Article Text
PDF
Abstract
Objective Cardiac amyloidosis (CA) is a rapidly progressive disease that portends poor prognosis. Our objective was to evaluate the prognostic impact of relative regional strain ratio (RRSR, a measure of the relative apical sparing of longitudinal strain (LS)) in CA.
Methods This is a retrospective study evaluating 97 patients with CA from 2004 to 2013. Patients were included if they met criteria for CA based on endomyocardial biopsy or advanced imaging criteria coupled with either extracardiac biopsy or genetic analysis. Baseline clinical and imaging data were collected and compared between light-chain amyloidosis (AL) (n=59) and transthyretin amyloidosis (ATTR) (n=38) subtypes. RRSR was defined as the average apical LS divided by the sum of the average mid and basal LS values. A Cox proportional hazards model was used to assess the effects of clinical and echocardiographic characteristics, including RRSR, on the outcome of time to death or heart transplantation.
Results Despite younger age, the AL subtype had a statistically significant association with the composite outcome as compared with ATTR (p=0.022). Log-transformed RRSR was independently associated with the composite outcome at 5 years (HR 2.45 (1.36 to 4.40), p=0.003). Patients with low ejection fraction and high RRSR had the worst prognosis. In multivariable analysis, RRSR remained predictive of the primary outcome (p=0.018). Addition of covariates related to systolic function (global LS and ejection fraction) to the model attenuated this effect.
Conclusions High RRSR is adversely prognostic in patients with cardiac amyloid. This novel tool is both diagnostic and prognostic and may have implications in management and suitability for treatment.
Statistics from Altmetric.com
Introduction
Amyloidosis is a systemic disease characterised by extracellular deposition of misfolded protein into various organs.1 Amyloid infiltration of the heart typically leads to a restrictive cardiomyopathy and progressive congestive heart failure. Cardiac involvement is almost exclusively due to deposition of either systemic light-chain amyloidosis (AL) or the transthyretin amyloidosis (ATTR). Differentiating between these subtypes of cardiac amyloidosis (CA) is essential to guide management and target therapy. Amyloid subtype is extremely prognostic, as AL CA has a mortality measured in months while ATTR survival is measured in years.2
While cardiac biopsy remains the reference standard for diagnosing CA, recent advances in multimodality imaging have allowed for a non-invasive approach. Echocardiographic patterns of increased wall thickness, diastolic dysfunction and a speckled appearance of the left ventricle (LV) myocardium have been used to raise suspicion for CA; however, these methods are not sensitive or specific enough to make a definitive diagnosis. Relative regional strain ratio (RRSR), a measure of the degree of apical sparing of longitudinal strain (LS), has been used to identify CA as part of the diagnostic algorithm for undifferentiated left ventricular hypertrophy.3 Patients with CA have regional sparing of the apical LS segments relative to the mid wall and basal segments. This ratio is a sensitive and specific non-invasive means of diagnosing amyloid involvement of the myocardium.4 However, the significance of this ratio on disease outcome has not been established. We sought to evaluate the prognostic utility of RRSR in patients with established CA.
Methods
Patient population
This is a retrospective cohort study of patients referred to the Cleveland Clinic between 2004 and 2013 with a confirmed diagnosis of CA. Patients were included if they met the diagnostic criteria described below. Patients with prior history of myocardial infarction, significant valvular disease (≥3+ valvular regurgitation or stenosis or prior history of valve surgery) and inadequate echocardiogram image quality were excluded from the study.
The diagnosis of CA was confirmed by the treating cardiologist and was based on (1) positive endomyocardial biopsy or (2) imaging confirmation of the disease (without the use of myocardial strain). Endomyocardial biopsies were reviewed by two cardiac pathologists; amyloid was confirmed with positive Congo red or Thioflavin S stains. Patients were further divided into the two major subtypes of CA, AL and ATTR, based on biopsy immunohistochemistry and/or molecular analysis for genetic mutation for ATTR. Patients included based on imaging criteria had to have a positive extracardiac biopsy in AL or a positive TTR gene mutation in ATTR. Echocardiographic features consistent with CA included diastolic dysfunction and an increase in LV wall thickness (>12 mm) in the absence of a history of hypertension.5 Features of CA on cardiac MRI included diffuse delayed enhancement in a non-ischemic pattern and abnormal T1 kinetics on Look-Locker sequences.6 Technetium-99m pyrophosphate scans were considered positive based on expert interpretation.7
Mortality was identified either through social security death index or the electronic medical record system. Patients were followed from the date of the initial echocardiogram to the date of death or last follow-up. The study was approved by the institutional review board as well as the ethics committee and patient-informed consent was not necessary due to this retrospective analysis.
Measurement techniques
Clinical, laboratory, imaging and mortality data were extracted from the electronic medical record (Epic, Verona, Wisconsin, USA). Estimated glomerular filtration rate was calculated by using the Chronic Kidney Disease Epidemiology Collaboration formula.
The ECG at the time of CA diagnosis was reviewed for each patient to identify characteristics suggestive of CA. Low voltage was defined as a QRS amplitude of ≤0.5 mV in all limb leads or ≤1 mV in all precordial leads.8 Pseudoinfarct pattern was defined as the presence of QS waves in two consecutive leads in patients with no prior myocardial infarction or, if available, cardiac catheterisation with non-obstructive coronary artery anatomy.
Echocardiography
Echocardiography was performed on Vivid 7 or Vivid 9 ultrasound systems (GE Medical, Milwaukee) and measurements were independently analysed offline in a blinded fashion. Only patients with studies done on a GE system were included to maintain consistency with the techniques used in our previous description of regional variation of LS in CA.4 ,5 Inadequate image quality was defined as a frame rate <40 frames per second or an inability to visualise or perform adequate tracking on more than two myocardial segments. Septal thickness, LV end diastolic diameter and posterior wall thickness were assessed by two-dimensional (2D) measurements at the level of the mitral valve leaflet tips in diastole. Left atrial volume and LV ejection fraction (EF) were measured using Simpson's method. Diastology parameters were collected in the apical four-chamber view using pulsed wave and tissue Doppler. All measurements were in accordance with American Society of Echocardiography guidelines.9 ,10
Peak systolic LS
Peak systolic LS assessment was performed using EchoPAC software (EchoPAC V.113, Advanced Analysis Technologies; GE Medical Systems). The three standard apical chamber views were used for LS measurements. Frame-by-frame tissue speckle tracking of the LV endocardium was performed and an automatically generated bulls-eye plot of segmental peak systolic LS values was obtained. Global LS was calculated using the average of the regional LS values. Strain values for the six basal, six mid and five apical views of the LV were averaged to obtain three regional LS values (basal, mid and apical). RRSR was calculated by dividing the average apical LS by the sum of the average basal and mid LS values, as described previously (figure 1).4
Longitudinal strain (LS) imaging in cardiac amyloidosis. This is a representative example of peak systolic LS imaging in a patient with biopsy-proven cardiac amyloidosis. There is apical sparing of LS as calculated by more negative values visualised on the target plot. The relative regional strain ratio (RRSR) in this example is 1.28.
Statistical methods
Continuous variables were expressed as either mean±SD or median (IQR) and analysed by the Student's t test (or analysis of variance) or Mann–Whitney U test for parametric and non-parametric variables, respectively. Categorical variables were expressed as n (%) and were compared by the χ2 test or Fisher's exact test, where appropriate. Variables were log transformed if not normally distributed.
Survival analysis was performed individually for each variable using Cox proportional hazards models after observing no trends with time for the Schoenfeld residuals. The end point used was time to all-cause mortality or heart transplantation. No patients received left ventricular device support. Cumulative mortality or heart transplantation was assessed by the Kaplan–Meier method and compared via the log-rank test. A fractional polynomial curve was fitted to show the continuous relationship between RRSR level and the composite outcome.
Multivariable adjustments were performed for amyloid type, estimated glomerular filtration rate, New York Heart Association (NYHA) class ≥III, LV mass index, early mitral deceleration time and log-transformed RRSR. Covariates were chosen for clinical factors, burden of myocardial infiltration and diastolic function based on prior prediction for mortality and the likelihood for confounding the relationship between RRSR and the composite outcome.11–14 Harrell's C-statistic was used to compare model discrimination. Additional measures of systolic function, global LS and EF, were substituted for RRSR in the multivariable models for qualitative comparison. Sensitivity analysis with a competing risk model as described by Fine and Gray was used to assess the competing risk of heart transplantation on mortality. In addition, a χ2 test for homogeneity was run to assess the interaction between amyloid type and RRSR on the composite endpoint.
All analyses used double-sided p values, and p<0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics V.20.0 (IBM, Armonk, New York, USA) and Stata V.13.1 software (StataCorp LP, College Station, Texas, USA).
Results
Patient population
A total of 269 patients with a diagnosis of CA were reviewed from 2004 to 2013. There were 112 patients excluded due to a prior history of myocardial infarction or significant valvular disease; a further 42 were excluded as their studies were performed on non-GE machines, and 8 studies were excluded due to inadequate image quality. Therefore, 97 patients with CA were identified: 59 patients with AL and 38 patients with ATTR. The most common subtype was AL lambda (48%), followed by wild-type ATTR (24%), hereditary ATTR (16%) and AL kappa (12%). Val122Ile was the most common mutation (10 patients, 67%), followed by Thr60Ala (4 patients, 27%) and Phe22Leu (1 patient, 6%). Out of the total CA population, 73 (75%) had a cardiac biopsy confirming CA diagnosis and 21 (22%) had an extracardiac biopsy that was consistent with amyloidosis. Three patients did not have any biopsy performed, but had positive TTR genetic testing. One or more extracardiac biopsies were performed in 21 patients: 12 (57%) bone marrow, 6 (29%) intestinal, 5 (24%) renal, 5 (24%) stomach/oesophagus, 2 (10%) skin/fat, 1 (5%) lung, 1 (5%) spleen, 1 (5%) liver, 1 (5%) bladder and 1 (5%) conjunctiva.
Baseline demographic and clinical characteristic data are outlined in table 1. Overall mean age was 64±11 years. Patients with ATTR were older, while those with AL were more likely to be Caucasian, have a higher incidence of haematological disorders and a higher N-terminal prohormone of brain natriuretic peptide (NT-proBNP) level.
Baseline patient demographics
ECG and echocardiography
Baseline echocardiographic and ECG data were collected closest to the date of CA diagnosis and are outlined in table 2. After excluding nine patients with paced rhythms on ECG, 47% had low-voltage pattern only, 38% had a pseudoinfarct pattern only and 22% had both ECG findings. There was no significant difference between the AL and ATTR subtypes for ECG patterns; however, patients with AL were more likely to have both low-voltage and pseudoinfarct pattern.
ECG and echocardiographic characteristics
The mean EF was 49.6±11.0% with 73% of patients having a preserved EF (defined as EF ≥45%). There were no significant differences between the two subtypes for all the other echocardiographic parameters. LS parameters are shown in table 2. Global LS was decreased in all patients including those with preserved EF (−10.38 (−13.54 to −8.24)), but there was no difference seen between the AL and ATTR subtypes. LS became less impaired with progression apically down the ventricle; basal LS was the most impaired and apical LS was relatively preserved. The RRSR was 1.19 (0.93 to 1.59) in the overall cohort and 1.14 (0.92 to 1.48) in those with preserved EF. There was no significant difference between amyloid subtypes for basal, mid, apical, global or RRSR parameters.
Outcomes
Median follow-up time to date of death, heart transplantation or censoring in days was 496 (51–967) in AL, 685 (433–1103) in ATTR and 638 (171–1028) in the combined cohort. Over the 194 person-years of follow-up, there were 53 deaths and 4 heart transplants in the total CA population (59%). In AL, there were 38 deaths and no transplantations (64%) while in ATTR there were 15 deaths and 4 transplantations (50%) (p=0.16). Patients with AL had significantly higher mortality or heart transplantation at 5-year follow-up (p=0.022).
Kaplan–Meier estimation of 5-year all-cause mortality or heart transplantation dichotomised by median RRSR is illustrated in figure 2. There was a significant increase in the composite endpoint at 5 years in patients with RRSR above the median (p=0.026). Figure 3 illustrates the continuous association of RRSR and the 5-year composite outcome. Higher RRSR values were associated with a higher probability of the composite outcome and risk increased at an RRSR of about 1.3. Figure 4 illustrates the contribution of LV EF and RRSR on mortality. In comparison with the patients with low EF and high RRSR, all other combinations have lower mortality risk: high LV EF and high RRSR (HR 0.464, 95% CI 0.241 to 0.895, p=0.022), low LV EF and low RRSR (HR 0.354, 95% CI 0.119 to 1.053, p=0.062), high LV EF and low RRSR (HR 0.388, 95% CI 0.206 to 0.733, p=0.004).
Kaplan–Meier curves stratified by median relative regional strain ratio (RRSR). This is an unadjusted Kaplan–Meier curve stratifying RRSR by the median value of 1.19. Values above the median were associated with increased progression to death or heart transplantation.
The continuous association between the relative regional strain ratio (RRSR) and mortality or heart transplantation risk. This is a fractional polynomial curve fitted to assess the relationship between the RRSR and the composite outcome of death or heart transplantation. The natural log of the HR per unit change in RRSR is plotted on the y axis against the RRSR on the x axis. Higher RRSR values are associated with a higher probability of the composite outcome.
Kaplan–Meier curves stratified by median relative regional strain ratio (RRSR) and ejection fraction (EF). Data are separated into four curves based on RRSR above and below the median and preserved EF. The median RRSR was 1.19; preserved EF was ≥45% and reduced EF was <45%. The largest difference was seen in the preserved EF and low RRSR versus reduced EF and high RRSR (p=0.004).
The unadjusted association between clinical and imaging variables and the composite outcome is shown in table 3. At the time of echocardiography, NYHA class ≥III, lower EF, lower albumin, lower average E/e′, higher global LS and higher log-transformed RRSR were associated with the composite outcome. As shown in table 4, after multivariable adjustment for amyloid type, NYHA class ≥III, estimated glomerular filtration rate, LV mass index and deceleration time, the association between log-transformed RRSR and mortality remained a significant predictor of death or heart transplantation. In a sensitivity analysis, risk-adjusted RRSR remained independently associated with mortality after accounting for the competing risk of heart transplantation (p=0.001). This model predicted the composite outcome with a C-statistic of 0.7065 (95% CI 0.632 to 0.781, p<0.001). When other measures of systolic function (EF and global LS) replaced RRSR in the multivariable model, there was a similar performance of the model with C-statistics 0.6991 (95% CI 0.629 to 0.769) and 0.7248 (95% CI 0.655 to 0.794), respectively. In addition, the amyloid subtype (AL vs ATTR) did not significantly interact with the association between RRSR and the composite outcome (p=0.369).
Univariable association with mortality or heart transplantation
Multivariable analysis of the association between log-transformed RRSR and mortality
Discussion
RRSR, a measure of the relative apical sparing of LS, is a sensitive and specific marker of cardiac amyloid involvement4 and aids in diagnosing patients with undifferentiated left ventricular hypertrophy.3 In this study, we assessed the ability of RRSR to predict all-cause mortality or heart transplantation in patients with an established diagnosis of CA. We found that RRSR is strongly predictive of death or heart transplantation in univariable and multivariable analyses and propose RRSR as a unique marker in CA that provides both diagnostic and prognostic value.
Differences between AL and ATTR CA
Morphological, functional, electrical and clinical parameters including prognosis were compared between the different subtypes of CA. There were no significant differences in echocardiographic parameters between amyloid subtypes, though a trend towards more left atrium (LA) enlargement in the ATTR subtype was noted. The combination of both low-voltage and pseudoinfarct pattern on ECG was noted more commonly in the AL subtype. There was no difference in median RRSR values between AL and ATTR subtypes and the amyloid subtype did not significantly interact with RRSR with regard to the composite outcome of all-cause mortality or heart transplantation.
Our study confirms the previously described tenet that patients with AL have a higher mortality and faster rate of progression as compared with ATTR. This is likely due to more systemic involvement of amyloid light chains including renal, bone marrow and autonomic nerve disease. Light-chain cardiac toxicity has also been thought to play a role and may be additive to the typical diastolic heart failure caused by fibril deposition.15 ATTR is a more indolent process than AL CA, which may allow for the development of local compensatory mechanisms prior to development of reduced EF.
Prognosis in CA and RRSR
Earlier studies have shown certain clinical and echocardiographic parameters to be associated with mortality in CA including NYHA class,11 early mitral inflow velocity, deceleration time16 and LV wall thickness.12 Echocardiographic assessment of systolic global LS has also been used to evaluate prognosis in CA.13–15 We confirmed the prognostic ability of NYHA class, EF and global LS, while finding a non-significant correlation between LV wall thickness with the composite endpoint. Regarding parameters of volume overload and diastolic dysfunction, average E/e′ was significantly associated with the composite outcome, while deceleration time and LA volume index trended towards significance. Albumin, a marker for nutritional status, was also associated with the composite outcome as anticipated.
After multivariable analysis using factors previously described as predictive in CA, only NYHA class, amyloid subtype and RRSR were found to be associated with the composite outcome. The association between RRSR and the outcome was muted with addition of other parameters of systolic function (EF and global LS); however, this was likely due to significant collinearity between the variables.
The value of using LS as an additive means of assessing LV systolic function and prognosis in CA when compared with EF is again highlighted in this study. As peak systolic LS is thought to be an earlier predictor of systolic dysfunction than EF, it follows that global LS and the RRSR were both significantly impaired in our study in the setting of a median EF of 50%. LS values became more abnormal with progression from the apex (−15.9) to the mid (−8.3) to the basal (−3.8) segments. This suggests that the RRSR, a ratio of apical strain to mid+basal strain, may become abnormal prior to significant perturbations in global LS or EF. Furthermore, in patients with preserved EF, there was a trend towards higher mortality in those with higher RRSR. This is hypothesis generating as we did not specifically test the temporal association of regional strain values in CA, and represents an area for further research.
It is rare to have a disease marker that is both sensitive and specific for the diagnosis of a disease and also prognostic for hard outcomes; glycosylated haemoglobin in diabetes and brain natriuretic peptide in heart failure are two prime examples. RRSR holds prognostic importance and has potential utility in the assessment of patients’ suitability for aggressive treatment such as chemotherapy, novel TTR pharmaceuticals and advanced heart failure therapies such as heart transplantation and left ventricular assist devices. For example, stem-cell transplantation is not recommended in AL CA patients with a troponin T level >0.06 ng/mL or NT-proBNP level >5000 pg/mL.17 These markers are not ideal as they are significantly affected by renal function and volume status. Prognostication with RRSR may help to refine risk stratification in these patients; further studies are required to assess this potential utility.
Study limitations
This is a retrospective study at a single large tertiary referral centre, which may limit external generalisability and introduce selection bias. Based on our sample size, we cannot exclude the possibility of type I error. However, our population size was comparable with other prior analyses in CA cohorts. Patients may have received interim therapies that may have affected prognosis. Not all patients underwent endomyocardial biopsy for a definitive diagnosis of CA; however, advanced multimodality imaging and genetic testing were used to confirm the diagnosis in the 25% of patients who did not undergo cardiac biopsy.
Conclusion
RRSR (a marker of the relative apical sparing of LS) is independently predictive of all-cause mortality or heart transplantation in CA despite multivariable risk adjustment for other adversely prognostic factors in CA. Patients with a low EF and high RRSR were found to have the worst prognosis. This study adds to the body of data justifying the use of systolic strain imaging by 2D speckle tracking in this disease. Specifically, the RRSR has now been found to be both diagnostic and prognostic in CA. Incorporation of the RRSR into prognostic algorithms in CA may have implications in management and suitability for treatment.
Key messages
What is already known on this subject?
Cardiac amyloidosis (CA) portends a poor prognosis and early detection is necessary given the limited therapies in advanced disease. Relative regional strain ratio (RRSR) is a unique strain pattern characterised by preserved apical longitudinal strain with reduced strain in the mid and basal segments. It is easily identifiable and highly sensitive and specific for CA, yet the prognostic implications are uncertain.
What might this study add?
In this study of 97 patients with confirmed CA we demonstrate the adverse prognostic impact of an elevated RRSR level. RRSR is predictive of death or need for heart transplantation.
How might this impact on clinical practice?
Assessment of strain in undifferentiated left ventricular hypertrophy for the assessment of RRSR should be considered a routine metric in all laboratories. This study further highlights the utility of this parameter. Further studies are needed to assess whether strain can be used to assess response to therapy.
References
Footnotes
AS and BWS contributed equally to this study.
Contributors PT, WJ, PC, MH, ZP and DP participated in the study design. AS and BS drafted the manuscript. AS, BS and KK participated in the data collection. BS and JG analysed the data. JG and DP critically reviewed the paper. All authors read and approved the final manuscript.
Competing interests None declared.
Ethics approval IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The original data are housed in a secure database accessible only to the authors who remain at the Cleveland Clinic. This data are part of an IRB-approved Cardiac Amyloid database at the Cleveland Clinic.