We enrolled consecutive patients referred to our Radiology Unit between March 2018 and June 2018 for suspected valvular heart disease. We stratified the patient population according to AS severity: mild (1.2 cm²/m² ≥ indexed EOA > 0.85 cm²/m²), moderate (0.85 cm²/m² ≥ indexed EOA ≥ 0.6 cm²/m²), or severe AS (indexed EOA < 0.6 cm²/m²). Each patient underwent TTE and CMR examinations on the same day.

Exclusion criteria were as follows: (i) age <18 years; (ii) left ventricular ejection fraction <50% by CMR; (iii) bicuspid aortic valve; (iv) at least moderate multi-valvular disease; (v) hemodynamic instability; (vi) atrial fibrillation or frequent supraventricular or ventricular premature contractions; and (vii) poor TTE image quality.

The study was conducted according to the criteria set by the Declaration of Helsinki. Patient's informed consent was obtained from all individual participants included in the study. All patients underwent a standard clinical assessment including medical history, physical examination, routine blood tests, transthoracic echocardiogram, and CMR.

CMR and TTE examinations were performed on the same day by four physicians accredited in each respective imaging modality, without knowledge of the results of the complimentary exam. Imaging and acquisition protocols were those recommended by the respective professional societies (15, 16). We performed the following measurements (Figure 1): (a) effective orifice area (EOA_echo) by TTE continuity equation; (b) geometric aortic valve area by CMR (GOA_cmr); (c) hybrid effective orifice area (EOA_hybrid) as the continuity equation using LVOT area measured by CMR and LVOT and aortic velocities measured by pulsed-wave (PW) and continuous wave (CW) Doppler echocardiography, respectively; and (d) aliased orifice area (AOA_cmr) planimetry using a low-VENC phase-contrast CMR sequence.

All patients had TTE by two accredited cardiologists (S.G. and E.D.) using a commercially available ultrasound system (Esaote My Lab 50 Gold, ESAOTE, Genoa, IT) with 1–5 MHz transducers, according to an established protocol (15). EOA_echo was calculated by the continuity equation according to the 2017 EACVI/ASE recommendations (3). TTE image quality was measured by the Image Quality Assessment Tool and poor quality defined by a score of <10 (17).

All patients were examined on Achieva 1.5-T scanner (Philips Medical System, Best, the Netherlands) using a dedicated eight-channel phased-array cardiac synergy coil for signal reception during end-expiratory breath holds. CMR cine assessment with steady-state free-precession sequences (SSFP) provided both morphological and functional data, including left and right ventricular volumes and ejection fraction.

All CMR studies were analyzed off-line using a dedicated semi-automated workstation and post-processing software (MR WorkSpace 2.6.3.2, Philips Medical Systems, Nederland B.V.). All measurements were performed off-line by two independent investigators (FR and CM) blinded to clinical and TTE results. We assessed left and right ventricle volumes and function, analyzed the velocity-encoded images and measured aortic valve planimetry on bSSFP and phase images of PC sequences, and LVOT diameters on bSSFP sequences. Geometric valve area and LVOT area were measured in mid-systole when the valve cusps reached the maximum degree of opening, the transvalvular flow was the widest, and LVOT had the largest diameters.

Continuous variables were expressed as mean ± standard deviation (SD) and categorical variables as counts and percentages. Categorical variables of patients in different groups were compared using Chi-squared test, and continuous variables were compared among groups using Student's t-test for unpaired data. Individual differences among groups were compared post hoc using Mann–Whitney U-tests. Correlation between EOA_hybrid and EOA_echo, GOA_cmr, and AOA_cmr was evaluated by ordinary least-square linear regression. Agreement between different imaging modalities was assessed by Bland–Altman analysis. To assess inter-observer variability, the measurements of GOA_cmr and AOA_cmr were repeated in all patients by two blinded observers. To further evaluate intra-observer variability of GOA_cmr and AOA_cmr, sequences were analyzed twice by the two observers. Intra- and interobserver reliability was computed by assessing absolute mean differences and using intraclass correlation coefficient (ICC).

A p-value of <0.05 was considered statistically significant. A sample size of 28 patients was calculated considering two observations per patient, with a p-value <0.05; statistical power of 80%, expected reliability of 0.80, and acceptable reliability of 0.50.

The mean values of LVOT_cmr area were significantly larger than the LVOT_echo both in patients with AS (mild, moderate and severe; p < 0.001) and in control group patients (p < 0.001) (Table 2).

We observed excellent pairwise positive linear correlations among AOA_cmr, EOA_hybrid, GOA_cmr, and EOA_echo (p < 0.001); AOA_cmr had the best correlation with EOA_hybrid (R² = 0.985, p < 0.001). There was also an excellent correlation between AOA_cmr and EOA_echo (R² = 0.975) (Figure 3). Bland–Altman analysis (Figure 4) demonstrated a good agreement between different techniques, with the lowest bias of 0.019 for the comparison between EOA_hybrid and AOA_cmr.

Figure 5 shows spider diagrams of controls and AS patients, differentiated on the basis of the AS severity. This graphic representation allows immediate comparison of the data obtained for each individual patient. It provides a visual display confirming that GOA_cmr values were significantly higher than corresponding values of EOA_echo in all degrees of stenosis (p < 0.001) as well as in controls (p < 0.001). On the other hand, the TTE method (EOA_echo) underestimates the valve area compared to the hybrid method (EOA_hybrid), because of different values for LVOT area used in the continuity equation, while AOA_cmr values are closer to the EOA_hybrid values.

A reclassification analysis was conducted to determine the number of patients whose AS severity was reclassified by using AOA_cmr, GOA_cmr, and EOA_echo compared with grading provided by EOA_hybrid. Compared with EOA_hybrid, AOA_cmr reclassified 3/28 (11%) subjects (Figure 6): two patients moved from the moderate to the severe category, with one patient moving from severe to moderate AS. EOA_echo and GOA_cmr reclassified, respectively, 6/28 (21%) and 9/28 (32%) subjects. When using EOA_echo as the reference, AOA_cmr underestimated AS severity in three patients (11%).

GOA_cmr and AOA_cmr yielded excellent intra-rater and inter-rater reliability both in the overall population (GOA_cmr, absolute mean difference: 0.07 and 0.09; ICC: 0.995 and 0.991; respectively; AOA_cmr: absolute mean difference 0.08 and 0.08; ICC: 0.997 and 0.998; respectively) and in the severe AS subgroup (GOA_cmr, absolute mean difference: 0.08 and 0.08; ICC: 0.997 and 0.994; respectively; AOA_cmr: absolute mean difference 0.07 and 0.08; ICC: 0.998 and 0.998; respectively).

TTE currently is the reference method for the diagnosis and quantification of AS. It is completely non-invasive and relatively simple to perform and allows an accurate diagnosis in the majority of patients (1, 4). The quantitative evaluation of AS relies on Doppler measurement of transvalvular gradients and the derivation of valve area. Both sets of metrics correlate with symptoms, need for valve intervention, and survival. TTE often underestimates LVOT area. Maes et al. showed that use of LVOT area by CMR into the continuity equation resulted in a 29% increase in the SVi and AVAi, leading to reclassification from echo-severe to CMR-moderate AS in 25% of patients (18). This is a consequence of the elliptical shape of the LVOT cross-sectional area, which is ignored by 2D TTE.

When echocardiographic parameters are discordant, inconclusive, or not accurate, CMR has proven to be a robust alternative imaging modality for grading AS severity by direct aortic valve orifice planimetry (19–22). Balanced SSFP sequences allow assessment of the anatomy and function of the AV, allowing a direct planimetry of the valve orifice (21). However, direct planimetry of the AVA can be challenging due to heavy calcifications of the tips, with image artifacts and partial volume effects. CMR also provides other relevant information including morpho-functional assessment of cardiac chambers, detection and quantification of other valvular lesions, tissue characterization, and extracardiac findings (23).

By fundamental hydrodynamic principles (the flow contraction phenomenon), the GOA is usually significantly larger than the EOA (10). This is not surprising, since direct planimetry reflects the anatomical orifice area, while the calculated EOA reflects the functional orifice area, measured after jet contraction downstream from the orifice. An in vitro study using stenotic bioprosthetic valves showed that an EOA of 1 cm² corresponds to a GOA of 1.2 cm², suggesting that a GOA measured with planimetry between 1 and 1.2 cm², by either CMR imaging, computed tomography, or transoesophageal echocardiography, should not discard the possibility of severe AS.

CMR has emerged as an alternative method for non-invasive estimation of EOA with the use of velocity-encoded CMR techniques, employing the continuity equation. CMR allows a direct measure of LVOT area avoiding its echocardiographic underestimation. However, CMR underestimates velocities compared to Doppler. This underestimation has been attributed to intravoxel dephasing, loss of signal, and lower temporal resolution than echocardiographic Doppler-based methods (10).

As mentioned earlier, accuracy in quantifying AS could be improved by the use of a multimodality metric, EOA_hybrid, combining strengths and avoiding the pitfalls of both TTE and CMR. The continuity equation can be obtained by using LVOT area derived from CMR and velocities measured by Doppler TTE (11–15). Although conceptually attractive, this approach is cumbersome and not easily applicable in clinical routine, as it would require a costly sequential acquisition of two different diagnostic tests possibly under the same hemodynamic conditions.

To overcome the aforementioned limitations, we are reporting a new, simple CMR method for measuring EOA directly, based on phase contrast velocity mapping of the aliased orifice area. The VENC (velocity-encoded value) was intentionally low (50 cm/s) in order to enhance aliasing. The transvalvular aliased flow surface area, which we termed aliased orifice area (AOA_cmr), is conceptually identical to the EOA. EOA is indeed equivalent to the cross-sectional area of the vena contracta of the transaortic flow jet. Despite the modest sample size of the study population, we obtained significant results demonstrating that AOA_cmr is a feasible, precise, and reproducible method to measure EOA by CMR.

We compared EOA_hybrid, conceptually the “gold standard” for measuring EOA, with other measurement techniques (EOA_echo, GOA_cmr), including our new proposed index (AOA_cmr). AOA_cmr proved the most accurate parameter: it had high linear correlation and excellent agreement with EOA_hybrid, as well as a good inter- and intra-rater variability.

There are several advantages to this method: (i) single direct measurement (with fewer sources of error, and thus more precise and reproducible than the continuity equation) from a single free-breathing sequence with no need for scout sequences to set any “correct” VENC (one-fits-all approach with fixed low VENC); (ii) purely non-invasive (no contrast required); (iii) good intra- and inter-rater reproducibility; (iv) simultaneous evaluation of effective regurgitation orifice area in mixed aortic valve disease; and (v) “Doppler-like” visual inspection of the aortic valve plane (i.e., detection of peri-valvular/prosthetic leak, cusp perforation, and central or eccentric origin of the regurgitant jet).

Possible disadvantages include the following: (i) low-quality images in case of poor ECG triggering, with potentially low applicability to patients with arrhythmias; (ii) intravoxel dephasing; (iii) low temporal and spatial resolution; (iv) 2D measurement; (v) unknown dependency on flow rate and magnet field intensity; and (vi) longer velocity mapping sequence due to high number of excitations (five averages: ~2 min).

We acknowledge a few limitations that must be addressed. Firstly, the small sample size of this pilot study prevented us from including a more heterogeneous population to assess various hemodynamic subsets of AS. Secondly, we used a 2D phase contrast sequence that is sensitive to the through-plane motion of vena contracta. Thirdly, we used an arbitrary very low VENC value of 50 cm/s; however, the selected threshold is lower than two standard deviations below normal mean aortic valve peak velocity by CMR (i.e., 80 cm/s) (24), which makes this a reasonable cut-point to achieve aliasing at the level of the aortic valve. Fourthly, we cannot foresee the influence of field strength, gradient echo sequences, and prospective triggering on the accuracy and precision of AOA. Further studies are warranted to assess reliability and accuracy of AOA across different hemodynamic patterns of AS and different VENC values, and to explore the feasibility of 3D vena contracta area by novel highly accelerated compressed sensing 4D and 5D respiratory-motion resolved flow sequences.