Study Design and Participants
The present study followed a single-center, retrospective design and was conducted in the Cardiology Department of the Sports Medicine Center of the Spanish Higher Sports Council Spanish (Madrid, Spain). In this center, Spanish elite athletes participating in a broad range of sport disciplines and who are members of the national team in their specialty and compete in major international events (Olympics, and European and World championships) undergo routine, in-depth cardiological evaluation (one or more per year, most frequently during the preparatory mesocycle), including medical history, physical examination, anthropometric measurements, 12-lead electrocardiogram (ECG), exercise testing, and M-mode and Doppler two-dimensional (2D) echocardiography.
Data were retrospectively analyzed from athletes who had attended the center over a 17-year period (from the start of year 1997 to the end of 2013). Exclusion criteria included being nonwhite, having tested positive for the use of banned substances and/or suspended from participation in official competitions due to violation or anti-doping rules, structural cardiomyopathy, abnormal ECG findings (i.e., not expected in athletes and suggestive of cardiomyopathy), sexual immaturity (< 18 and < 16 years for men and women, respectively), hypertension (baseline systolic or diastolic blood pressure ≥ 140 and ≥ 90 mm Hg, respectively), or an abnormal blood pressure response to exercise. For the sake of consistency, when this was possible, we attempted to choose for the present study in each athlete those evaluations corresponding to the aforementioned preparatory period (e.g., usually during the fall for classical individual ‘Olympic’ sports such as track and field, swimming, or canoeing, among others or July–August for team ball sports). In those athletes with data available for more than one season, we used the evaluation from the last season because this was deemed to reflect the highest degree of adaptation to the sport in question. The study was approved by the local Ethics Committee (#1385226-1) and complies with the Declaration of Helsinki and its later amendments. Oral or written consent was obtained from all participants.
Athletes were categorized according to the modified Mitchell classification into nine groups attending to the relative dynamic/static component of their sport specialty, as recently done by us for normative values of aortic root dimensions (with the inclusion of some sports not included in the original Mitchell’s classification, i.e., mountaineering, freestyle skiing, indoor soccer, motorboat racing, modern pentathlon, and water polo) [13]:
-
IA, low static (< 20% of maximum voluntary contraction [MVC] and low dynamic (< 40% of maximum oxygen uptake [VO2max]) component
-
IB, low static and moderate dynamic (40–70% VO2max)
-
IC, low static and high dynamic (> 70% VO2max)
-
IIA, moderate static (20–50% MVC) and low dynamic
-
IIB, moderate static and moderate dynamic
-
IIC, moderate static and high dynamic
-
IIIA, high static (> 50% MVC) and low dynamic
-
IIIB, high static and moderate dynamic
-
IIIC, high static and high dynamic.
Measures
Echocardiography evaluations were conducted using a Toshiba SSH-140A system (Toshiba Medical Systems, Tochigi, Japan) equipped with 2.5- and 3.75-MHz probes, or a Phillips Sonos 7500 system (Advance Diagnostics, Palo Alto, CA) equipped with a color, tissue Doppler, multifrequency 2–4 MHz transducer. All measurements were taken independently by two experienced sonographers (AB and MEH, 15 years working together). All LV dimensions were measured using 2D-guided M-mode imaging following ASE recommendations [14]. All participants were assessed under resting conditions (i.e., during morning hours or early afternoon, after a rest period from the last exercise training session of at least 12 h). Height and weight were measured (accuracy of 0.1 cm and 0.1 kg, respectively) for the computation of BSA (see below).
Septal wall thickness (SWT), LV posterior wall thickness (LVPW), LVEDD, and LV end-systolic diameter (LVESD) (all in mm) were measured in the parasternal long-axis view, directly from the screen using the scale of the device itself, with the 2D-guided M-mode approach in real time and also guided by the ECG signal in bipolar lead CM5. All the echocardiographic measures corresponding to diastole and systole were obtained coinciding with the start of the QRS complex and with the maximal posterior displacement of the interventricular septum, respectively. Special care was taken when measuring SWT and LVPW to avoid including as part of the wall the different trabeculae both from the LV (false or ‘true’ tendinous chords) and right ventricle (mitral subvalvular apparatus and moderator band (or ‘septomarginal trabecula’)), because inclusion of these structures could erroneously reflect LV hypertrophy. All LV measures were obtained using the mean of three (or five, in case of doubt) cardiac cycles. We used the following equations to measure LV end-systolic volume (LVSV), LV end-diastolic volume (LVEDV), LV mass and LV ejection fraction (LVEF), respectively: LVESV (mL) = [7.0/(2.4 + LVESD)] × LVESD3; LVEDV (mL) = [7.0/(2.4 + LVEDD)] × LVEDD3; and LV mass (g) = 0.8 × {1.04 [(LVEDD + SWT + LVPW)3 − LVEDD3]} + 0.6, where LVEDD, SWT and LVPW are measured in cm; and LVEF (%) = [(LVEDV – LVESV)/LVEDV] × 100.
Diastolic function was assessed by measuring the transmitral flow rate (pulsed-wave Doppler, apical four-chamber view) and determining E and A wave velocities (both in cm/s).
Relative wall thickness (RWT) was calculated with the formula RWT = (SWT + LVPW)/(LVEDD), which allowed grouping the athletes into four categories [14]:
-
Normal geometry, RWT ≤ 0.42 cm and LV mass/BSA ≤116 (males) or ≤ 96 g/m2 (females)
-
Concentric remodeling, RWT > 0.42 cm and LV mass/BSA ≤ 116 (or 96) g/m2
-
Concentric hypertrophy, RWT > 0.42 cm and LV mass/BSA > 116 (or 96) g/m2
-
Eccentric hypertrophy, RWT ≤ 0.42 cm and LV mass/BSA > 116 (or 96) g/m2
LV dimensions were expressed relative to BSA (in m2, calculated as 0.007184 × height (cm)0.725 × weight (kg)0.425 [15]). We also assessed the number of athletes with LV dimensions (thickness and cavity) above those considered ‘normal’ for the general population [14].
In addition, all participants underwent a cardiopulmonary exercise test until volitional exhaustion to determine VO2max using a breath-by-breath metabolic cart (Jaeger Oxycon Pro System; Jaeger, Wuerzburg, Germany), as detailed elsewhere [13, 16]. Depending on the athlete’s sports discipline, the test was performed on a treadmill, cycle-ergometer, or rowing-ergometer.
App “Online Calculator”
We developed a web application using Google Sheets as a database, Javascript for statistical calculations, and HTML 5 for presentation. The data of the athletes that are entered into the application user interface are statistically evaluated against (but not stored in) our reference database.
Statistical Analyses
Data are shown as mean (standard deviation (SD)), and the 95th percentile (P95) is also shown for each variable, as a measure of the upper limit of normality. The χ2 test (or Fisher’s exact test if > 20% of the cells in the cross-table had an expected frequency < 5) was used to compare the proportion of the four types of cardiac geometry between the two sexes and also attending to the dynamic/static component of each sport. Unpaired Student’s t test and one-way analysis of variance (ANOVA) were used for comparisons of LV dimensions for sex and sport category, respectively, with the Bonferroni test used post hoc for pairwise comparisons when a significant group (i.e., sex or sports category) effect was found. Effect size was determined with partial eta squared (η2p, for comparisons of geometry proportions) and Cohen’s d (for comparisons of the different LV dimensions) and was considered small (η2p ≥ 0.01 or d ≥ 0.2), medium (η2p ≥ 0.06 or d ≥ 0.5) or large (η2p ≥ 0.14 or d ≥ 0.8) [17]. We also determined Pearson’s correlation coefficients between VO2max and the different LV dimensions. Finally, for the cardiac dimension variables shown in the App online calculator, we reported Z-scores (i.e., an indicator of how far [that is, how many SD above or below] from the population mean [μ] a data point [X] is), where Z = (x – μ)/SD. Statistical analyses were performed with Statistical Package for Social Sciences (SPSS) (IBM, Armonk, NY).