Application of the proximal isovelocity surface area method for estimation of the effective orifice area in aortic stenosis

Although the echocardiographic effective orifice area (EOA) calculated using the continuity equation is widely used for the assessment of severity in aortic stenosis (AS), the existence of high flow velocity at the left ventricular outflow tract (LVOT) potentially causes its overestimation. The proximal isovelocity surface area (PISA) method could be an alternative tool for the estimation of EOA that limits the influence of upstream flow velocity. EOA was calculated using the continuity equation (EOACont) and PISA method (EOAPISA), respectively, in 114 patients with at least moderate AS. The geometric orifice area (GOA) was also measured using the planimetry method in 51 patients who also underwent three-dimensional transesophageal echocardiography. Patients were divided into two groups according to the median LVOT flow velocity. EOAPISA could be obtained in 108 of the 114 patients (95%). Although there was a strong correlation between EOACont and EOAPISA (r = 0.78, P < 0.001), EOACont was statistically significantly larger than EOAPISA (0.86 ± 0.33 vs 0.75 ± 0.29 cm2, P < 0.001). Both EOACont and EOAPISA similarly correlated with GOA (r = 0.70, P < 0.001 and r = 0.77, P < 0.001, respectively). However, a fixed bias, which is hydrodynamically supposed to exist between EOA and GOA, was not observed between EOACont and GOA. In contrast, there was a negative fixed bias between EOAPISA and GOA with smaller EOAPISA than GOA. The difference between EOACont and GOA was significantly greater with a larger EOACont relative to GOA in patients with high LVOT flow velocity than in those without (0.16 ± 0.25 vs − 0.07 ± 0.10 cm2, P < 0.001). In contrast, the difference between EOAPISA and GOA was consistent regardless of the LVOT flow velocity (− 0.07 ± 0.12 vs − 0.07 ± 0.15 cm2, P = 0.936). The PISA method was applied to estimate EOA in patients with AS. EOAPISA could be an alternative parameter for AS severity grading in patients with high LVOT flow velocity in whom EOACont would potentially overestimate the orifice area.


Introduction
Aortic stenosis (AS) is the most common valvular heart disease in developed countries and its prevalence is growing with the aging population [1,2]. In clinical practice, accurate assessment of AS severity is essential for appropriate management and therapeutic decision making. Echocardiography is a key tool in determining AS severity, and the effective orifice area (EOA) assessed by the continuity equation is one of the main variables that is recommended for assessment in the guidelines [3]. Owing to its feasibility and reproducibility, the continuity equation is widely used in the assessment of AS severity; however, some potential sources of error have been identified. Among them, increased flow velocity at the left ventricular outflow tract (LVOT) due to upper septal hypertrophy (USH) was reported to cause overestimation of the stroke volume (SV) and subsequent EOA [4,5]. Because USH is common in elderly AS patients [6], an alternative method to quantify the EOA is expected to improve the diagnostic accuracy in these patients.
The proximal isovelocity surface area (PISA) method is widely used to quantify the effective regurgitant orifice area in valve regurgitation [7]. By applying the PISA method to AS, the EOA could be estimated in patients who are potentially unsuitable for applying continuity equation. In terms of the application of the PISA method for the stenotic valve, it has been used for mitral stenosis [8]; however, the feasibility and diagnostic accuracy of the PISA method for AS severity has not been reported. Accordingly, the present study aimed to validate the PISA method for the estimation of EOA in AS and to test whether the PISA method could overcome the overestimation of the continuity equation caused by high flow velocity at the LVOT in patients with AS.

Study population
We prospectively enrolled 160 consecutive patients who had been diagnosed with at least moderate AS (peak velocity across the aortic valve > 3.0 m/s or EOA < 1.5 cm 2 ) referred for transthoracic echocardiography from May 2018 to March 2019 at Hokkaido University Hospital. After the exclusion of patients with irregular heart rhythm due to arrhythmias (N = 38) and those with poor echocardiographic images (N = 8) in whom EOA by continuous equation could not be obtained, 114 patients were included in the final analysis. In 51 of the 114 patients who also underwent transesophageal echocardiography (TEE) after transthoracic echocardiographic examination (median distance: 27 days: TEE cohort), the geometric orifice area (GOA) of the aortic valve was used as the reference for the valve area. Study approval was obtained from the institutional review board of Hokkaido University Hospital (No. 018-0179) and an opportunity to opt out was given to each participant through a published disclosure document on the web site of the institute.

Two-dimensional and Doppler echocardiography
Comprehensive transthoracic echocardiography was performed in the left decubitus position using a commercially available ultrasound equipment system (Vivid E9, GE Healthcare, Horton, Norway; iE33, Philips Medical Systems, Andover, Massachusetts; Acuson SC2000 prime, Siemens Healthineers, Erlangen, Germany; Aplio Artida, Canon Medical Systems, Otawara, Japan) according to the guidelines [9]. Left ventricular (LV) mass was determined using the Devereux formula. LV ejection fraction (EF) was calculated using the biplane method of disks. USH was defined as a ratio of upper septal wall thickness to mid-septal wall thickness ≥ 1.3. The maximal aortic valve velocity (Vmax) was measured using continuous-wave Doppler imaging in the view that showed the highest flow velocity, and the mean transvalvular gradient was estimated based on the simplified Bernoulli equation. The LVOT diameter was measured in the parasternal long-axis view approximately 3-10 mm from the aortic annulus at mid-systolic phase. Flow velocity at the LVOT was measured using pulsed-wave Doppler imaging in the apical long-axis view, in which the sample volume was placed at the level where the LVOT diameter was measured and proximal to the region of flow acceleration, and the time velocity integral (TVI LVOT ) was measured. SV was then calculated as SV Cont = (LVOT diameter/2) 2 × Π × TVI LVOT . EOA of the aortic valve was calculated using the continuity equation: EOA Cont = SV/TVI AV , where TVI AV was the time velocity integral of the maximal aortic valve flow [3].

PISA method for estimation of EOA
Measurements of EOA by PISA methods were performed by the sonographers (M.M. or M.N) who were blinded to the clinical parameters before the calculation of EOA Cont . To obtain a clear visualization of the proximal convergence of the aortic stenotic jet, a cut plane was selected in each patient from the apical long-axis, apical four-chamber, right parasternal long-axis, and suprasternal long-axis views (Fig. 1A). Care was taken to minimize the angle between the centerline of the flow convergence and ultrasound beam. Because the radius of the PISA was better recognized in cine loops, the radius was carefully measured using a magnified image by selecting an optimal time phase from the series of the images. The appearance of PISA was then optimized by shifting the baseline of the color Doppler scale to forward direction to adjust for aliasing velocity around 50-70 cm/s as appropriate. The radius of PISA (r) was then measured, and the instantaneous flow rate was calculated as 2Πr 2 × aliasing velocity (mL/s). EOA was then calculated using the PISA method: EOA PISA = flow rate/Vmax (cm 2 ) based on the principle of conservation of mass (Fig. 1A). The SV was also calculated by EOA PISA × TVI AV (SV PISA ).

Measurement of GOA
Transesophageal echocardiographic images were obtained using an iE33 (Philips Medical Systems, Andover, Massachusetts) with X7-2t probe or an Acuson SC2000 prime (Siemens Healthineers, Erlangen, Germany) with Z6Ms probe. At first, the two-dimensional images were clearly obtained in the midesophageal long-axis and short-axis views. Then volume data sets were obtained using four-beat full-volume mode (N = 38, median volume rate: 27 Hz) or the live three-dimensional zoom mode (N = 13, median volume rate: 21 Hz) focused on the aortic valve complex. Fig. 1 Measurement of the PISA radius (A) and analysis of three-dimensional transesophageal echocardiographic data (B). A Color Doppler images of the apical long-axis view (upper panel) and right parasternal long-axis view (lower panel) are indicated. Proximal isovelocity surface was visualized using a zoom mode and radius of PISA (r) was measured from the sequent image series of the images. B From the threedimensional volume data set, the short-axis view of the aortic valve orifice (left panel) was adjusted using the corresponding long-axis plane (right upper panel) and coronal plane (right lower panel). The geometrical orifice area was then measured at the mid-systolic frame by tracing the inside of the calcified cusps. PISA proximal isovelocity surface area, EOA effective orifice area, EOA PISA EOA calculated using the PISA method During acquisition of full-volume images, gain and compression settings were optimized to display a magnified image of the aortic valve. After choosing the mid-systolic frame, in which maximal aortic valve excursion was observed, fine adjustments of the short-axis plane were performed to obtain the smallest aortic valve orifice and the GOA was measured (Fig. 1B). All the measurements of GOA were performed by trained cardiovascular physicians (S.I, Y.C., and S.T.) and designated by a certified reader (H.I.).

Classifications of the patients
To test the influence of LVOT flow velocity and LV morphology on the accuracy of measuring the EOA, we divided the patients by the median value of LVOT flow velocity and the presence or absence of USH and compared the differences in EOA Cont to aortic valve area-related variables between the groups.

Statistical analysis
Continuous variables are expressed as mean ± standard deviation and compared by paired or unpaired t test. Categorical data are presented as number (percentage). Agreements between aortic valve areas assessed using different methods were evaluated using Pearson's correlation coefficient and Bland-Altman plot analysis. The Bland-Altman procedure for comparing two methods was also used to determine the bias (95% confidential interval: CI). A two-sided P value < 0.05 was considered statistically significant. All statistical analyses were performed using JMP version 14.0 (SAS Institute Inc., Cary, North Carolina).

Patient characteristics
The clinical and echocardiographic characteristics of the studied patients are summarized in Tables 1. In the overall cohort, relatively young patients with AS who were predominantly female were enrolled. More than half of the patients had experienced some symptoms of heart failure. While 69% of the patients were diagnosed as hypertension, the blood pressure was well controlled at < 140 mmHg during the echocardiographic examination [10] in the majority of these patients. LV hypertrophy defined as an increased LV mass index was observed in 57% and USH was found in 46% of patients. Since the most common reason for TEE was to decide on surgical or transcatheter aortic valve replacement, the 51 patients in the TEE cohort consisted of elderly patients with a high frequency of heart failure symptoms and more advanced AS severity (Table 2), whereas the LV geometry was similar to that in the overall cohort.

Comparisons of EOA Cont and EOA PISA
According to EOA Cont , 79 patients were classified as severe AS and remaining 35 patients as moderate AS. In the 114 patients in whom the severity of AS was assessed based on EOA Cont , estimation of EOA PISA was possible in 108 patients (95%) and impossible due to inadequate image quality of PISA radius in 6 patients (5%). Optimal approaches to measure PISA radius were apical long-axis view in 62 (57%) patients, apical five-chamber view in 18 (17%) patients, right parasternal long-axis view in 18 (17%), and suprasternal  (Fig. 2A), EOA PISA was statistically significantly smaller than EOA Cont (Fig. 2B). Accordingly, 14 of 33 patients (42%) who were diagnosed with moderate AS based on EOA Cont were classified with severe based on EOA PISA . When AS was classified according to mean transvalvular gradient, and LV ejection fraction, roughly half of the moderate AS patients classified by EOA Cont were reclassified to paradoxical low gradient severe AS by EOA PISA (Fig. 3). Interestingly, the difference between EOA Cont and EOA PISA was greater in patients with USH (Fig. 2C). Further, the difference was also greater in patients with high LVOT flow velocity than in those without (Fig. 2D).

TEE cohort
In the 51 patients who underwent TEE, both EOA Cont and EOA PISA similarly correlated with GOA (r = 0.70, P < 0.001 and r = 0.77, P < 0.001, respectively) (Fig. 4A, C). However, a negative fixed bias, which is hydrodynamically supposed to exist between effective and geometric orifice areas, was not observed between EOA Cont and GOA (95% CI − 0.03 to 0.08; Fig. 4B). In contrast, there was a negative fixed bias between EOA PISA and GOA, with smaller EOA PISA than GOA (95% CI − 0.11 to − 0.03; Fig. 4D). Importantly, the difference between EOA Cont and GOA was significantly greater with a larger EOA Cont than GOA in patients with high LVOT flow velocity than in those without (Fig. 5A).
In contrast, the difference between EOA PISA and GOA was consistent regardless of the LVOT flow velocity (Fig. 5B). Similar results were observed in the comparison between patients with and without USH (Fig. 5C, D). This overestimation was clarified by larger difference to GOA in EOA Cont than in EOA PISA in the population showing USH or that showing high LVOT velocity (Fig. 5E, F). In a typical case   (Fig. 6).

Reproducibility analysis
The reproducibility for EOA PISA was assessed in 30 randomly selected patients. Two independent observers (M.N. and M.M.) analyzed the same cine loops of zoomed color Doppler images and one blinded observer (M.N.) repeated the analysis on a separate day. The intra-and interobserver intraclass correlation coefficients of EOA PISA were 0.98 and 0.95, respectively. Similarly, the reproducibility for GOA was assessed in 20 randomly selected patients using the stocked 3D volume datasets. The intra-and interobserver intraclass correlation coefficients of GOA were 0.91 and 0.83, respectively.

Discussion
In the present study, we tested the feasibility and reliability of EOA PISA in AS patients and found that (1) estimation of EOA using the PISA method was possible in up to 95% of AS patients, in whom EOA Cont could be obtained, with acceptable consistency with EOA Cont , (2) EOA PISA was as well correlated with GOA as EOA Cont was, and importantly, (3) the concordance between EOA PISA and GOA was consistent in patients with high LVOT flow velocity and those with USH in whom EOA Cont potentially overestimated the valve orifice area. Although the PISA method has been applied for other valvular heart diseases [11][12][13][14], this is the first study to demonstrate the feasibility and reliability of this method in AS.

PISA method for assessment of AS severity
In the same manner as the continuity equation, the PISA method is based on the principle of conservation of mass, where flow rate during the ejection period through the proximal isovelocity surface is equal to that through effective orifice area of the narrowing orifice. The physical basis of this method is quite solid, and use of the PISA method to quantify the severity of mitral regurgitation, mitral stenosis, and aortic regurgitation is recommended by the guidelines [15, 16]. However, the use of PISA method for AS has not been mentioned probably because of the spread and retention of EOA Cont . In the present study, we demonstrated an acceptable feasibility and reproducibility of EOA PISA in clinical AS patients.

EOA Cont for assessment of AS severity
Transvalvular gradients and EOA Cont are the principal measures for the echocardiographic grading of severity in AS [3]. Among them, estimation of EOA is important because transvalvular gradients vary depending on the flow conditions [17]. An accurate estimation of EOA Cont is, thus, mandatory for managing patients with AS; however, sources of errors have been noted in the measurement of its components [18,19]. As the LVOT cross-sectional area is calculated based on the anterior to posterior diameter assuming a circular shape, it can underestimate the area, which has an elliptical shape in most patients [20][21][22][23][24]. At the same time, a study showed that TVI LVOT was overestimated using pulsed-wave Doppler, resulting in compensation for the underestimation of the LVOT area with subsequent good concordance between EOA Cont and cardiac magnetic resonance-derived EOA [5]. This overestimation of TVI LVOT was thought to be due to the alteration of the flow velocity profile from flat to skewed in the presence of high flow velocity at the LVOT, as seen in USH [5,25]. Koto et al. [4] demonstrated that AS patients with USH had greater SV and EOA calculated using the Doppler method than those without. Consistent with these previous findings, when SV was compared between patients with and without USH or those with and without high LVOT flow velocity, apparent differences were observed only in SV Cont but not in SV PISA between groups (Supplemental Table). Therefore, the significant differences in EOA Cont and EOA PISA (Fig. 2C, D) and those in EOA Cont and GOA (Fig. 5A, C) between patients with and without USH or those with and without high LVOT flow velocity, along with the absence of a negative fixed bias between EOA Cont and GOA, suggests overestimation of EOA Cont in patients with high LVOT flow velocity.

Clinical implication
Owing to the increasing elderly population with AS, a substantial population of the patients could show a narrowing LV cavity with USH, resulting in increased LVOT flow velocity [26,27]. As observed in a previous study [4] and ours, the continuity equation would overestimate SV and subsequent EOA in these patients. In typical cases with USH or LVOT obstruction, it is generally impossible to avoid sampling high flow speed by placing the sample volume at anywhere of the proximal site of the aortic valve annulus due to the turbulent flow. In that case, EOA PISA could be an alternative parameter to determine AS severity. EOA PISA is expected to be particularly effective in patients showing EOA Cont that is larger than expected from the transvalvular flow velocity as shown in Fig. 6. As the diagnosis of severity is critical in AS [3], many reliable and feasible variables are needed for its assessment. Based on the additional time of approximately 50 s in other AS population for measurement of EOA PISA (although we did not record the time in the present population), we believe that EOA PISA is in line with other recently proposed markers of AS severity [4,27,28].

Study limitations
This study had several limitations. First, because of the cross-sectional nature, this study could not show prognostic superiority of EOA PISA and thus the present results need to be interpreted with a caution. Second the relatively small number of studied patients limits the strength of our findings; thus, the results of our study should be confirmed in a larger study. Third, test-retest reproducibility between the two examinations was not assessed, although the consistency of the EOA PISA obtained from the same cine loops was acceptable. Finally, while the PISA method assumes the shape of the proximal isovelocity area to be hemispherical, we could not confirm its exact geometry in AS. As noted in a recent computational simulation [15], an ellipsoidal PISA geometry results in substantial underestimation. In addition, based on the funnel shape geometry of the stenotic orifice, PISA might require correction based on an angle such as the PISA method in mitral stenosis. However, shape of stenotic orifice in AS, especially in sclerotic degenerative valves, is symmetrical and far from funnel shape compared to that of mitral stenosis, which might have reduced the alteration of PISA geometry in AS. In fact, the EOA PISA was 10-30% smaller than GOA in our study, which was within the previously reported difference ranges between EOA and GOA (10-30%) [29], and can justify the application of PISA method without angle correction for AS.

Conclusions
In conclusion, the PISA method was applied for the estimation of EOA in patients with AS. EOA PISA can be used as an alternative variable when assessing the AS severity in patients with high LVOT flow velocity or USH in whom EOA Cont would potentially overestimate the orifice area.