Summary
Background
An accurate assessment of left ventricular (LV) mass is important for the detection of LV hypertrophy.
Aims
To assess the accuracy of four echocardiographic imaging modalities for assessing LV mass compared with cardiac magnetic resonance (CMR).
Methods
We prospectively studied 40 consecutive patients, who underwent an echocardiographic examination using four imaging modalities (M-mode fundamental imaging [FI], M-mode harmonic imaging [HI], two-dimensional [2D] FI and 2D HI) and CMR (our gold standard for LV mass measurement). All echocardiographic measurements were performed by two independent observers.
Results
All echocardiographic modes significantly overestimated LV mass compared with CMR ( P ≤ 0.04), except 2D FI ( P = 0.25). This overestimation was significantly higher with HI (up to 15.5%) compared with FI (up to 5.7%; P ≤ 0.04). Significant correlations were observed between the different echocardiographic methods and the two observers. The interobserver agreement over LV mass measurement was lower with FI (intraclass coefficient [ICC] range, 0.66–0.73) than with HI (ICC range, 0.72–0.82), and the best agreement was obtained with 2D HI (ICC, 0.82). Good agreement between CMR and all echocardiographic methods was observed among the smallest LV diameters (ICC range, 0.62–0.85), but not among the largest LV diameters (ICC range, 0–0.22).
Conclusions
HI overestimates LV mass compared with FI and CMR; this leads to overestimation of prevalence of LV hypertrophy in a population of hypertensive patients. HI improves interobserver reproducibility of LV mass measurement compared with FI, leading to a significant decrease in the number of patients required for clinical trials evaluating LV mass regression. Accuracy of LV mass measurement by echocardiography is affected by LV geometry.
Résumé
Contexte
La mesure de masse ventriculaire gauche doit être la plus précise possible pour la détection de l’hypertrophie ventriculaire gauche.
Objectif
L’objectif de cette étude est de comparer les quatre différentes modalités de mesure de masse VG en échocardiographie entre elles et à l’IRM.
Méthodes
Quarante patients ont été inclus prospectivement et ont bénéficié d’une échocardiographie comprenant les différentes modalités de mesure (imagerie fondamentale TM, imagerie harmonique TM, imagerie fondamentale 2D, imagerie harmonique 2D), et d’une IRM, considérée comme la méthode de référence de mesure de masse VG. Toutes les mesures échocardiographiques ont été faites par deux observateurs indépendants.
Résultats
Toutes les méthodes de mesure échocardiographiques ont surestimé la mesure de masse VG par rapport à l’IRM ( p ≤ 0,04), exceptée la méthode FI 2D ( p = 0,25). Cette surestimation était supérieure en imagerie harmonique (jusqu’à 15,5 %), comparée à l’imagerie fondamentale (jusqu’à 5,7 %; p ≤ 0,04). Des corrélations significatives ont été constatées entre les différentes modalités échocardiographiques et entre les deux observateurs. La reproductibilité inter-observateur était plus faible en imagerie fondamentale (coefficient intraclasse [CCI] 0,66 à 0,73) qu’en imagerie harmonique (CCI 0,72 à 0,82), et la meilleure reproductibilité a été obtenue en imagerie harmonique 2D (CCI = 0,82). Une bonne reproductibilité entre l’IRM et toutes les modalités échocardiographiques a été observée parmi les petits VG (CCI 0,62 à 0,85), contrairement aux grands VG (CCI 0 à 0,22).
Conclusions
L’imagerie harmonique surestime la mesure de masse VG en comparaison à l’imagerie fondamentale et à l’IRM, ce qui peut entraîner une surestimation du diagnostic d’hypertrophie ventriculaire gauche dans les populations d’hypertendus. L’imagerie harmonique améliore la reproductibilité de mesure inter-observateur comparée à l’imagerie fondamentale, ce qui pourrait permettre des échantillons de patients significativement moins grands pour les études cliniques de régression de masse VG. La précision de mesure de masse VG est influencée par la géométrie VG.
Background
LV hypertrophy is an independent risk factor for cardiovascular mortality and morbidity ; its regression in hypertensive patients is associated with a reduced risk of events . Therefore, measurement of LV mass is of importance for cardiovascular risk stratification. CMR is the reference method for measuring LV mass. However, CMR is not available in most centres and its accessibility is restricted. Therefore, in routine practice, LV mass is mainly estimated by echocardiography, which is considered to be less accurate and reproducible than CMR.
The only anatomically validated echocardiographic method for LV mass measurement is M-mode FI . However, LV mass is routinely measured by other less validated echocardiographic methods: M-mode HI, which is currently used because of improved quality of imaging ; or 2D parasternal long-axis views with FI or HI using M-mode equations . In clinical trials evaluating LV mass modification with drugs, all these approaches are indifferently used and are generally not specified . The aim of this prospective study using echocardiography and CMR was:
- •
to assess the potential impact of different echocardiographic imaging methods on the accuracy and reproducibility of LV mass measurement;
- •
to analyse the potential influence of LV geometry in the assessment of LV mass by echocardiography (Penn and ASE methods).
Methods
Population
We prospectively included 40 consecutive patients with hypertension who were referred for clinical consultation in the hypertension clinic. Entry criteria included age greater or equal to 18 years and sufficient quality of echocardiography for LV measurement. Patients were excluded in case of contraindication to CMR. All included patients underwent an echocardiographic examination and a CMR study on the same day. CMR was considered as our gold standard for the measurement of LV mass. The study was approved by the ethics committee of our university medical institution, and all participating patients gave informed consent.
Cardiac magnetic resonance protocol
CMR examinations were performed using a 1.5-T imager (Signa LX; GE Medical Systems, Milwaukee, WI, USA) with a phased-array torso coil. Breath-hold electrocardiogram-gated segmented cine fast imaging employing steady-state excitation (FIESTA; GE Medical Systems) was performed in long-axis views (four- and two-chamber views) and finally in short-axis views. In all patients, 10 to 16 short-axis cine loops were obtained from base to apex with a slice thickness of 8 mm with a 2-mm gap. CMR cine loops were analysed offline with commercial software (Mass Analysis; Medis, Leiden, The Netherlands). In each short-axis slice, endocardial and epicardial contours were manually traced at end-diastole, while papillary muscles were included in the LV cavity. LV volumes were derived by summation of discs and LV mass was calculated by subtracting endocardial from epicardial volume at end-diastole and multiplying by 1.05 g/cm 3 . All CMR measurements were interpreted by an experienced investigator (E.M.) who was blinded to the echocardiographic results. For each patient, the traced contours were used to calculate LV mass, which served as the reference.
Echocardiography
The same senior physician (E.A.) performed all echocardiographic examinations using an HDI 5000 (Advanced Technology Laboratories, Bothell, WA, USA) echocardiographic system equipped with a multifrequency transducer. For each patient, measurement of LV mass was performed in the parasternal long-axis view using two different modalities (M-mode imaging and 2D imaging) and for each modality, acquisitions were performed using FI and HI. Thus, for each patient, echocardiographic images were captured in four modes: M-mode FI, M-mode HI, 2D FI and 2D HI. All recorded images were set up anonymous and randomly analysed on nine separate tapes. In particular, the FI and HI tracings for each patient were always recorded on different tapes. Readings were performed after all recordings had been completed. All measurements were read in a blind fashion and independently by two experienced physicians (observer 1 [O1; L.P.] and observer 2 [O2; N.M.]). The following variables were systematically measured in M-mode and in 2D mode: end-diastolic IVS, end-diastolic PW and end-diastolic LVD. In M-mode, all measurements were made using Penn and ASE conventions and in 2D mode, all measurements were made using the ASE convention. For each variable, the mean of three different measurements was calculated. The M-mode LV mass was calculated using corresponding equations (ASE: LV mass = 0.8[1.04(LVD + IVS + PW) 3 – LVD 3 ] + 0.6 g; and Penn: LV mass = 1.04[(LVD + IVS + PW) 3 – LVD 3 ] – 13.6 g), and the 2D LV mass was calculated by applying the ASE M-mode equation. Thus, for each patient, we obtained six sets of LV measurements for O1 and six sets of LV measurements for O2: M-mode FI (Penn), M-mode HI (Penn), M-mode FI (ASE), M-mode HI (ASE), 2D FI (ASE) and 2D HI (ASE) ( Fig. 1 ). LV hypertrophy was defined as LV mass greater than 111 g/m 2 in men and greater than 106 g/m 2 in women .
Statistical analysis
Statistical analysis was performed with the Stata statistical software package, version 8.0 (Stata Corp, College Station, TX, USA) and Nquery advisor software, version 4 (Statistical Solutions Ltd, Cork, Ireland). A paired t test was used to compare LV mass estimates obtained by the different echocardiographic modes and to compare echocardiographic measurements (mean LV estimates obtained by O1 and O2) with CMR measurements. Linear regression analysis was used to investigate the relationship between continuous variables. Interobserver variability was assessed by using:
- •
the presence of a systematic difference between two investigators (absolute value or percentage: [O1 – O2]/O1 × 100);
- •
the ICC and the 95% limits of agreement calculated using the method of Bland and Altman .
To assess the interobserver variability in the diagnosis of LV hypertrophy, a paired Chi 2 test with one degree of freedom was applied. The sample size required to detect a clinical change between paired observations in an experimental study was calculated by using the formula for determination of sample size for the comparison of two means, taking paired data into account. These sample sizes were calculated using an alpha value of 0.05, a power of 80% or 90%, to expect either 10 or 20 g LV mass differences. For each echocardiographic mode, we used the SD of the mean difference between O1 and O2, where each measure was itself the mean of three measurements. The variability of echocardiographic LV mass measurement was expressed by:
- •
the presence of a systematic difference between echocardiography and CMR (absolute value or percentage: [LV mass echocardiography – LV mass CMR]/LV mass echocardiography × 100; P < 0.05 was considered significant);
- •
the ICC.
Results
Forty consecutive patients (32 men and eight women) were included. Mean (SD) age was 51 (10) years (range, 28–72 years). Mean (SD) systolic blood pressure was 148 (20) mmHg (range, 102–203 mmHg); mean (SD) diastolic blood pressure was 89 (13) mmHg (range, 61–121 mmHg).
Comparisons of left ventricular mass calculated by echocardiography
For LV mass, significant correlations were observed between the different echocardiographic methods ( r = 0.99 between ASE FI and Penn FI; r = 0.96 between ASE HI and Penn HI; r = 0.80 between Penn FI and Penn HI; r = 0.85 between ASE FI and ASE HI; r = 0.82 between 2D FI and 2D HI; P < 0.0001) and between O1 and O2 ( r = 0.94 for Penn HI; r = 0.81 for Penn FI; r = 0.79 for ASE HI; r = 0.75 for ASE FI; r = 0.81 for 2D HI; r = 0.75 for 2D FI). Regardless of the mode used (M-mode or 2D) or the observer (O1 or O2), LV mass was significantly higher with HI than with FI ( Table 1 ). Moreover, LV hypertrophy was more frequently found with HI compared with FI: using Penn M-mode, LV hypertrophy was found in 33% of patients using HI vs 13% of patients using FI ( P = 0.04).
| M-mode Penn | M-mode ASE | Two-dimensional | ||||
|---|---|---|---|---|---|---|
| Fundamental imaging | Harmonic imaging | Fundamental imaging | Harmonic imaging | Fundamental imaging | Harmonic imaging | |
| LV mass by O1 (g) a | 174.3 (42.1) | 199.3 (47.5) * | 171.5 (35.6) | 196.3 (45.2) * | 162.3 (34.9) | 180.1 (38.8) * |
| LV mass by O2 (g) a | 169.7 (46.1) | 184.4 (46.9) ** | 174.5 (42.5) | 182.9 (41.3) *** | 175.0 (37.7) | 188.5 (39.7) **** |
| SD difference | 32.6 | 28.7 | 29.2 | 30.4 | 27.9 | 22.3 |
| 95% limits of agreement | [–69.9;60.6] | [–72.3;42.4] | [–55.4;61.4] | [–74.3;47.6] | [–43.2;68.6] | [–36.2;53.2] |
| ICC | 0.73 | 0.78 | 0.70 | 0.72 | 0.66 | 0.82 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
