DISCUSSION
In this study we documented for the first time that both SVD and LVEDD (or LVEDV) were significantly correlated with LVOTD in addition to patient BSA and height (or gender) in patients undergoing echocardiography. SVD had the strongest association with LVOTD. On the basis of these observations, we derived a comprehensive equation including SVD, LVEDD and patient BSA and height, which showed a high correlation with LVOTDM and better performance than LVOTDBSA in a validation patient cohort. Alternative equations including LVEDV instead of LVEDD with/without gender were also obtained but did not improve the results of the primary analysis.
Limits of anthropometric variables in determining chamber and aortic root dimensions in adults. Aortic root normal-reference studies have notoriously shown low correlation between anthropometric variables (mainly BSA and height) and SVD in adults (18,19). For example, in the EACVI NORRE aortic root study, linear models considering age, gender, and body size barely explained around one-quarter of the total variance in determining aortic root dimension in adults (19). This is why regression equations for prediction of aortic size (and derived nomograms) based only on these parameters should be interpreted with caution, taking into account this limitation (19). In our derivation study, BSA showed moderate correlation with LVOTD and low correlation with SVD. Even in our validation study, LVOTDBSA had moderate correlation with LVOTDM, reflecting moderate correlation between BSA and LVOTD. These observations are in line with previously reported correlations (14,18,19). All these findings indicate that the wide biological variability in aortic and LVOT dimension is not entirely explained by simple demographic and anthropometric variables (19).
In our study, a regression equation including anthropometric and biologic (LV and aortic root) dimensions had high correlation with LVOTDM in the validation group. These results may be explained by a proximity anatomical effect in distorting the LVOT independently from BSA. To our knowledge, this correlation has never been explored before. Interestingly, both intra-and inter-observer variability of LVOT area derived from regression equations were lower than those related to the measured LVOT, which were similar to previous findings (22). This may be due to the well-known amplification of the error derived from squaring the LVOTD measurement when calculating the LVOT area.
Clinical implications and perspectives . There are two main clinical messages from this study: 1) measures of LV output and aortic stenosis severity that avoid use of LVOTD measurement may be misleading in particular anatomical conditions, namely abnormal LV and aortic root dimensions; 2) a well-performing formula for predicting LVOTD is now available accounting not only for anthropometric data but also for LV and aortic root dimensions.
Regarding the first message, it should be underlined that, while LVOT is a BSA-dependent measure (14,15), LVOT velocity-time integral (or stroke distance) has been hypothesized to be independent on BSA and to represent, therefore, a SVI analog (15). However, two issues limit this assumption. Firstly, the BSA independency was originally supposed in normal individuals (15): in these subjects a minor role of LV cavity and aortic root dimension is expected. However, in patients with aortic root enlargement, normal BSA and reduced stroke distance suggesting a low flow status, a relatively high LVOTD assures a normal SVI. For example, with a BSA of 1.8 m2, a stroke distance of 12 cm and a LVOTD of 27 mm, SVI is 38 ml/m2. Secondly, the accuracy of stroke distance has never been compared for assessing LV output with invasive measurement, whereas SV (or other LVOTD-including measures such as cardiac output) has shown high correlation with the invasive measurement in several studies (22-31). In addition, comparing the usefulness of SVI and stroke distance as prognostic markers, a better stratification capability of SVI was shown in one study (16).
The LVOTD measurement is also avoided in the aortic dimensionless index, which has been proposed in addition to aortic valve area calculation for aortic stenosis grading (7). Although the dimensionless index has shown prognostic value in asymptomatic and minimally symptomatic patients with aortic stenosis with preserved LV ejection fraction and without significant valve regurgitation (32), it was not better than other markers of aortic stenosis severity, including aortic valve area, for grading aortic valve stenosis in another study (33).
Our findings evidence pathophysiological reasons to encourage calculation of SVI and aortic valve area over stroke distance and dimensionless index whenever possible in the assessment of LV output and aortic stenosis severity grading. In particular, the interpretation of LVOTD-free indexes should be taken with caution in cases of abnormal LV and aortic root dimensions.
Previous studies tried to overcome the limits of LVOTD measurement by using alternative techniques, such as three-dimensional echocardiography (34) and multidetector computed tomography (35) for the assessment of aortic stenosis severity. However, it should be noticed that different techniques could imply different cut-off values to refer to for prognostic assessment: for example, an aortic valve area cut-off value of 1.2 cm2 should be used for assessing aortic stenosis severity with a hybrid multidetector computed tomography-Doppler approach rather than the established 1 cm2 obtained with the standard ultrasound method (7,35). Thus, in current clinical echocardiographic practice measuring LVOTD remains an almost inevitable, although sometimes difficult task.
Study limitations. 1) We recognize that our formula uses four different variables (BSA, height, SVD and LVEDD), thus it might seem too complicated for clinical practice. However, those measures are commonly obtained in standard echocardiography and the equation could be easily integrated in the echocardiograph to give an automatic response once all the measurements are performed, without additional time consume. In addition, many reporting software can provide derived calculations (such as the aortic valve area) automatically and could therefore integrate our formula as an assistance tool for cases in which LVOTD is scarcely or incorrectly visualized. 2) Our formula includes both BSA and height. Although height is one of the variables used for the calculation of BSA, there was no significant multicollinearity in the regression model (variance inflation factor for height and BSA were 2.2 and 2.1, respectively), thus both variables were included in the formula. 3) Multidetector computed tomography was not performed in this study as an external reference for TTE LVOTD evaluation. However, we documented that in our patients the LVOTD measurements at TTE and TEE were similar, thus indicating accuracy of TTE measurements. Our data are in line with those of Leye et al., who showed a very high correlation between TTE and TEE for measuring the LVOTD (R=0.95, P<0.001; P=0.26 for difference between techniques) (14). 4) LVEDD was chosen as LV dimension parameter for the primary endpoint rather than LVEDV because of the proximity between LVEDD and LVOTD measurement sites (the basal LV and the LVOT, respectively). Whether three-dimensional echocardiography would have given a higher correlation between LV volumes and LVOTD is unknown. However, the regression equation including LVEDD had the highest correlation with LVOTDM. 5) In the present study LVOTD was measured at the level of the annulus, which is considered the preferred site by current guidelines on aortic stenosis assessment (7), even though there is lack of general consensus and some laboratories use a more proximal site (up to 1 cm apical to the annulus) for this measure. Our findings may not be applied to predicting LVOTD at different levels of the LVOT. Evidences in support of the measurement at the level of the aortic annulus have been recently reviewed (36), in addition to a very recent phase-contrast cardiovascular magnetic resonance comparative study (37).
Conclusions . We documented that aortic root and LV dimensions affect LVOTD independently from anthropometric data in an adult heart failure patient cohort. On this ground, it is evident that bypassing LVOTD measurement with surrogate markers of LV output (stroke distance) and dimensionless indexes in aortic stenosis severity grading may be misleading, especially in presence of abnormal LV cavity and aortic root dimensions. A comprehensive equation for predicting LVOTD, including aortic root and LV dimensions, was derived and validated in the present study to help the physician in the calculation of LVOTD when it is difficult to calculate. Further multicenter investigations are needed to support a more extensive application of our formula in routine echocardiographic clinical practice.
Conflicts of Interest: Nothing to disclose.