interventricular dyssynchrony predicted improvement in ventricular function at 6 months. In a study of 54 patients, Yu et al. reported that tissue Doppler imaging predicted improvement in ventricular function at 6 months. Similarly, Bax et al. studied 80 patients who received CRT using tissue Doppler imaging and reported that the risk of death or hospitalization with heart failure at 1 year was only 6% in the 49 patients with intraventricular dyssynchrony >65 ms but 50% in those with less marked dyssynchrony. This could reflect an intrinsically better prognosis in patients with dyssynchrony or a treatment effect. It is not clear whether the use of short-term improvements in ventricular function as a surrogate measure of long-term clinical response is appropriate, and the few attempts using data from randomized controlled trials have, so far, been unable to corroborate these findings, although the techniques used might now be considered outmoded. Recently, a large prospective, observational study (PROSPECT [Predictors of Response to CRT] trial) was unable to find clinically useful predictors of the response to CRT over 6 months in terms of clinical status or improvement in LV function.
The complex and varied reasons for failure to respond to CRT may be the reason that no robust pre-implantation marker of therapeutic response to CRT has yet been found. Patient factors may include the severity and type of dyssynchrony, the extent and location of myocardial scarring, and the severity of and reasons for mitral regurgitation. It is also likely that dyssynchrony may develop, and perhaps disappear, as ventricular dysfunction progresses. The success of CRT will also depend on the technical success of lead positioning and device programming. An alternative approach to try to assess the role of dyssynchrony in determining the long-term response to CRT is to investigate whether the initial response to therapy predicts long-term outcome, because this simultaneously tests both the patient substrate and the adequacy of implementation of CRT. For example, Yu et al. noted that changes in ventricular function 3 to 6 months after implantation but not changes in symptoms predicted long-term survival in an observational study of 141 patients. Kubanek et al. noted that change in natriuretic peptides at 3 months predicted clinical improvement at 1 year in a study of 43 patients. However, responses to concomitant pharmacological treatment, the existence of comorbid conditions, intercurrent events, and "placebo" effects will be important additional determinants of the apparent clinical response to CRT in observational trials.
An observational study can address the question of whether a patient who receives CRT will do well or not, but cannot tell whether the intervention changed the course of the disease because it lacks a control group. Randomized trials are required to prove whether an intervention alters the natural history of disease. A patient who dies 1 year after receiving CRT might be considered to have had an excellent response if similar patients in the control group die within a few weeks. Similarly, some patients who appear to respond very well to CRT might do just as well without it.
Clearly, CRT reduces long-term mortality. However, if the effects of CRT on mortality can be explained by either baseline measures of dyssynchrony and/or the initial response to therapy, then being randomized to CRT should no longer be associated with reduced mortality after adjustment for these factors. Accordingly, we undertook an analysis to identify whether baseline variables and the early response to CRT could explain its effects on long-term mortality in the CARE-HF (Cardiac Resynchronization in Heart Failure) trial.
Source...