The limited usefulness of real-time 3-dimensional echocardiography in obtaining normal reference ranges for right ventricular volumes

Right ventricular (RV) function is an important prognostic factor in both congenital and acquired heart disease [1]. In clinical practice the assessment of right ventricular dysfunction is important to a variety of conditions [2]. As early as 1982 an attempt was made to determine RV volume from the apical window by 2-dimensional echocardiography (2DE) using the Simpson's biplane method [3]. In that study, however, only "body" volumes were obtained, reflecting inflow tract dimensions which represented 55% of the total RV volume as obtained by RV angiography. In order to compensate for these problems, Levine et al. [4] incorporated a combination of apical four-chamber (for inflow) and subcostal views (for outflow). This approach, however, has not gained widespread acceptance within clinical practice. According to present guidelines, assessment of RV size is best performed in the apical 4-chamber view, with reference limits provided for both RV diastolic and systolic areas and for RV area fraction (AF) [5]. The introduction of real-time 3-dimensional echocardiography (RT3DE) has provided promising results for RV volume measurements from one single modified apical view using offline analysis with Tomtec^® 3D software by disc summation [6].

The recent introduction of RT3DE with fast online analysis may allow bedside measurements of RV volumes and ejection fraction (EF). The purpose of the present study was to obtain normal reference values for RV volumes and EF with RT3DE parallel to reference ranges for 2DE-derived areas corrected for body surface area (BSA). In addition, a blinded and unblinded reproducibility study of the two methods was carried out.

The present study has provided gender-specific normal reference ranges for both RV volumes and EF measured with RT3DE and 2DE-derived RV areas and AF corrected for BSA. In addition, we observed a negative correlation between age and RV dimensions with RT3DE, but not with 2DE. The intraobserver variability, however, was poor with blinded measurements using both methods, indicating that follow-up measurements of individual patients should be performed after careful inspection of previous loops and tracings of the RV endocardium. This dataset has been obtained from a large population of healthy volunteers, each of whom was carefully screened for disease, including subclinical cardiac disorders.

During the progress of this study, the outflow tract, including the pulmonary artery valves, was not visualised in any of the examinations. This corresponds with the problems reported in previous 2DE studies using the apical approach [3, 4].

The notion of an underestimation of RV volumes in our study is verified from reference values in studies incorporating offline analysis allowing inclusion of the RV outflow tract. In Gopal et al's. [6] study of 71 healthy individuals upper normal values for RVEDVI were 100 ml/m² in males and 92 ml/m² in females. Their respective references for RVESVI were 53 ml/m² and 51 ml/m². Their lower limits for RVEF; 30% in males and 38% in females; are, however, comparable to our data. The underestimation of RV volumes in our study may therefore have been of a similar magnitude in end-diastole and end-systole, resulting in meaningful RVEF measurements.

RT3DE using off-line analyses with dedicated software (e.g. Tomtec^®) seem to underestimate RV volumes to a less extent compared with cardiac magnetic resonance (CMR) than the methodology used in our study [10, 11].

In Kjaergaard et al's. [12] study of 54 healthy volunteers upper normal RVEDVI was 88 ml/m² in males and 80 ml/m² in females. The respective limits for RVESVI were 46 ml/m² and 38 ml/m². There was a tendency for smaller volumes to be associated with increasing age. The lower references for RV volumes may be related to differences in methodology, since Kjaergaard et al. used disk summation from a short axis view whilst Gopal et al. used an apical view.

CMR is regarded as the best measure of cardiac dimensions. In Gopal et al's. study [6], CMR derived max RVEDVI was 97 ml/m², max RVEDVI 53 ml/m² and min RVEF 35%. These values are quite similar to their RT3DE derived references. Maceira et al. [9] studied 120 healthy volunteers comprising of 10 females and 10 males with CMR in each decade between 20 and 80 years. Their upper normal values were 100 ml/m² for RVEDVI and 41 ml/m² for RVESVI, whilst the lower limit for RVEF was 54%. BSA corrected RV volumes were larger in males and a significant negative correlation between age and RV volumes was found. RVEF did not differ between genders but was positively correlated with age.

Hudsmith et al. [13] studied 108 healthy volunteers with CMR aged between 21 and 68 years. Upper normal values were 123 ml/m² for RVEDVI and 56 ml/m² for RVESVI, whilst lower normal RVEF was 49%. BSA corrected RV volumes were larger in males than females and tended to be smaller in participants aged >35 years.

Within these studies there are considerable inter- and intra- methodological discrepancies between upper reference values for RV volumes, despite the fact that all measurements claim to have included both RV inflow and outflow tracts. Especially concerning is the large discrepancy between the lower normal limit for RVEF when derived from RT3DE as opposed to CMR. All studies, including ours, emphasize the need for gender-specific reference values corrected for BSA. The data supporting age-related corrections are not convincing. The correlations between age and RV volumes in our study were so weak that such a correction is not deemed to be necessary.

Our 2DE results also emphasise the need to use reference values for RV areas corrected for age and BSA. The normal ranges presented in the ASE/EAS guidelines are obtained from a group of 72 adult patients aged from 15 to 76 (mean 38) years with a mean BSA of 1.75 (range 1.4 – 2.1) m² [14]. For RV areas, however, only 41 participants were included. BSA corrected reference values are not presented. Max RVAEDI (mean +2SD) was 24 cm² and max RVAESI 17 cm², whereas minimum RVAF (mean -2SD) was 31%. In our study the respective values were 31 cm², 18 cm² and 27%. Since the measurement techniques were quite similar, the discrepancies between the upper normal diastolic areas and RV AF may be related to the differences in the study populations. These differences might have been smaller if BSA corrected values had been available as in Weyman et al's. study [14].

The blinded reproducibility study indicates unacceptable intraobserver variability with both methods. To optimize reproducibility the unblinded approach should be introduced into routine clinical practice. The intraobserver variability of 2DE derived areas was slightly better than that of RT3DE derived volumes in the unblinded study. In addition, the interobserver variability for the 2DE method was of clinical importance, emphasizing the need for repeated unblinded measurements performed by the same investigator during follow-up examinations of RV dimensions. Due to anticipated learning curve problems, we chose to use only one experienced investigator to perform the RT3DE recordings and measurements.

RT3DE with automatic border detection and fast online analysis is hampered by the underestimation of RV volumes when compared with MRI and RT3DE with offline analysis. The RVEF data presented may be more realistic for RV function since the error in volume determination may be equally pronounced in ED and ES. The serial reproducibility is only acceptable if previous recordings and tracings are available. 2DE measurements of RV areas and AF are simple to perform and are fairly accurate and reproducible provided the unblinded approach is practiced. Both methods are suitable as a bedside tool for immediate clinical decisions. The present study has expanded the knowledge of normal RV dimensions and contractility obtained with apical RT3DE and 2DE recordings from a large series of individuals, allowing the presentation of BSA and gender corrected values.