Functional mitral regurgitation (FMR) is a poor prognostic factor in patients with dilated and ischaemic cardiomyopathy.1^–3 Geometric alterations of the left ventricle and mitral valve apparatus, including left ventricular dilatation, mitral valve tethering/tenting and annulus remodelling, have been reported to determine the severity of FMR.4^–15 The proximal isovelocity surface area (PISA) method, based on the assumption that PISA shape should be hemispherical, has been regarded as a clinically useful tool for quantification of FMR.16 In some cases, however, PISA appears to be ellipsoidal, and two separate FMR jets have been also demonstrated in patients with ischaemic cardiomyopathy by two-dimensional (2D) echocardiography.10 2D echocardiography has limitations in evaluating FMR, because it cannot visualise the entire shape of FMR PISA and FMR features are completely dependent on the imaging planes. Since the coaptation line of the mitral valve is curved along the margin of the anterior and posterior leaflets, an elongated ellipsoidal PISA might appear like two separate PISAs on 2D echocardiography. However, three-dimensional (3D) analysis of FMR still remains to be clearly demonstrated. We have therefore evaluated the 3D features of FMR and their geometric determinants by 3D analyses of PISA and geometry of the left ventricle and mitral valve apparatus using real-time 3D echocardiography and 3D colour Doppler image.

METHODS

Study population

Fifty-two patients (mean age, 60 (SD 13) years, 20 females) with severe left ventricular systolic dysfunction (ejection fraction ⩽35%) and sinus rhythm were enrolled. Mean left ventricular ejection fraction was 25% (6%). Severe left ventricular dysfunction was caused by ischaemic cardiomyopathy in 17 patients (33%) and non-ischaemic cardiomyopathy in 35. All patients had FMR persisting throughout the entire systolic period. Exclusion criteria were organic mitral valve diseases, a vanished PISA in mid-systole phase, dyspnoea precluding breath holding for image acquisition, atrial fibrillation, implantation of a pacemaker or defibrillator and poor image quality for accurate geometric analysis.

Real-time 3D echocardiography

Image acquisition

A real-time 3D echocardiography system (Sonos 7500, Philips Medical Systems, Andover, MA, USA) with a 2–4 MHz 3000-element xMATRIX transthoracic transducer was used to acquire 3D full-volume and colour Doppler images. Images were obtained from the apical views with gain, compression controls and time gain compensation settings optimised for image quality. Care was taken to include the entire left ventricle and mitral valve apparatus in the full-volume dataset. Full-volume datasets were acquired in the wide-angled acquisition (93°×80°) mode, in which four wedge-shaped subvolumes (93°×20° each) were obtained from four different cardiac cycles during held respiration. Acquisition was triggered to the R wave of every other cardiac cycle to allow time for storage of each subvolume, resulting in a total acquisition time of eight heartbeats. Three-dimensional colour Doppler image volumes were acquired as seven subvolumes gated to form a full Doppler volume (30°×30°). The aliasing velocity of the colour Doppler was set between 35 cm/s and 45 cm/s, and the entire FMR PISA and mitral annulus were included in the colour Doppler image set. All images were stored digitally and analysed offline.

Image analysis

QLab (Philips Medical Systems, Andover, MA, USA) software was used for PISA analysis. The medial, central and lateral cross-sectional antero-posterior planes, perpendicular to the commissure-commissure plane, were generated (fig 1). Images on the antero-posterior planes were magnified appropriately, and PISA radii were measured on the medial, central and lateral planes at a mid-systole frame when the mitral valve tenting was smallest. The largest PISA radius among those three planes was defined as PISAmax. Eccentric PISA was defined as PISAmax not located on the central plane, while central PISA was defined as PISAmax on the central plane. Both-sides dominant PISA was defined as the appearance of two major PISAs on the medial and lateral planes and no or small PISA on the central plane. Separate PISAs were defined as multiple interrupted PISAs when no PISA was observed between the two PISAs while the antero-posterior plane was moved from the medial to the lateral end of the mitral valve.

Geometric analyses of the left ventricle and mitral valve were performed using Tomtec software (Munich, Germany). Left ventricular end-diastolic and end-systolic volumes were measured, and ejection fraction was calculated. The commissure-commissure and central antero-posterior planes were obtained at mid-systolic frame when the mitral valve tenting was smallest. Left ventricular spherical index was defined as the medial to lateral length divided by the mitral annulus to apex length on the commissure-commissure plane. Commissure-commissure and antero-posterior mitral annulus diameters were measured (fig 2). Mitral valve tenting height and tenting area, tenting angles of the anterior and posterior leaflets, bending angle of the anterior leaflet angulation, anterior leaflet proximal and distal lengths from the angulation point and posterior leaflet length were measured on the central antero-posterior plane.

Two-dimensional colour Doppler echocardiography

Two-dimensional colour Doppler echocardiography was performed on the same day as real-time 3D echocardiography. The aliasing velocity was set at 35–45 cm/s on colour Doppler image, and the PISA radius was measured from the apical view at a mid-systole frame when the mitral valve tenting was smallest.

Statistical analysis

Statistical analyses were performed using software (SPSS 12.0, SPSS Inc, Chicago, IL, USA). Data were expressed as mean (SD). Group comparison of continuous variables was performed using the unpaired Student t test. Comparison of the ratio between two groups was performed by the Fisher’s exact test. Receiver-operating characteristic curve analysis was performed to test the diagnostic power of each parameter and determine the best cut-off values for the identification of specific 3D PISA features. Bland-Altman analysis was performed to evaluate the difference between PISA measured by 2D colour Doppler echocardiography and PISAmax determined using real-time 3D echocardiography. Univariate linear regression analysis was used to evaluate the correlation between those two continuous variables. A p value <0.05 was considered significant.

RESULTS

Eccentric PISA was observed in 29 patients (56%, representative examples in fig 3C, D, E), where as central PISA was observed in 23 patients (fig 3A, B). Of the 29 patients with eccentric PISA, 18 showed both-sides dominant PISA (fig 3C, D). Separate multiple PISAs were present in 14 patients (27%, fig 3C) including 11 with eccentric PISA and 3 with central PISA, whereas single PISA was present in 38 patients (fig 3A, B, D, E). Of the 14 patients with separate PISAs, 10 (71%) had two PISAs on the medial and lateral planes.

The only geometric variable that differed significantly between patients with eccentric and central PISA was the anterior leaflet distal length from the angulation point, which was longer in patients with eccentric PISA (table 1). An anterior leaflet distal length from the angulation point ⩾1.70 cm differentiated eccentric PISA with a sensitivity of 62% and a specificity of 78% (fig 4, left).

Patients with both-sides dominant PISA showed longer anterior leaflet distal length from the angulation and smaller anterior leaflet bending angle than other patients (table 2). An anterior leaflet bending angle ⩽140° identified both-sides dominant PISA with a sensitivity of 61% and a specificity of 77% (fig 4, centre).

Patients with multiple separate PISAs tended to have higher left ventricular ejection fraction and smaller antero-posterior mitral annulus diameter than those with single PISA, but the difference was not significant (table 3). The former, however, had significantly smaller mitral valve tenting height, tenting area and PISAmax than the latter. PISAmax ⩽0.40 cm differentiated separate PISAs with a sensitivity of 79% and a specificity of 50% (fig 4, right).

In all patients, there was a good correlation between PISA measured using 2D colour Doppler echocardiography and PISAmax determined using 3D colour Doppler echocardiography (r = 0.781, p<0.001, fig 5). However, the former was significantly smaller than the latter (−0.06 (0.11) cm, p<0.001), while the aliasing velocities used for PISA measurements were not different between 2D and 3D colour Doppler echocardiography studies (41 (5) vs 40 (4) cm/s, p = 0.45).

DISCUSSION

Functional mitral regurgitation in patients with ischaemic and dilated cardiomyopathy is caused by 3D geometric alterations of the left ventricle and mitral valve apparatus, such as left ventricular dilatation,11 12 14 15 mitral valve tethering,4 5 10 13 and mitral annulus remodelling.6^–8 Because of the pathogenesis of FMR, not only FMR regurgitant orifice size but also orifice shape should be determined by geometric changes of the left ventricle and mitral valve apparatus.17 Although FMR is generally believed to stem from the entire mitral valve coaptation line, the evidence for two separate jets originating from the medial and lateral sides in FMR has been reported by in vitro experiment and clinical 2D colour Doppler study.10 17 It is difficult, however, to confirm the 3D configuration and main origin of FMR by 2D colour Doppler imaging, because of the dependence on the imaging plane.

Using real-time 3D colour Doppler imaging, we have demonstrated that 3D PISA shapes, which must be determined by regurgitant orifice shapes, are quite diverse in FMR patients with severe left ventricular dysfunction. More than 50% of our study population showed PISAmax out of the central plane, and 35% and 27% showed both-sides dominant PISA and separate multiple PISAs, respectively. A longer anterior leaflet distal length from the angulation point was associated with eccentric PISA, while a longer anterior leaflet length from the angulation and more acute anterior leaflet-bending angle were associated with both-sides dominant PISA. Other geometric variables of the left ventricle and mitral valve apparatus and the aetiologies of the left ventricular dysfunction were not associated with these specific PISA shapes. These results suggest that the regurgitant orifice shape in FMR is determined by the configuration of anterior mitral valve angulation. Because anterior leaflet angulation is mainly determined by secondary chordae, insertion sites of the secondary chordae on the anterior mitral valve leaflet and major tethering force/direction on the secondary chordae may determine regurgitant orifice shape in FMR. This suggestion is supported by an animal study demonstrating the importance of secondary chordae in generating FMR,18 and an in vitro study showing that anterior leaflet remodelling caused by chordal force distribution generated a non-uniform regurgitant orifice area.19

We also demonstrated that multiple separate PISAs are associated with smaller tenting of the mitral valve, and that PISAmax of multiple separate PISAs are smaller than PISAmax of single PISA. These results suggest that small tenting in FMR tends to generate multiple small regurgitant orifices. Most of these patients showed separate PISAs on the medial and lateral sides of the mitral valve, a finding consistent with an in vitro study showing that less mitral regurgitation resulted from both-sides regurgitant orifices than from the central orifice.17

Clinical implications

The severity of FMR may be underestimated by quantification using conventional 2D colour Doppler PISA, because only a part of PISA may be visualised on the 2D plane. The findings reported here show that maximal PISA radius was underestimated by 2D colour Doppler compared with real-time 3D colour Doppler echocardiography, perhaps because the 2D imaging plane may deviate from the plane of the largest PISA. Therefore, compared to 2D colour Doppler, real-time 3D colour Doppler echocardiography can provide more accurate 3D information about PISA of FMR. Furthermore, even if the 2D imaging plane is properly positioned at the PISAmax, the PISA surface area may be underestimated by 2D PISA radius measurement and geometric hemispherical assumption in patients with separated or elongated PISA. The 2D PISA quantification method has been shown to underestimate elongated elliptic regurgitant orifice areas.20 Therefore, in patients with FMR evaluated by 2D PISA quantification, the possibility of eccentric, both-sides dominant or separate PISAs should be considered. The geometric determinants and cut-off values determined here may provide clues to identify these specific features of PISA.

Our results emphasise the importance of 3D PISA quantification in FMR patients, because a substantial proportion of patients did not show the single hemispherical shape of PISA, as assumed in the 2D PISA quantification method. Sitges et al demonstrated by an animal experiment that the 3D measurement of PISA could provide reliable quantification of mitral regurgitation.21

Study limitations

We used different software programs for 3D colour Doppler analysis and geometric analyses of the left ventricle and mitral valve, owing to the advantages and limitations of these specific 3D echocardiography software programs, including colour Doppler display image quality and availability of angle measurements. Since we used a colour Doppler image set and a full-volume image set for these analyses, we could not completely exclude their possible discordance. Because of the dynamic change in PISA during the systolic phase,22 23 we selected a mid-systole frame when the mitral tenting was smallest. However, we could not confirm that the same frame was selected for analysis in both image sets.

Predictive values of geometric variables for specific types of 3D PISA shapes were not very good and there were substantial overlaps. These results suggest that the regurgitant orifice shape of the FMR may not be determined entirely by a single geometric factor and may be influenced by multiple 3D geometric factors and their interactions.

Finally, we could not quantify 3D PISA surface area and FMR severity, because of the absence of a non-invasive ideal gold standard method to compare. Further clinical studies of 3D PISA quantification involving more invasive quantitative verification in FMR patients may be necessary to verify the accuracy of this method and its clinical usefulness.

CONCLUSIONS

Using 3D colour Doppler imaging, we have demonstrated that the 3D features of FMR are quite diverse. The shape and site of the anterior leaflet bending determine the shape of the regurgitant orifice, and small mitral valve tenting generates separate small regurgitant orifices of FMR in patients with severe left ventricular dysfunction.