In this cohort of light chain amyloidosis subjects, presenting heart failure class was a strong and independent predictor of mortality. Among echocardiographic structural, systolic and diastolic functional indices, only left ventricular ejection time had additive independent prognostic value to heart failure class. Ejection time of ≤240 ms was associated with increased mortality and had good sensitivity and specificity for predicting 1-year all-cause mortality as well as 1-year cardiac death. Liver function assessed by alkaline phosphatase level was also an independent predictor of mortality in AL.
Our findings are consistent with prior studies that reported adverse consequences in the presence of heart failure, with median survival of less than 5 months 4, 21
. In AL, extent of cardiac involvement is the most important determinant of clinical outcome 22
. Cardiac involvement is thought to be present in ~50% of light chain amyloidosis cases 2
, but recent studies using late gadolinium enhancement magnetic resonance imaging suggest cardiac involvement could be as high as 70% or more 23, 24
. Echocardiography is an established noninvasive means to evaluate the presence and extent of cardiac involvement. On 2D echo, thickened myocardium with “granular sparkling” appearance and reduced left ventricular systolic function are seen in amyloid patients 8
. Doppler echo demonstrate diastolic dysfunction with increased E/A ratio as a result of impairment in ventricular compliance leading to shorter deceleration time, reduced pulmonary vein peak systolic velocity and progressively more restrictive physiology 6, 7
. Pulsed tissue Doppler also show impaired longitudinal myocardial velocity 9, 10
. More recently, early systolic dysfunction has been reported in AL patients in the form of intraventricular regional dyssynchrony measured by strain 11
or 3-dimensional echocardiography 12
The prognostic value of Doppler assessment of diastolic function has been described by Klein and colleagues when they reported increased cardiac mortality in AL patients with deceleration time less than 150 ms (mean duration of follow-up of 18 months) and found that reduced deceleration time together with increased E/A ratio were stronger prognosticators than ventricular thickness and fractional shortening using bivariate analyses 13
. Their findings were consistent with studies showing the lack of prognostic value of ventricular thickness and ejection fraction 5
. Mitral annular peak systolic velocity is a measure that parallels left ventricular ejection fraction; the values obtained in our cohort is closer to the values reported by previous investigators in normal subjects than to the values in subjects with systolic dysfunction and heart failure 19, 20
, in parallel to the normal ejection fraction observed in the group. Our results from a longer duration of follow-up were similar to Klein and colleagues’ findings in terms of the importance of heart failure class in prognostication but also differ because deceleration time and E/A were not found to be independent predictors of mortality on multivariable analysis. This difference can perhaps be explained by the fact that abnormal deceleration time and E/A were associated with degree of heart failure (); when placed in a multivariable model, these diastolic indices fell out as having any additive prognostic value to presenting heart failure class. In contrast, ET did not differ significantly in patients without and with heart failure () and multivariable analysis confirmed its independent and additive valu e to heart failure class. Among 18 patients who initially presented with no heart failure (Class I), 3/3 (100%) with ET≤240 ms and 3/15 (20%) with ET>240 ms died during follow up, suggesting the potential of ET for early identification of at-risk AL patients without overt clinical heart failure.
Recently, Bellavia and colleagues followed AL subjects (median follow up 13 months) and found that the only independent predictors of mortality were ET and level of brain natriuretic peptide with greater reduction in ET in more advanced cases 11
. Our study differs from the study by Bellavia and colleagues in having a longer median time to follow-up (29.4 months); our study therefore complements the previous work in showing the robustness of ET as an independent predictor of AL mortality for both short term and long term follow-up. Our study, together with the study of Klein and colleagues 13
point to heart failure class as the strongest predictor of adverse outcome in this disease. This differs from the finding of Bellavia and colleagues showing heart failure class did not predict mortality. This difference may be related to two things: one, brain natriuretic peptide is potentially correlated to heart failure class and multivariable analysis may have resulted in this surrogate marker of heart failure to come out instead of heart failure class as an independent factor. Second, a greater proportion of AL patients in our cohort had advanced heart failure (NYHA Class II-IV, 57% versus 33% from previous study) and this may account for some of the study differences. Our results taken together with the previous study strengthen the robustness of ET as a predictor of mortality in both advanced and less advanced forms of AL. We similarly demonstrated that compared to conventional echo indices of cardiac amyloidosis (mass, thickness, ejection fraction, degree of diastolic function), ET is the most important predictor of adverse outcome. In the current study, ET ≤240 ms had good sensitivity (61%) and high specificity (90%) for predicting 1-year all-cause mortality, with even better prediction of 1-year cardiac death (73% sensitivity and 90% specificity). It is not surprising that ET has superior ability to predict 1-year cardiac death versus 1-year all cause mortality because it is a measure of cardiac function. AL is a multiorgan disease and as shown by our cohort, patients’ mode of exit is frequently cardiac plus other organ failure. From a clinical standpoint, it is useful that ET as a noninvasive measure can predict not only cardiac mortality but also all-cause mortality. Because a 1-point change in ET is a small change, it is expected that the change in hazard or risk is also small. However, when we consider a 10 ms change, likely a more clinically meaningful change, an increase in ET by this amount reduces risk of mortality by 13%. To put this into context of other cardiovascular diseases, a recent large hypertension study involving 180,000 patients showed that every 10 mm Hg increase in systolic blood pressure is associated with a 9% increased risk of death in older patients 25
and health professionals consider this degree of risk sufficient to warrant aggressive public health promotion program aimed at lowering blood pressure. In a disease such as AL that is associated with >40% 1-year mortality, a noninvasive test that can predict 13% reduction in risk is extremely relevant.
The utility of ET may also potentially be applicable to other forms of cardiac amyloidosis. In a retrospective study of 45 patients with cardiac amyloid (6 of whom had senile or familial amyloidosis), the Tei index which incorporated ET together with isovolumetric contraction and isovolumetric relaxation time was shown to be the only parameter aside from heart failure class that was independently predictive of survival 14
The mechanism underlying the adverse prognostic consequence of reduced left ventricular ejection time is unknown and our prospective observational study is limited in providing mechanistic insight. Left ventricular ejection time is an old established measure of ventricular performance and is known to be sensitive to changes in inotropic state although it is also affected by preload and afterload 26-28
. It has been used in the diagnosis and assessment of valvular disease, coronary artery disease, pericardial disease and hypertensive heart disease 27
. The severity of ET abnormality paralleled the increase in functional class and reduction of resting cardiac index 29
. Decreased ET was noted in subjects with hypovolemia 30
. In cardiac amyloidosis, reduced ET was described using phonocardiogram and carotid pulse wave tracing 31
as well as Doppler echocardiography 14
. The physiologic basis may be related to the fact that since systole is fixed in duration, myocardial dysfunction from amyloidosis can prolong preejection period leading to reduced ejection time. Indeed, prior studies on amyloid patients demonstrated both prolonged preejection period and decreased ejection time 14, 31
. In addition, it has been postulated that the inability of the infiltrated myocardium to increase end-diastolic volumes shortens ejection time leading to reduced stroke volume and adverse clinical outcomes 14
. A reduction in ET was recently found useful in identifying abnormal ventriculoarterial coupling in stable heart failure subjects 32
. Reduced ET was also found in advanced AL subjects who demonstrate hypersynchronization of regional systolic function, in contrast to dyssynchrony in less advanced stages of the disease 11
. These recent studies suggest that ET may be a sensitive marker of dysfunction in contractile synergy among ventricular segments and between atria and ventricles and that mechanical synchrony may play an importa nt role in the pathophysiology of AL amyloidosis.
It remains to be seen whether ET can be used to risk-stratify AL subjects in terms of aggressive treatment that currently includes chemotherapy, autologous stem cell transplantation or cardiac transplantation. Ejection time is currently not used to evaluate candidacy for stem cell transplantation (patients with advanced cardiac involvement are often ineligible for stem cell transplantation because of increased transplant related mortality 33-35
). Therefore its role in the risk-adapted approach 33, 36
that is currently utilized by oncologists to determine eligibility for stem cell transplantation needs to be further studied. Similarly, ET may have potential usefulness in evaluating physiologic response to treatment on serial follow up.
A major limitation of the study was the small sample size. Light chain amyloidosis, although increasingly recognized clinically, nevertheless remains a comparatively rare disease 2
. The longer follow up period of the study and frequency of outcome mitigated the relatively small sample size. The sample size limited the ability to add more clinical variables to test in a Cox model because of resulting unstable parameter estimates. Furthermore the small sample size, the temporal delay between echocardiography and chemotherapy/stem cell transplantation and the risk-adapted approach that oncologists use to decline stem cell transplantation in patients with advanced cardiac involvement due to excessive risk 33, 36
precluded performing sophisticated and complicated multivariable modeling to assess the independent role of treatment options in patients’ survival. The results of this study therefore need to be validated further in a larger prospective study. We also lack data on myocardial strain that was shown to be abnormal in light chain amyloidosis 9, 10, 37
as well as right ventricular diastolic function, so these variables were not included in the multivariable analysis. However, a prior study revealed that myocardial strain was not an independent predictor of outcome in AL in a multivariate analysis that included ET; the latter, on the other hand, was found to be prognostic of outcome 11
Summary and clinical implications
Light chain amyloidosis in this cohort was associated with high 1-year and long-term mortality. Left ventricular ejection time measured by pulsed Doppler echocardiography predicted long-term mortality in light chain amyloidosis independent of heart failure status. It was a sensitive and specific test in assessing 1-year all-cause mortality as well as cardiac death. Ejection time may be useful in risk-stratification of AL amyloid subjects by identifying vulnerable patients who may benefit from aggressive treatment or enhanced surveillance.