Regression of Left Ventricular Mass in Athletes Undergoing Complete Detraining Is Mediated by Decrease in Intracellular but Not Extracellular Compartments

Background: Athletic cardiac remodeling can occasionally be difficult to differentiate from pathological hypertrophy. Detraining is a commonly used diagnostic test to identify physiological hypertrophy, which can be diagnosed if hypertrophy regresses. We aimed to establish whether athletic cardiac remodeling assessed by cardiovascular magnetic resonance is mediated by changes in intracellular or extracellular compartments and whether this occurs by 1 or 3 months of detraining. Methods: Twenty-eight athletes about to embark on a period of forced detraining due to incidental limb bone fracture underwent clinical assessment, ECG, and contrast-enhanced cardiovascular magnetic resonance within a week of their injury and then 1 month and 3 months later. Results: After 1 month of detraining, there was reduction in left ventricular (LV) mass (130±28 to 121±25 g; P<0.0001), increase in native T1 (1225±30 to 1239±30 ms; P=0.02), and extracellular volume fraction (24.5±2.3% to 26.0±2.6%; P=0.0007) with no further changes by 3 months. The decrease in LV mass was mediated by a decrease in intracellular compartment volume (94±22 to 85±19 mL; P<0.0001) with no significant change in the extracellular compartment volume. High LV mass index, low native T1, and low extracellular volume fraction at baseline were all predictive of regression in LV mass in the first month. Conclusions: Regression of athletic LV hypertrophy can be detected after just 1 month of complete detraining and is mediated by a decrease in the intracellular myocardial compartment with no change in the extracellular compartment. Further studies are needed in athletes with overt and pathological hypertrophy to establish whether native T1 and extracellular volume fraction may complement electrocardiography, echocardiography, cardiopulmonary exercise testing, and genetic testing in predicting the outcome of detraining.


Abstract Objectives
To establish if athletic cardiac remodelling assessed by cardiovascular magnetic resonance (CMR) is mediated by changes in intracellular or extracellular compartments, and whether this occurs by one or three months of detraining.

Background
Athletic cardiac remodelling can occasionally be difficult to differentiate from pathological hypertrophy. Detraining is a commonly used diagnostic test to identify physiological hypertrophy which can be diagnosed if hypertrophy regresses.

Methods
Twenty-eight athletes about to embark on a period of forced detraining due to incidental limb bone fracture underwent clinical assessment, electrocardiogram and contrast enhanced CMR within a week of their injury, and then one month and three months later.
High LV mass index, low native T1 and low ECV at baseline were all predictive of regression in LV mass in the first month.
Introduction guidelines for ECG interpretation in athletes. 11 LV mass was estimated from ECG by the 1 Sokolow-Lyon product, the voltage sum of the greatest S wave in V1/2 and R wave in V5/6. 12 2 A full blood count, for measurement of haematocrit, was taken at the time of intravenous 3 cannulation prior to each CMR study. 4

Cardiovascular Magnetic Resonance Acquisition 5
Participants underwent CMR on a dedicated cardiovascular 3 Tesla Philips Achieva system 6 equipped with a 32 channel coil and MultiTransmit® technology. Data were acquired during 7 breath-holding at end expiration. Balanced steady state free precession (SSFP) cine images 8 covering the entire heart in the LV short axis were acquired prior to contrast administration 9 (repetition time (TR) 2.7ms, echo time (TE) 1.3ms, matrix 320 x 320, slice thickness 10mm 10 with no gap, 30 cardiac phases). 11 T1 maps were acquired in a three short axis slices. Native T1 mapping used a breath-held 12 Modified Look-Locker Inversion recovery (MOLLI) acquisition (ECG triggered 5s(3s)3s, 13 single-shot, SENSE factor 2, prepulse delay 350ms, trigger delay set for end-diastole 14 (adaptive), flip angle 20°, matrix 400 x 400, slice thickness 10mm, giving a reconstructed voxel 15 size of 1.17 x 1.17mm). 16 0.15 mmol/kg of gadobutrol was administered through an intravenous cannula with a 10ml 17 saline flush (Gadovist®, Bayer Pharma, Berlin, Germany). 18 Tissue tagging by spatial modulation of magnetization (SPAMM) (spatial resolution 19 1.51 × 1.57 × 10 mm3, tag separation 7 mm, ≥18 phases, TR 5.8ms, TE 3.5ms, flip angle 10°, 20 typical temporal resolution 55 ms) was acquired in the three short axis slices. 13 21 Late gadolinium enhancement (LGE) in matching LV short axis planes were carried out more 1 than 6 minutes after contrast administration . Typical parameters were TR 3.7ms, TE 2.0ms,  2   flip angle 25 o , matrix 512 x 512, slice thickness 8mm with 2mm gap.  3 Post contrast T1 mapping was carried out exactly 15 minutes following last contrast injection 4 using 4s(3s)3s(3s)2s MOLLI acquisition with identical positioning and planning to the native 5 T1 map. 6 Cardiovascular Magnetic Resonance Analysis 7 CMR data were assessed quantitatively using commercially available software blinded to 8 detraining status (CVI42, Circle Cardiovascular Imaging Inc. Calgary, Canada). Epicardial and 9 endocardial borders were traced offline on the short axis cine stack at end-diastole and end-10 systole to calculate LV and right ventricle (RV) end-diastolic volume (EDV), end-systolic 11 volume (ESV), stroke volume (SV), ejection fraction (EF) and LV mass. Papillary muscles 12 were excluded from all measurements. Indexed cardiac parameters were divided by body 13 surface area calculated by the Mosteller equation at baseline. 14 LGE imaging was analysed 14 visually to assess for the presence of scarring. 15 Pre and post contrast myocardial T1 values with a 3-parameter exponential fit with Look-16 Locker correction were measured from short axis slices in the septum.Average measurements 17 from the basal and mid ventricular slices were used. Data from the apical slice was not used 18 because it was vulnerable to partial volume effects due to decreased wall thickness. ECV was 19 calculated from native and post contrast T1 times of myocardium and blood pool and 20 haematocrit as previously reported. 15 21 Intracellular compartment volume was calculated by multiplying (1-ECV) x (LV mass/1.05). 22 Tagging analysis was conducted using inTag (v1.0 CREATIS lab, Lyon, France). Endocardial 1 and epicardial contours were drawn on the short axis SPAMM sequences using a semi-2 automated process as reported previously. 13 Peak LV circumferential strain was measured for 3 the three slices. 4

Statistical analysis and power calculation 5
Statistical analysis was performed using IBM SPSS® Statistics 22.0 (IBM Corp., Armonk, 6 NY). Continuous variables were expressed as mean ± SD or median (interquartile range) 7 depending upon normality. Categorical variables were expressed as N (%). Paired data at 8 baseline one and one month were compared by paired t test. When comparing three paired 9 groups, analysis of variance (ANOVA) with repeated measures was used. P<0.05 was 10 considered statistically significant. 11 Receiver operating characteristic analysis was used to determine the diagnostic accuracy 12 baseline imaging parameters to predict regression of left ventricular hypertrophy (>10g) or 13 cavity dilatation (>10ml). The diagnostic accuracy is expressed as area under the curve (AUC) 14 and 95% confidence interval. Optimal sensitivity and specificity were calculated using Youden 15 index. Variables were combined by binary logistic regression. AUCs were compared by using 16 validated methods described by DeLong et al 17 . 17 The study was powered to detect a 7.5% decrease in indexed intracellular compartment volume 18 after one month of detraining. Assuming that baseline indexed intracellular compartment 19 volume would be comparable to low performance athletes in our previous study, which was 20 47±6ml/m 2 a minimum sample size of 25 athletes would be required (power=0.8, =0.05). 3 21

1
Thirty-five athletes agreed to take part in the study between November 2016 and March 2018. 2 One athlete was unable to complete the study because of claustrophobia, one withdrew because 3 of possible pregnancy and five withdrew consent after the first scan but before the second scan. 4 The final cohort of 28, included 23 male and 5 female athletes with a median age of 24 (IQR: 5 21 -30) years. Twenty-three athletes completed the whole protocol, with five athletes 6 withdrawing after their one month scan because they had resumed full training. Baseline 7 characteristics and their progression throughout the study are shown in Table 1. There were 31 8 ± 5 days between the baseline and one month scan, and 94 ± 10 days between the baseline and 9 three month scans. Athletes trained in a wide range of sports including soccer 9, rugby 5, 10 running 4, mixed sports 4, cycling 3, hockey 1, netball 1 and triathlon 1. Prior to their injury 11 athletes trained median 7 hours per week (IQR 5-9). 12

Changes in Surface Electrocardiogram 13
On one month detraining there was a significant decrease in the voltage of the R wave in chest 14 lead V5 and the Sokolow-Lyon product, both electrical markers of left ventricular mass, Table  15 2. There were no significant changes in heart rate, PR interval or QTc. 16

Changes in Ventricular Morphology 17
After one month of complete detraining there was a 9.3g (7%, P<0.0001) reduction in LV mass 18 with no further reduction between one and three months, Figure 1 and Table 3. This remained 19 significant when indexed to baseline body surface area. In the first month there were significant 20 increases in native T1 and ECV, Figure 2. There was a decrease in intracellular compartment 21 volume (8.4ml, 9%, P<0.0001) with no significant change in the extracellular compartment 22 mass, Figure 3. 23 After one month of complete detraining there were significant comparable decreases in the end 1 diastolic volumes of both ventricles (∆LV -8.2ml, -4.3%, P=0.003; ∆RV -7.8ml, -4.1%, 2 P=0.03). By three months of detraining there was no further decrease in end diastolic volume 3 of either ventricle, Table 3. There was no difference in these temporal changes when they were 4 indexed to baseline body surface area. There was no significant change in LV EF throughout 5 detraining, but there was a reduction in RV EF after one month due to decreased RV EDV. 6 No athlete had scarring detected on LGE imaging on any scan. 7 After one month of detraining there were non-significant absolute increases in peak 8 circumferential strain all three levels (∆apex 0.4%, P=0.61; ∆mid LV 1.1%, P=0.30; ∆base 9 1.4% P=0.06). There were no further changes at three months. 10

Comparison of those who had and who had not resumed training. 11
Between the one month and three month scan 11/23 athletes were able to restart light training, 12 but were still not able to resume full training. When athletes were split according to those who 13 had resumed light training (N=11) and those who had not (N=12) there was no difference in 14 any LV or RV volumetric parameter, native T1 or ECV, Supplementary T1 or ECV were predictive of absolute LV EDV regression in one month of detraining of more 23 than 10ml. Only RV EDV index, but not native T1 or ECV, were predictive of absolute RV 1 EDV regression in one month of detraining of more than 10ml. 2

3
We have shown that in athletes after just one month of complete detraining there is regression 4 of LV mass, LV EDV and RV EDV. There was no further regression of any measure by three 5 months of detraining. The regression of LV mass is mediated by a decrease in intracellular 6 compartment volume (predominantly cardiac myocytes) with no change in extracellular 7 compartment volume. High baseline LV mass is the strongest predictor of regression of LV 8 hypertrophy after one month of detraining. Low native T1/ECV were also predictive of LV 9 mass regression at one month and may have a role in the diagnosis of athlete's heart. 10

Insights into the mechanisms of athletic ventricular remodelling 11
We have demonstrated that regression in LV mass is mediated by a decrease in the intracellular 12 myocardial compartment with no change in the size of extracellular compartment. Previous 13 studies have demonstrated that athletes have lower ECV than sedentary controls and that the 14 fittest athletes have the lowest ECV. 3, 4 These studies were cross-sectional and therefore cannot 15 be used to attribute causality. Our present study is the first to show a longitudinal relationship 16 between LV mass, ECV and training, confirming the hypothesis that athletic hypertrophy is 17 mediated by an increase in the cellular compartment. 18 When T1 mapping data and LV mass are combined it is possible to dichotomise the 19 myocardium into cellular and extracellular compartments. This pattern is particularly relevant 20 in hypertrophic phenotypes and has been validated most comprehensively in aortic stenosis 21 where the derived extracellular compartment volume has a strong correlation with diffuse 22 fibrosis on biopsy 18 and there is regression of both the cellular and extracellular compartments 23 after aortic valve replacement. 16 The most important differential diagnosis in the young athlete with LV hypertrophy is 1 hypertrophic cardiomyopathy. CMR tissue characterisation has been histologically validated 2 in hypertrophic cardiomyopathy and can be used to detect both diffuse fibrosis (increased ECV) 3 and replacement fibrosis (focal LGE). 19, 20 High level athletes with hypertrophic 4 cardiomyopathy are reported to have an altered phenotype with more prominent cavity 5 dilatation but replacement fibrosis is still identified in 33%. 21  with LV hypertrophy at baseline (>12mm interventricular septum) and included athletes at the reflecting the fitness of the athletes studied. The extent of LV mass regression was therefore 1 greater (24-28% vs 7%). These studies defined detraining as reduction in exercise intensity 2 rather than complete cessation perhaps explaining why regression of LV mass not reported 3 until three months compared to one month in our study. 4 Pedlar et al performed echocardiography in 21 amateur runners after an 18 week training 5 programme and then after 4 and 8 weeks when participants were limited to <2 hours of training 6 a week. 27 Similarly to our findings they reported a 10.4% reduction in LV mass after 4 weeks 7 with no change in LV EDV even 8 weeks post-race. 8 The finding of early regression of LV mass is not unique to athletes, and has been reported in 9 by CMR in healthy individuals (N=5) after 6 weeks complete voluntary bed rest 28 and by 10 echocardiography in astronauts (N=38) immediately after a 9-16 day spaceflight. 29 11 The mean LV mass in the present study (130 ± 28g) was comparable to low performance male 12 athletes in our previous study (129 ± 17g) who had a mean VO2max of 60 ± 8mls/kg/min. If we 13 had been able to recruit higher performance athletes with higher LV mass at baseline we may 14 have been able to detect a further decrease in LV mass between one and three months. An 15 alternative explanation is that the pattern of regression reflects the nature of detraining. Athletes 16 were most incapacitated immediately after their fracture leading to most regression in this 17 period. Throughout the subsequent recovery the levels of physical activity gradually increased 18 affecting the regression response. 19 Using the same CMR tagging methodology we have previously shown that athletes have lower 20 peak circumferential strain than sedentary controls. 13 In the current study, we found that in all 21 three levels there was a no significant increase in strain on detraining, despite significant 22 decreases in LV mass and intracellular compartment in the same period. 23

Limitations 1
In this study we relied upon self-reported abstinence from training and it is therefore possible 2 that athletes carried out training that was not reported to the research team. We have not 3 conducted an objective assessment of fitness using cardiopulmonary exercise test but this was 4 not possible due to the nature of the participants' injuries. Athletes in this study participated in 5 a range of sports which giving different patterns of athletic remodelling at baseline. We did not 6 collect data on non-steroidal anti-inflammatory use which may have caused fluid retention and 7 altered the myocardial extracellular compartment. We have not studied athletes with an 8 abnormal ECG, overt LV hypertrophy (12-15mm) or cardiomyopathy and patterns of 9 regression in these groups remains to be established. 10 T1 mapping has only been validated histologically in disease and is difficult to validate in 11 athlete's heart. Native T1 (and less so ECV) vary by field strength, manufacturer and pulse 12 sequence. At present it is recommended that normal values specific to the scanner and 13 acquisition protocol are used to determine ECV in the athlete with unexplained LV hypertrophy 14 30 . 15   Figure 2. Native T1 and ECV (extracellular volume fraction) maps before and after one month of detraining. Native T1 (above) and ECV (below) maps from a rugby player before and after one month of detraining. Over this period native T1 increased from 1160ms to 1213ms, ECV increased from 19.5% to 23.3% and LV mass decreased from 186g to 164g. Typical myocardial and blood pool contours are shown in the lower left panel.   T wave inversion, n (%) 1 (4) 1 (4) 0 (0) 25