Article Text
Abstract
Objective Echocardiographic cut-off values are often used for cardiac MRI in athletic persons. This study investigates the difference between echocardiographic and cardiac MRI measurements of ventricular and atrial dimensions and ventricular wall thickness, and its effect on volume and wall mass prediction in athletic subjects compared with non-athletic controls.
Methods Healthy non-athletic (59), regular athletic (59) and elite athletic (63) persons, aged 18–39 years and training 2.5±1.9, 13.0±3.0 and 25.0±5.4 h/week, respectively (p<0.001), underwent echocardiography and cardiac MRI consecutively. Left ventricular (LV) and right ventricular (RV) dimensions were measured on both modalities. LV and RV end-diastolic and end-systolic volumes and LV wall mass were determined on cardiac MRI. Echocardiographic M-mode LV volumes (Teichholz formula) and LV wall mass (American Society of Echocardiography formula) were calculated.
Results LV and RV dimensions were smaller on echocardiography (p<0.001), and although the correlation with the cardiac MRI volume was good (p<0.01), the difference in volume was large (LV end-diastolic volume difference 93±32 g, p<0.001). LV wall thickness and calculated wall mass were significantly (p<0.001) larger on echocardiography (wall mass difference −101±34 g, p<0.001). Differences in absolute dimensions did not change significantly between non-athletic and athletic persons; however, the difference in echocardiographic estimations of LV volumes and wall mass did increase significantly with the larger athlete's heart, requiring possible correction of the standard echocardiographic formulas.
Conclusions Echocardiography shows systematically smaller atrial and ventricular dimensions and volumes, and larger wall thickness and mass, compared with cardiac MRI. Correction for the echocardiographic formulas can facilitate better intertechnique comparability. These findings should be taken into account in the interpretation of cardiac MRI findings in athletic subjects in whom cardiomyopathy is suspected on echocardiography.
Statistics from Altmetric.com
Introduction
Preparticipation screening of endurance athletes has gained interest during the past decade. Its main focus is to prevent sudden cardiac death (SCD) from unrecognised cardiac pathology, including hypertrophic cardiomyopathy (HCM) and arrhythmogenic right ventricular dysplasia/cardiomyopathy (ARVD/C) in individuals <40 years, and predominantly coronary artery disease in ≥40 years of age.1,–,3
If the results of clinical evaluation or preparticipation screening (including medical history, assessment of symptoms and signs and ECG) of athletes warrant further investigation, non-invasive imaging is typically used to identify the presence of structural heart disease.4 5 The most frequently used imaging modality is echocardiography, which can accurately assess cardiac function and morphology, while being inexpensive, rapid and widely available.
Unfortunately, physiological changes due to long-term remodelling in response to the increased volume load during endurance training (the athletes' heart) can resemble relevant cardiac disorders, associated with SCD in athletes,6 7 especially when left ventricular (LV) wall thickness is increased to an extent to fulfil HCM criteria, or when the right ventricle (RV) becomes enlarged, a hallmark feature of ARVD/C on echocardiography.7,–,12 This distinction is relevant when a potentially career changing decision must be made for the individual athlete with already suspect findings on the ECG.5 7 13 If echocardiographic results remain inconclusive or warrant further investigations, cardiac MRI can be considered.14 Well-established cut-off values to identify cardiomyopathy have almost exclusively resulted from studies using echocardiography as the non-invasive research tool.14 15
Although numerous papers have studied the difference between echocardiography and cardiac MRI in healthy controls and patients, the effect of this difference in athletes is still unclear, as few studies have included athletes.6 9 16 17 This is a relevant issue in athletes, as differences between echocardiography and cardiac MRI may place them in different risk categories.
Our aim was to establish the difference between echocardiographic and cardiac MRI measurements of ventricular and atrial dimensions as well as ventricular wall thickness in the athlete's heart in a head-to-head fashion using state-of-the-art imaging techniques. We investigated the degree of difference in athletic persons as compared with non-athletic controls and calculated how many athletes would qualify for cardiac pathology using echocardiographic cut-offs for both modalities. Using cardiac MRI ventricular volumes and wall mass as reference, we also studied if the conventionally measured dimensions on echocardiography and cardiac MRI are a reliable prediction of cardiac MRI volume and wall mass.
Methods
Study population
Healthy endurance athletes (national and international competition level) and non-athletic persons, aged 18–39 years, were prospectively enrolled. All subjects were healthy with no history of cardiovascular disease (including hypertension and diabetes). The echocardiographic examination and the cardiac MRI study were performed consecutively (head to head) during one visit in all subjects by AJT and NHJP, respectively. ECG and blood pressure measurement was also performed in all subjects during the same visit. Unexpected hypertension or pathological findings on the ECG, echocardiography or cardiac MRI during the study resulted in exclusion.10 11
The final study population consisted of 181 healthy individuals (mean age 27±5.2 years, 39% women): 122 endurance athletes (59 regular athletic persons training 9–18 h/week and 63 elite athletic subjects training >18 h/week over the past year) and 59 non-athletic controls exercising ≤3 h/week. The study was approved by the Institutional Ethics Committee of the University Medical Center Utrecht. All persons gave written informed consent.
Echocardiography
The echocardiographic examination was performed with the subject at rest, lying in a left lateral decubitus position. Echocardiographic data were acquired using a Vivid 7 scanner (GE Vingmed Ultrasound, General Electric, Milwaukee, Wisconsin) with an M3S broadband transducer. A complete echocardiographic study of both standard parasternal and apical views was performed in two-dimensional (B-mode) and M-mode.
LV dimensions were measured by AT according to the standards of the American Society of Echocardiography (ASE).18 The parasternal two-dimensional measurements included the left atria (LA) and aortic root diameter (Ao), the LV internal diameter at end-diastole (LVIDd) and at end-systole (LVIDs) and the septal (IVSd) and posterior wall thickness (PWd) obtained in the long-axis M-mode (leading edge to leading edge method). Special care was taken to avoid the inclusion of the moderator band in the septal wall thickness measurement. LV wall mass was calculated from these measurements according to the ASE convention: LV wall mass (gr)=0.8×(1 .04((LVIDd+IVDs+PWd)3−(LVIDd)3))+0.6. The LV end-diastolic and end-systolic volumes were calculated according to the Teichholz method: V (ml)=(7/(2.4+LVID))×(LVID)3, with LVID being the internal dimension at end-diastole and end-systole, and V the corresponding volume.18 The RV outflow tract (RVOT) end-diastolic diameter was determined in the long-axis recording, perpendicular to the septum. On the apical four-chamber view, LA and right atrial (RA) end-systolic areas and RV end-diastolic area were measured by tracing the endocardial border. The end-diastolic diameter of the inflow tract of the LV (LVIT) and RV (RVIT), respectively at the level of the mitral and tricuspid valve tips, were also measured on the four-chamber view (figure 1).
Cardiac MRI
All persons were examined on a 1.5 T MRI scanner (Achieva, Philips Medical Systems, Best, The Netherlands) using a five-element phased-array cardiac coil and a vector-ECG triggering set-up. The protocol included ECG-gated breathhold transversal T1-weighted black-blood series, and Steady-State Free-Precession cines with two-chamber LV, four-chamber, short-axis and LV outflow tract (LVOT) views that have been described in detail earlier.11 All acquired cardiac images were viewed to rule out cardiac pathology.
Analysis was performed with a workstation and semiautomated contour-tracing software (View Forum cardiac package version R5.1V1L1 2006, Philips, Best, The Netherlands). An experienced blinded observer (AB) performed a cardiac MRI data analysis of atrial and ventricular diameters, wall thickness and atrial area contours. LVIDd, LVIDs, IVSd, LVPWd, RVOT-parallel long-axis (PLAX) and LA-PLAX were measured on the LVOT cine view, and the LVIT and RVIT were measured on the four-chamber view conform the echocardiographic analysis (figure 1). LA- and RA-area contours were measured on the four-chamber view using the area-length ejection fraction tool.19
Endocardial contours for the ventricular end-diastolic and end-systolic volumes (EDV, ESV) were traced on the short-axis slices, including the outflow tracts.11 20 Epicardial borders were drawn in end-diastolic phase for the calculation of ventricular wall mass.11 20 Papillary muscles and trabeculae were included in the endocardial blood volume.21 Our contour analysts used a reproducible contour-tracing protocol for the segmentation method with an intraobserver and interobserver disagreement of ≤8% and ≤5%, respectively (R2 0.93–0.99). Operators were trained with test cases before a blinded analysis of study cases.21 All measurements were checked by a second blinded observer (AT and NP) experienced in cardiac MRI before finalising the results.
Statistical analysis
Continuous data are presented as mean±SD, and categorical data are presented as frequencies and percentages. For the comparison between the two modalities, we used a paired Student t test, Pearson correlation for linear regression and Bland–Altman analysis for difference estimation on the pooled data of all 181 subjects. To ensure the acquired data were normally distributed (Gaussian), histograms were constructed of all parameters differences. One-way analysis of variance with Bonferroni correction was used to identify differences in baseline characteristics, absolute values and the mean echocardiogram–cardiac MRI differences between the three groups. A p value of <0.05 was considered statistically significant. All statistical analyses were performed using commercially available software (SPSS version 16.0).
Results
Study population
The baseline characteristics of the study population have been described previously and are summarised in table 1.8 10 11 More men than women were included in this study; however, the distribution among the groups was equal. Of all athletes, 37% were rowers (58% men), 29% were triathletes (54% men), 20% were cyclists (92% men), 10% were runners (58% men), and 4.2% participated in other endurance sports (20% men).
Echocardiographic and cardiac MRI dimensions per group are summarised in table 2. All mean values are significantly larger in athletic than in non-athletic persons. Although not significant, all mean values were also higher in elite than in regular athletic subjects. This has been reported in detail previously.8 10 11
Echocardiography versus cardiac MRI
The mean differences between echocardiography and cardiac MRI are shown in table 3. Absolute values on echocardiography compared with cardiac MRI were significantly smaller for ventricular and atrial dimensions, while wall thickness values were significantly larger. The high and significant correlations between the modalities indicate that the observed difference is a systematic over-/underestimation. For distribution analysis, absolute differences were plotted in a histogram (figure 2A,B) and showed a normal Gaussian distribution for all parameters. The regression coefficients, ranging from 0.7 to 1.1, signify that all reported differences in table 3 apply to the entire range of absolute values and did not change significantly when the absolute measurements were larger, as seen in athletes. A subgroup analysis based on the amount of athletic activity in the three groups revealed no significant variance in difference.
Echocardiographic and cardiac MRI dimensions were correlated to cardiac MRI volumes and wall mass as a reference value, and the correlations are presented as a result of the pooled data of all participants in table 4.
The Teichholz formulas for calculation of LV volume resulted in much smaller echocardiographic estimates compared with cardiac MRI volumes for both the LVEDV (125±27 on echocardiography vs 218±53 ml on cardiac MRI, difference 93±32, p<0.001) and the LVESV (31±7.8 vs 94±28 ml, difference 64±23, p<0.001). On the other hand, calculation of LV wall mass using the ASE formula resulted in twofold higher values on echocardiography (205±53 vs 104±33 g, difference −101±34, p<0.001).
The correlations (r) between the techniques were relatively good (LVEDV 0.8, LVESV 0.7 and LV wall mass 0.8). Difference (regression slope b) increased with higher values using these formulas (LVEDV 1.5, LVESV 2.4 and LV wall mass 0.5), which resulted in a significantly higher difference for these parameters in regular and elite athletic as compared with non-athletic subjects.
Correction factor
In order to achieve a 1:1 relation between cardiac MRI and M-mode estimation of LV wall mass, the ASE value has to be divided by approximately 2 to obtain cardiac MRI values (LV wall mass (g): cardiac MRI=(0.49ASE)+3.2). For LVEDV (ml) this would be cardiac MRI=(1.53Teichholz)+26.3 and for the LVESV: cardiac MRI=(2.38Teichholz)+21.6.
Cut-off values
The echocardiographically measured absolute IVSd exceeded a 12 mm threshold in one (3%) non-athletic, four (12%) regular athletic and 11 (26%) elite athletic men, whereas cardiac MRI exceeded 12 mm in two (6%) regular athletic and 10 (23%) elite athletic men. No one exceeded a IVSd threshold of 15 mm on either imaging modality. The echocardiographically measured absolute LVPWd exceeded a 10 mm threshold in five (15%) non-athletic, 15 (46%) regular athletic and 26 (61%) elite athletic men. Cardiac MRI exceeded 10 mm in five (15%) regular athletic and six (14%) elite athletic men. In none of the women did IVSd exceed 12 mm or LVPWd 12 mm on either imaging modality.
The echocardiographically measured absolute LVIDd exceeded a 60 mm threshold in one (3%) regular athletic and three (7%) elite athletic men, but in none of the women. Cardiac MRI exceeded 60 mm in five (15%)/2 (8%) non-athletic, 13 (39%)/2 (8%) regular athletic and 23 (54%)/3 (15%) elite athletic men/women. The echocardiographically measured absolute LVIT exceeded a 60 mm threshold only in one (2%) elite athletic man, and in none of the women. Cardiac MRI exceeded 60 mm in two (6%) non-athletic, 10 (30%) regular athletic and 24 (56%) elite athletic men and three (15%) elite athletic women.
The echocardiographically measured absolute RVIT exceeded a 50 mm threshold only in two (5%) elite athletic men and in none of the women. Cardiac MRI exceeded 50 mm in 15 (44%)/two (8%) non-athletic, 21 (64%)/five (19%) regular athletic and 35 (81%)/two (10%) elite athletic men/women.
Discussion
The present study shows that compared with echocardiography, cardiac MRI ventricular and atrial dimensions and ventricular volumes are larger, and wall thickness and wall mass are smaller.
Although there was a good linear correlation of obtained dimensions with the actual volume on cardiac MRI, the difference in volume and wall mass measurements between echocardiography and cardiac MRI was large. We have provided a possible correction factor on the echocardiographic formulas to facilitate better intertechnique comparability.
Comparison to previous literature
Although echocardiographic image quality has improved greatly in recent years, cardiac MRI still provides a higher LV and RV volume and wall mass measurement reproducibility and accuracy owing to its high spatial resolution.16 17 20 22,–,36 Several studies confirm that RV and LV volumes and dimensions on 2D echocardiography are significantly lower, and LV wall mass and wall thickness higher as compared with cardiac MRI measurements, with moderate agreement per measurement in healthy subjects, as well as in patients.16 17 20 22,–,27 31,–,36 In particular, the large differences between LV volumes and wall mass on cardiac MRI compared with those derived from M-mode echocardiography show the limited accuracy of the ASE and Teichholz formulas.16 27
To our knowledge, this is the first paper using a large cohort of healthy athletic and non-athletic persons, containing both men and women, to compare echocardiographic and cardiac MRI data in a head-to-head fashion. Previous studies comparing echocardiography results with cardiac MRI in athletic subjects included only a small group of either athletic males or females and allowed for much larger time windows between the two examinations.17 28
Even though we minimised the methodological difference for individual variations by performing both examinations consecutively during one session, our results showed larger mean differences between echocardiographic measurements and cardiac MRI than mostly reported in the literature.22 23 25 29 30 One explanation is that the ventricular trabecularisation is recognised better on cardiac MRI and is therefore included in the ventricular volume diameter instead of inclusion in the LV wall thickness as partly occurs on echocardiography (figure 1).23 Another explanation is the use of different cardiac MRI contour tracing protocols, including or excluding the papillary muscles and trabecularisations as ventricular wall mass, as well as the option to include the RVOT or LVOT in the blood volume on cardiac MRI or not.16 17 21 23
Clinical relevance
A reliable assessment of ventricular dimensions by non-invasive imaging is paramount to rule out potentially lethal cardiomyopathies.4 5 To this end, a large body of evidence has been published producing specific cut-off values specifically for echocardiography.7 37 38 Using these echocardiographic cut-off values for cardiac MRI is unjustified without establishing the difference between the modalities. Our results suggest that unmodified implementation of echocardiographic absolute reference values and cut-off values for cardiac pathology on cardiac MRI measurements is not recommended. In athletic persons, there is additional concern because of the sport-heart-related larger ventricular volumes and wall mass, especially in elite athletic men.11 12
For example, in elite athletic men, the absolute RVIT exceeded a 50 mm threshold, sometimes used for arrhythmogenic RV cardiomyopathy more frequently on cardiac MRI (81%) compared with echocardiography (5%).38 39 The absolute LVIDd exceeded a 60 mm threshold, a commonly used echocardiographic cut-off for dilated cardiomyopathy more frequently in cardiac MRI (54%) versus echocardiography (7%).7 These differences suggest that the cut-off value should be adjusted for cardiac MRI (using the 95th percentile).11 A septal wall thickness of >12 mm on echocardiography has generally been regarded as a cut-off value to indicate LV hypertrophy.7 15 The difficulty in recognising the trabecularisation border on echocardiography is illustrated by the similar measurements between echocardiography and cardiac MRI in septal wall thickness (26% vs 23% above 12 mm cut-off) and large variation in PWd (61% vs 14% above 10 mm cut-off).
In clinical practice, a single dimension is often used to get an impression of a three-dimensional parameter. Our results indicate that most of these commonly applied measurements do provide a good insight into the volume and wall mass as calculated on cardiac MRI in both echocardiography and cardiac MRI. The difference between these two modalities seems to be systematic, not influenced by the value of the measure itself, as indicated by the comparable difference in the controls (normal dimensions and LV wall thickness) and athletic persons (ventricular enlargement and LV hypertrophy). Echocardiographic estimation of the LV wall mass and LV volumes using commonly applied formulas do, however, show an increasing difference as the absolute values increased. Although this is to be expected owing to the exponential contributions of the 2D measurements in the different formulas, the relationship was linear (figure 2C). This problem might be overcome using 3D echocardiography, which is free from geometrical assumptions.20 33 35 36 40
Limitations
Several factors, other than the dissimilarity in spatial resolution, could have provided additional variation to the observed difference between the two techniques. First, the measurements were performed by two different observers. Although we checked all measurements for overall consistency with the guidelines, this could have resulted in a systematic difference in the application of the ASE guidelines, and explain the variation in difference between RA (−1.9 mm) and LA (+10.1 mm) where the exclusion of the pulmonary vein ostia could have been performed differently. Second, some reference points used for echocardiographic measurements were less clear on cardiac MRI, such as the tips of the atrio-ventricular valves. This could have resulted in a slightly different measurement location within the heart, which was the case for the LVIT and RVIT. Third, image planes were probably not identical during acquisition. While cardiac MRI follows a protocolised image acquisition sequence, echocardiography is a more user-dependent method, where different cardiac compartments could be recorded separately to obtain the optimal image.30 41 Nevertheless, the observed differences between echocardiography and cardiac MRI are too large to be solely attributed to these limitations and showed a typical systematic difference with a Gaussian distribution, suggesting a relevant clinical difference.
Conclusion
In healthy non-athletic and athletic persons, echocardiography shows systematically smaller atrial and ventricular dimensions and volumes, and a larger wall thickness and wall mass, compared with cardiac MRI. While the differences in absolute dimensions do not change significantly between non-athletic and athletic subjects, the difference in echocardiographic estimations of LV volumes and wall mass does increase significantly with the larger athletic heart, requiring possible correction of the standard echocardiographic formulas. It is important that these findings be taken into account in the interpretation of cardiac MRI findings in athletic persons in whom cardiomyopathy is suspected on echocardiography.
What is already known on this topic
Previous studies show that ventricular volumes and dimensions on two-dimensional echocardiography are significantly lower, and left ventricular wall mass and wall thickness higher as compared with cardiac MRI measurements in healthy subjects and patients.
What this study adds
The linear difference between echocardiographic and cardiac MRI values for physiological enlargement of the athlete's heart and pathological left ventricular (LV) hypertrophy has been established using a head-to-head comparison. Correction factors are offered in order to achieve a 1:1 relation between echocardiographic and cardiac MRI estimation of LV volumes and wall mass.
References
Footnotes
NHJP and AJT have joint first authorship.
-
Competing interest None.
-
Patient consent Obtained.
-
Ethics approval Ethics approval was provided by the Institutional Ethics Committee of the University Medical Center Utrecht.
-
Provenance and peer review Not commissioned; externally peer reviewed.