2.1 Study population

A group of 143 healthy volunteers were prospectively recruited from May 2021 to July 2021. The inclusion criteria were as follows: age older than 18 years; body mass index < 30 kg/m²; normal physical examination results; normal electrocardiogram results; normal 2D echocardiography results; and absence of any cardiovascular or respiratory diseases. The Organizational Ethics Committee approved the protocol, and all participants provided informed consent before undergoing the examinations.

2.2 Echocardiographic acquisition

Transthoracic echocardiography was performed using a Vivid E95 scanner (GE Vingmed Ultrasound, Norway) according to the recommended protocols.^{9, 10} The 2D echocardiography images were acquired using an M5Sc probe, and the four-dimensional (4D) echocardiography RV images were acquired using a 4Vc probe. All electrocardiogram-triggered echocardiography images were obtained over three to five consecutive cardiac cycles during breath holding. Datasets were digitally stored and analyzed offline using EchoPAC v204 (GE Vingmed Ultrasound).

2.3 Echocardiographic measurements

The RV volume and RV ejection fraction were obtained using a software package (4D Auto RVQ). The TAPSE, RV FAC, and RV index of the myocardial performance according to pulsed-wave Doppler (PD) and TD imaging (PD RV myocardial performance index [RMPI] and TD RMPI, respectively) and peak systolic velocity of the tricuspid annulus according to pulsed-wave TD imaging and color TD imaging (TA pulsed-wave TD S′ and TA color TD S′, respectively) were measured according to the current guidelines.⁵ TA isovolumetric acceleration (TA IVA) was measured by color TD imaging.¹¹ In the RV-focused apical four-chamber view, RV GLS was evaluated by tracing the RV free wall and interventricular septum.¹⁰ The tricuspid flow pattern with E and A wave velocities were obtained with the sample volume positioned at the tricuspid leaflet tips. The tricuspid deceleration time was defined as the time from the top point to the end point of the E wave. Isovolumetric relaxation time (IVRT) was measured at the tricuspid lateral annulus using pulsed-wave TD imaging.¹² TA velocities according to TD imaging included early and late diastolic annular velocities (E′ and A′, respectively).

Pulmonary artery systolic pressure (PASP) was computed as follows: PASP = 4 × [tricuspid regurgitation (TR) velocity]² + mean right atrial (RA) pressure.¹³ The mean RA pressure was estimated based on the diameter and collapsibility of the inferior vena cava.⁴ The mean RV–RA gradient was estimated by tracing the TR velocity-time integral (Figure 1A).¹⁴ The pulmonary artery mean pressure equals the mean RV–RA gradient plus the mean RA pressure. According to the theoretical formula, pulmonary artery diastolic pressure (PADP) was assessed as follows: PADP = 1.5 × [pulmonary artery mean pressure − (PASP/3)].⁴ RVMW indices were analyzed using a software package (AFI) designed to assess left ventricular (LV) myocardial work (LVMW). The prognostic validation of LVMW has been performed during several studies.^15-17 The tricuspid and pulmonic valve event timings were derived from direct visualization of the parasternal short-axis views (Figure 1B). Then, measurements of RV GLS, PASP, and PADP were synchronized by tricuspid and pulmonic valve event timings to produce a noninvasive RV pressure–strain loop (Figure 1A–D).

2.4 Statistical analysis

The normality of the variables was tested using the Kolmogorov–Smirnov test. Normally distributed variables are expressed as the mean ± SD, and those that were not normally distributed are represented as the median (first quartile, third quartile). Differences between sex groups were analyzed using the two-tailed independent Student's t-test and Mann–Whitney U-test, as appropriate.

Univariate and multivariate linear regression analyzes were performed to examine the independent correlations among the RVMW indices and other RV echocardiographic parameters. Multicollinearity was tested by calculating the variance inflation factors in the multiple linear regression models. The collinear variables included the variables with the highest correlation coefficients.

The intra-observer and inter-observer variabilities of the RVMW were estimated for 20 random subjects using the Bland–Altman analysis. One observer analyzed the same echocardiographic images at two different times to evaluate intra-observer variability. Two independent blinded observers analyzed the echocardiographic images to assess inter-observer variability.

All data were analyzed using SPSS (version 26.0; SPSS Inc., IBM Corp). When p < .05, the difference between variables was considered significant.

3.1 Clinical characteristics

During the study, seven subjects were excluded from enrollment because of poor 2D echocardiography images or RV 4D echocardiography images. A total of 30 subjects were excluded because the TR Doppler envelope could not be obtained or the TR Doppler envelopes were of poor quality. Therefore, the feasibility of the RVMW assessment was 74.1% in the study population. The clinical characteristics of the enrolled patients are summarized in Table 1.

3.2 Parameters of RV function

Table 2 summarizes the RVMW and RV systolic and diastolic function parameters of the enrolled population. The RVGWI and RVGCW were lower for men (255 mmHg% [215, 332] vs. 291 mmHg% [263, 341]; p = .018 and 316 mmHg% [268, 387] vs. 357 mmHg% [308, 407]; p = .024, respectively), whereas the RVGWW and RVGWE showed no significant differences between sexes.

3.3 Correlations among RVGWI and other parameters of RV function

The RVGWI showed good correlation with PASP (r = 0.749; p < .0001), moderate correlations with PADP (r = 0.449; p < .0001), TD RMPI (r = −0.220; p = .024), TAPSE (r = 0.294; p = .002), TA pulsed-wave TD S′ (r = 0.234; p = .016), TA color TD S′ (r = 0.283; p = .003), RV FWLS (r = −0.321; p = .001), RV GLS (r = −0.395; p < .0001), and TA A′ (r = 0.306; p = .001), and weak correlation with TA E′ (Table 3). The multivariate analysis showed that the RVGWI was significantly correlated with PASP (standardized β-coefficient = 0.691; p < .0001), TAPSE (standardized β-coefficient = 0.143; p = .036), and RV GLS (standardized β-coefficient = −0.361; p < .0001) (Table 3 and Figure 2A–C).

3.4 Correlations among RVGCW and other parameters of RV function

The RVGCW showed good correlations with PASP (r = 0.810; p < .0001) and PADP (r = 0.541; p < .0001) and moderate correlations with TAPSE (r = 0.275; p = .004), TA pulsed-wave TD S′ (r = 0.230; p = .018), TA color TD S′ (r = 0.248; p = .010), TA IVA (r = 0.288; p = .003), RV FWLS (r = −0.308; p = .001), RV GLS (r = −0.330; p = .001), and TA A′ (r = 0.349; p < .0001) (Table 4). The multivariate analysis showed that the RVGWI was significantly correlated with PASP (standardized β-coefficient = 0.716; p < .0001) and RV GLS (standardized β-coefficient = −0.279; p < .0001) (Table 4 and Figure 2D,E).

3.5 Correlations among RVGWW and other parameters of RV function

The RVGCW showed moderate correlations with PADP (r = 0.214; p = .028), PD RMPI (r = 0.297; p = .002), TA IVA (r = 0.257; p = .008), and RV GLS (r = 0.247; p = .011; Table 5). The multivariate analysis showed that the RVGWI was significantly correlated with TA IVA (standardized β-coefficient = 0.279; p = .008) and RV GLS (standardized β-coefficient = 0.306; p = .008; Table 5 and Figure 2F,G).

3.6 Correlations among RVGWE and other parameters of RV function

The RVGCW showed moderate correlations with PD RMPI (r = −0.62; p = .007), TD RMPI (r = −0.234; p = .016), RV GLS (r = −0.394; p < .0001), and tricuspid IVRT (r = −0.200; p = .040; Table 6). The multivariate analysis showed that the RVGWI was significantly correlated with TD RMPI (standardized β-coefficient = −0.232; p = .043) and RV GLS (standardized β-coefficient = −0.471; p < .0001; Table 6 and Figure 2H).

3.7 Inter-observer and Intra-observer variabilities

The intra-observer and inter-observer variabilities of the RVMW parameters are summarized in Table 7, Figures 3 and 4. Good intra-observer reproducibility and inter-observer reproducibility were indicated.

4.1 Study limitations

Acquiring and quantifying RVMW datasets using a single-provider platform may affect the applicability of these reference values to data obtained from other provider platforms. Additionally, the commercial software required to measure the RVMW is specifically designed to measure the LVMW. The RV afterload may have been inaccurate because it was estimated by calculating the pulmonary pressure rather than by performing direct measurements. Therefore, the calculation of the RVMW is not as precise as that of the LVMW because of the complicated and irregular RV anatomy and shape.^{28, 29} The noninvasive RV pressure–strain loop may need to be validated by the invasively derived RV pressure–strain loop in the future. Moreover, for ethical reasons, cardiac magnetic resonance and right heart catheterization were not performed to verify the validity of the RVMW. Additionally, because the TR Doppler envelope may not be acquired, and because imaging of the TR Doppler envelope is obscure in some healthy subjects, selection bias might have occurred. Furthermore, although all subjects were asymptomatic during regular examinations, we cannot rule out the possibility of subclinical cardiovascular or respiratory diseases, especially in the elderly.