This study demonstrates range and distribution of echocardiographic parameters in the elderly population with high grade aortic stenosis. Independent predictors of elevated NT-proBNP levels (> 500 ng/L) are PAP, mitral peak E, DT and septal E/e´ or septal e´ which can be used interchangeably. Parameters which are recommended for the assessment of diastolic dysfunction according to the 2016 ASE and EACVI guidelines (i.e. high average E/e´, low septal e velocity, high TR velocity and high LAVI) result in comparable AUC in ROC curve analyses as reported in the guidelines. However, specificities at sensitivity >60%, >45% could be increased from 23% to >75%, 85%, respectively, by appropriate change of threshold-values.
Diastolic dysfunction in aortic stenosis, hemodynamic considerations
Aortic stenosis modifies ventricular diastolic properties with three different mechanisms:
- Delayed relaxation and relaxation velocity of the mitral ring might be associated with increased afterload and left ventricular hypertrophy(14). Filling pressures can be normal in delayed relaxation (14). Delayed relaxation velocity is expressed by reduced e´ in the septal or lateral ventricular wall. Low pressure gradient due to low ventricular pressure fall and prolonged early filling period are reflected by low E velocity, prolonged IVRT and prolonged DT(15).
- Increased filling pressures can be determined by invasive measurement of ventricular pre-A pressure, which is a condition often followed by increased end-diastolic pressure and pulmonary capillary wedge pressure (PCWP). Increased filling pressures in AS are often reversible and can occur at superimposed afterload in a ventricle with reduced relaxation properties and initially normal filling pressures(14). Filling pressures can typically occur at decreased stroke volume, shortened diastole during tachycardia or at missing compensatory atrial filling in atrial fibrillation. Increased diastolic pressure gradients are associated with higher E velocities, shortened DT and IVRT, reduced PV SFF and higher PAP(1,15). Additionally, LA size and E/é are associated with increased PCWP at rest and during exercise(16).
- Ventricular hypertrophy, fibrosis and molecular mechanisms of diastolic dysfunction (17) might affect ventricular relaxation, and might also increase ventricular stiffness. These conditions, in turn, increase LV filling pressures with a non-reversible component after afterload reduction, and might contribute to persistently elevated NT-proBNP levels after TAVR/SAVR.
Impaired relaxation in elderly AS patients
Most echocardiographic parameters of diastolic dysfunction have a U-shaped function with normal values between two extremes of either impaired relaxation or increased filling pressures(15). For correct interpretation of indicators on increased filling pressures, diastolic properties of the LV at normal filling pressures in AS need to be considered. Results of our study and previous investigations(2,16,18) on diastolic parameters in AS, indicate that most AS LVs predominantly display signs of impaired relaxation. This is reflected by low e´, which increases after TAVR/SAVR, though without complete normalization (7,19).
Age alone is one important factor for reduced relaxation properties. However, the patient population with TAVR, typically ages > 75 years, is scarcely represented in epidemiological studies(20) or in control groups of clinical studies. Probably due to the low number of study subjects, the 2016 ASE and EACVI guidelines refer to the high age group as > 60 years(1).
Compared to the high age group of the guidelines, the septal and lateral e´ are significantly lower, and LAVI and DT are significantly higher in the present study. To a smaller extent, age might reduce relaxation properties in higher age groups, however, most of the diastolic parameters were not different between the two age groups ≥ 80 and <80 years in the present study.
Steine et al (2) investigated patients (65±12 years) with moderate AS, showing decreased septal e´ and increased atrial velocity compared to age-matched controls. Furthermore, E velocity, DT and E/e´ ratio were highly elevated compared to age-matched controls (17,4±10 vs 11±4)(2). These results demonstrate that AS has age independent effects on ventricular diastolic properties.
Even though the present study represents an older patient population with higher degree of AS compared to Steines and other younger study populations (2,7,19,21), we found a similar reduction of e´. Our results stress the predominance of factors other than age like LV hypertrophy, changed LV stiffness and increased afterload as the main reasons for decreased relaxation properties in AS patients. PAP was the only diastolic echocardiographic parameter which was significantly different between age-groups. One explanation of this could be low LV compliance with elevated filling pressures due to diffuse myocardial fibrosis in the elderly patients. Another explanation might be the high number of COPD patients in the elderly TAVR group.
Changes of diastolic properties after TAVR/SAVR
Even though e´ increases after TAVR/SAVR (7,19,21-23), average e´ stayed lower than 10 cm/s in 90% of postoperative AS patients, indicating persistent impaired relaxation also after afterload reduction with AVR. Postoperatively, relaxation velocities in the lateral but not septal wall were significantly higher than pre-operative values, suggesting regionally inhomogeneous response to reduced afterload.
The correlation of e´ and improvement of s´ and their increase after AVR, indicate an association between ventricular contraction and relaxation velocities, which are both influenced by afterload changes (14,24). These findings point to afterload dependent, reversible reduction of contraction and relaxation velocities of the lateral wall, whereas the bulky, structurally changed septum of hypertensive ventricles(25) have less ability to recover. The majority of previous studies report a similar degree of diastolic dysfunction in AS patients with reduced recovery of relaxation properties(7,19,21-23), while only one study could be identified describing normalized E/e´, e´ and LA volume post TAVR (26).
Indicators for increased filling pressures
The present study refers to the 2016 ASE/EACVI recommendations for the evaluation of LV diastolic dysfunction(1), which propose a simplified approach of estimating increased filling pressures (i.e. grade II and grade III of diastolic dysfunction). For assessment of impaired relaxation (i.e. Grade I diastolic dysfunction) we refer to the better definition of this state in the recommendations of 2009(15).
The current results reflect the difficulties of estimating filling pressures in ventricles with highly impaired relaxation. Impaired relaxation will naturally change thresholds for increased filling pressures due to the U-shaped function of many Doppler-based parameters. This applies for mitral peak E velocity, DT, IVRT and PV SFF (9). In opposite, the parameters e´ increases and E/e´ decreases with an additive effect of both, impaired relaxation and increased filling pressures. . In AS patients, E/e´ seem to be predominantly influenced by ventricular relaxation properties (i.e. e´) rather than mitral pressure gradients (i.e. E velocity). Thus, e´ is lower in AS patients, while E-wave velocities are not elevated compared to age-matched controls (1,2). Similarly, comparing pre- and post AVR parameters shows that reduction of E/e´ was mainly related to postoperatively increasing e´, while E velocity was unchanged.
Plasma NT-proBNP levels correlate well with elevated filling pressures in many different settings (8,10,27). For this reason, in the present study, NT-proBNP was chosen as a marker for increased end-diastolic pressures. The cut-off NT-proBNP value of 500 ng/L was derived from the clinical normalcy range for females between 70-80 years and males ≥70 years, appropriate for the majority of our patients.
Significantly reduced NT-proBNP levels after AVR indicate moderately improved but not normalized filling pressures following afterload reduction. ROC curve analyses for several diastolic parameters distinguish high and low NT-proBNP values with AUC values between 65% and 76%, which indicates moderate diagnostic value when values are measured in the intermediate range. In line with our findings, Sasaki et al showed that E/e´ was a highly sensitive and specific predictor of NT-proBNP levels, even after adjustment for clinical and systolic parameters(11). Notably, in the present study, E/e´ was not superior to low septal e´ as a marker for high NT-proBNP levels. Even though high E/é is a widely used marker for elevated filling pressures (8,11,27), changes in filling are thought to be driven by changing peak E velocity, while low e´ is thought to be closer related to impaired relaxation(15). However, the close correlation of e´ to NT-proBNP indicate that afterload and filling pressures have a direct influence on e´. A previous experimental study(14), tried to explain this inverse correlation by showing partially irreversible relation of impaired relaxation to increased filling pressures. Interestingly, peak E velocity showed low correlation with E/e´ or e´, and appeared to be an independent predictor of high NT-proBNP.
High PAP is a consistent and independent marker of elevated filling pressures(1,11,27) and a predictor for outcomes after TAVR and SAVR (21,28). In accordance with previous findings, high PAP values showed independent correlation with elevated NT-proBNP levels (27). In consistency with the 2016 guidelines, this study shows that PAP is the most specific prognosticator with incremental effect to other parameters for detection of high filling pressures.
Preoperatively and postoperatively, 91% and 78% of patients with AS, respectively, displayed at least 2-3 of 3 criteria of grade I-III diastolic dysfunction, which include septal e´, lateral e´, LAVI, according to the 2009 ASE/EACVI guidelines.
Following the 2016 guidelines, all mitral flow and tissue-Doppler based parameters indicated elevated filling pressures in the majority of AS patients with unacceptably low specificity (25%). Only TR velocity >2.8m/s indicated high NT-proBNP levels with high specificity at acceptable sensitivity.
According to the results of the present study, we suggest the use of an adjusted model for elderly patients with AS by taking the following considerations into account:
First, E/e´ and e´are good indicators of elevated filling pressures. Because of high correlation with each other, they have no significant additive value and can be used interchangeably. Septal or average E/e´ and e´ seemed to be more accurate than lateral e´ measurements in assessment of increased filling pressure.
Second, LA volume was not a significant indicator of increased NT-proBNP levels and thus of lesser value in the assessment of increased filling pressures in AS patients. LA size changes with atrial fibrillation, which was present in 25% of the AS patients, and it increased due to impaired relaxation which is present in the majority of AS patients. LA size might be a better indicator for long-term increased filling pressures when seperatedly assessed for patients with sinus-rhythm or atrial fibrillation.
Third, our study confirms a highly specific PAP cut-off level of 40 mmHg which is the equivalent of TR peak velocity of 2.8 m/s suggested in the 2016 guidelines.
Fourth, E wave velocity and E DT, are independent indicators of elevated NT-proBNP levels in the multiple logistic regression analysis. E wave velocity decreases and DT is prolonged after AVR, indicating a closer relationship with reversible pressures. However, E velocity, DT, E/e´ and e´ cut-off values have to be adjusted towards cut-off values with higher specificities to be relevant for clinical use. According to our results, at least one of the following parameters need to cross their thresholds to indicate high filling pressures in patients with pre- or postoperative aortic stenosis: E/e´sept > 20, E velocity > 100 cm/s, DT <220ms , PAP > 40 mmHg or e´sept <5.0 cm/s.
Limitation of the study
The gold standard for diastolic filling pressures is invasive pressure measurements by right or left-heart catheter, which were not available in the present study. We used NT-proBNP as a surrogate marker of increased filling pressure, as previously practiced in other studies (17,19). Even though NT-proBNP correlates well with diastolic pressures, it is not known whether age alone increases proBNP. Setting the cut-off value of NT-proBNP at 500 ng/L did not take age-adjusted normalcy into account. However, a model with age-adjusted cut-off values for NT-proBNP was tested and rendered similar results.
The present study was performed on one ultrasound-system and the same reader, resulting in highly robust tissue velocity measurements. However, tissue Doppler indices are known to differ between ultrasound-systems with vendor specific machine-settings(29). Comparison of E/e´ and e´ and cut-off values with guidelines or other studies are therefore challenging as varying results might be due to systematic errors.