Left Atrial Function Measured by cMRI in Heart Failure
In patients referred with suspected HF, lower LAEF measured by CMRI is associated with higher plasma concentrations of NTproBNP, despite higher doses of diuretics, and with worse LV and RV function. LAEF is also a powerful predictor of incident AF, hospitalization for HF and of mortality in patients with HF whether or not they have LVSD. In contrast, measures of either LV or RV structure or function by MRI were only weak predictors of outcome.
MRI measurements of LA volumes and function can be done in most patients using either two- or four-chamber views with low intra- and inter-observer variation. Our findings confirm echocardiographic studies showing that measures of LA function are strongly related to outcome not only in patients with systolic HF but also in those with a normal LVEF. However, LAEF might be more strongly related to morbidity and mortality than LA dimension or volume.
Measuring LAEF using CMRI might add prognostic information in patients at risk of developing HF, such as those with hypertension. Kaminski and colleagues studied 210 patients with chronic hypertension and a normal LVEF. Amongst all variables studied, including age, LVEF, and LA volume, LAEF was most strongly related to major adverse cardiac events in the following 2 years. Furthermore, in a recently published study, Gupta and colleagues showed that lower LAEF was associated with an adverse outcome even in the general population but that LA maximal volume corrected for BSA (LAmax/BSA) was only weakly associated with mortality. Our analysis also suggests that LAmax is not a powerful prognostic marker. Because BSA is itself associated with mortality both in this and much larger HF datasets, we decided to include it in the multivariable model as an independent variable rather than to correct our MRI measurements for BSA.
The mechanism(s) linking LA function with outcome are not fully understood. Exposure to high LV filling pressures will cause an increase in LA volume and it may be that a decreased LAEF is just a marker of a sicker LV, agnostic to and integrating the effects of a decline in both systolic and diastolic performance. Short-term increases in LA myocardial stretch, such as may occur during exercise, will invoke the Frank-Starling mechanism, increasing LA contractility. However, with sustained increases in LV and LA pressure, eccentric hypertrophy and dilation occurs and LA contractile reserve becomes exhausted. In end-stage HF, the LA becomes mainly a passive conduit dictated by ventricular distensibility. These alterations are often mirrored on the right side of the heart, where chronic exposure to higher pulmonary venous and subsequently pulmonary arterial pressures leads to RV hypertrophy, dysfunction and dilation and eventually to worsening peripheral congestion.
Many studies show that markers of congestion are strongly related to outcome and LAEF may be a useful measure of this pathophysiology. LAEF also predicts incident AF, which may trigger decompensation and increase mortality.
Conceptually, the prognosis of HF is driven by sudden death that may be due to arrhythmias or acute coronary events, by worsening congestion and by co-morbid conditions that may act alone or in concert. It is likely that atrial dilatation and dysfunction and increasing plasma concentrations of natriuretic peptides are different aspects of the same process of increasing congestion. Although we did not observe an association between clinical features of congestion and atrial dysfunction or natriuretic peptides, this most likely reflects a confounding effect of diuretics. If all patients were given just 40 mg/day of furosemide, we suspect that clinical congestion would become much more obvious and severe in patients with low LAEF or raised NTproBNP. Observational studies, such as this, can only speculate on the nature of associations, but we suspect that declining atrial function and rising plasma concentrations of natriuretic peptides may have a common origin, reflecting rising LV filling pressures. However, some factors may differentially influence natriuretic peptides. Renal dysfunction may cause diuretic resistance and add to congestion but will also reduce the clearance of natriuretic peptides, causing a disproportionate rise relative to changes in LA volumes or function.
Although this is a substantial study of patients with suspected HF referred for CMRI and followed up for a long period of time, with a large number of events and several measures of outcome considered, it has several limitations.
The indications for referral for cardiac MRI were not recorded and a referral bias for investigation by CMRI is likely; for instance, patients with implanted cardiac resynchronization or defibrillator devices will have been excluded. Younger patients may be more likely to be referred for and agree to MRI compared with frail elderly patients with advanced comorbidities. However, >25% of patients enrolled in this study were older than 75 years.
Most of the patients in whom HF was refuted as a diagnosis complained of exertional breathlessness. It is possible that some of their symptoms reflected reversible myocardial ischaemia or that some of them were simply unfit with high BMI. Their LAEF overlapped with those with HF in the highest quartile of LAEF but they had a higher LVEF, lower plasma concentrations of NTproBNP, and a better prognosis, suggesting that rejecting the diagnosis of HF was correct. The diagnosis of HF based on a raised NTproBNP but normal LVEF might also be questioned. The prognosis of these patients was substantially worse than patients in whom the diagnosis of HF was refuted [HR for the combined endpoint of hospitalization for HF and total mortality: 2.38 (95%CI: 1.53–3.71; P < 0.001); graph not shown], again suggesting that the diagnosis was correct.
In the clinical setting, it is probably simpler to measure LAEF by echocardiography, and newer tools are being developed to study LA deformation. Results from other large prospective studies are warranted not only to support or refute our findings.
We measured LAEF only in patients with sinus rhythm. One in five patients referred had AF. It is technically possible to measure the 'passive' LAEF caused by LA distension during mitral valve closure but we felt that there were too few patients with AF to provide a robust analysis that might only serve to complicate and confuse the results from patients in sinus rhythm.
Discussion
In patients referred with suspected HF, lower LAEF measured by CMRI is associated with higher plasma concentrations of NTproBNP, despite higher doses of diuretics, and with worse LV and RV function. LAEF is also a powerful predictor of incident AF, hospitalization for HF and of mortality in patients with HF whether or not they have LVSD. In contrast, measures of either LV or RV structure or function by MRI were only weak predictors of outcome.
MRI measurements of LA volumes and function can be done in most patients using either two- or four-chamber views with low intra- and inter-observer variation. Our findings confirm echocardiographic studies showing that measures of LA function are strongly related to outcome not only in patients with systolic HF but also in those with a normal LVEF. However, LAEF might be more strongly related to morbidity and mortality than LA dimension or volume.
Measuring LAEF using CMRI might add prognostic information in patients at risk of developing HF, such as those with hypertension. Kaminski and colleagues studied 210 patients with chronic hypertension and a normal LVEF. Amongst all variables studied, including age, LVEF, and LA volume, LAEF was most strongly related to major adverse cardiac events in the following 2 years. Furthermore, in a recently published study, Gupta and colleagues showed that lower LAEF was associated with an adverse outcome even in the general population but that LA maximal volume corrected for BSA (LAmax/BSA) was only weakly associated with mortality. Our analysis also suggests that LAmax is not a powerful prognostic marker. Because BSA is itself associated with mortality both in this and much larger HF datasets, we decided to include it in the multivariable model as an independent variable rather than to correct our MRI measurements for BSA.
The mechanism(s) linking LA function with outcome are not fully understood. Exposure to high LV filling pressures will cause an increase in LA volume and it may be that a decreased LAEF is just a marker of a sicker LV, agnostic to and integrating the effects of a decline in both systolic and diastolic performance. Short-term increases in LA myocardial stretch, such as may occur during exercise, will invoke the Frank-Starling mechanism, increasing LA contractility. However, with sustained increases in LV and LA pressure, eccentric hypertrophy and dilation occurs and LA contractile reserve becomes exhausted. In end-stage HF, the LA becomes mainly a passive conduit dictated by ventricular distensibility. These alterations are often mirrored on the right side of the heart, where chronic exposure to higher pulmonary venous and subsequently pulmonary arterial pressures leads to RV hypertrophy, dysfunction and dilation and eventually to worsening peripheral congestion.
Many studies show that markers of congestion are strongly related to outcome and LAEF may be a useful measure of this pathophysiology. LAEF also predicts incident AF, which may trigger decompensation and increase mortality.
Conceptually, the prognosis of HF is driven by sudden death that may be due to arrhythmias or acute coronary events, by worsening congestion and by co-morbid conditions that may act alone or in concert. It is likely that atrial dilatation and dysfunction and increasing plasma concentrations of natriuretic peptides are different aspects of the same process of increasing congestion. Although we did not observe an association between clinical features of congestion and atrial dysfunction or natriuretic peptides, this most likely reflects a confounding effect of diuretics. If all patients were given just 40 mg/day of furosemide, we suspect that clinical congestion would become much more obvious and severe in patients with low LAEF or raised NTproBNP. Observational studies, such as this, can only speculate on the nature of associations, but we suspect that declining atrial function and rising plasma concentrations of natriuretic peptides may have a common origin, reflecting rising LV filling pressures. However, some factors may differentially influence natriuretic peptides. Renal dysfunction may cause diuretic resistance and add to congestion but will also reduce the clearance of natriuretic peptides, causing a disproportionate rise relative to changes in LA volumes or function.
Limitation
Although this is a substantial study of patients with suspected HF referred for CMRI and followed up for a long period of time, with a large number of events and several measures of outcome considered, it has several limitations.
The indications for referral for cardiac MRI were not recorded and a referral bias for investigation by CMRI is likely; for instance, patients with implanted cardiac resynchronization or defibrillator devices will have been excluded. Younger patients may be more likely to be referred for and agree to MRI compared with frail elderly patients with advanced comorbidities. However, >25% of patients enrolled in this study were older than 75 years.
Most of the patients in whom HF was refuted as a diagnosis complained of exertional breathlessness. It is possible that some of their symptoms reflected reversible myocardial ischaemia or that some of them were simply unfit with high BMI. Their LAEF overlapped with those with HF in the highest quartile of LAEF but they had a higher LVEF, lower plasma concentrations of NTproBNP, and a better prognosis, suggesting that rejecting the diagnosis of HF was correct. The diagnosis of HF based on a raised NTproBNP but normal LVEF might also be questioned. The prognosis of these patients was substantially worse than patients in whom the diagnosis of HF was refuted [HR for the combined endpoint of hospitalization for HF and total mortality: 2.38 (95%CI: 1.53–3.71; P < 0.001); graph not shown], again suggesting that the diagnosis was correct.
In the clinical setting, it is probably simpler to measure LAEF by echocardiography, and newer tools are being developed to study LA deformation. Results from other large prospective studies are warranted not only to support or refute our findings.
We measured LAEF only in patients with sinus rhythm. One in five patients referred had AF. It is technically possible to measure the 'passive' LAEF caused by LA distension during mitral valve closure but we felt that there were too few patients with AF to provide a robust analysis that might only serve to complicate and confuse the results from patients in sinus rhythm.
SHARE
