Effects of High- Versus Moderate-Intensity Training on Neuroplasticity and Functional Recovery After Focal Ischemia
Background and Purpose—This study was designed to compare the effects of high-intensity interval training (HIT) and moderate-intensity aerobic training (MOD) on functional recovery and cerebral plasticity during the first 2 weeks after cerebral ischemia.
Methods—Rats were randomized as follows: control (n=15), SHAM (n=9), middle cerebral artery occlusion (n=13), middle cerebral artery occlusion at day 1 (n=7), MOD (n=13), and HIT (n=13). Incremental tests were performed at day 1 (D1) and 14 (D14) to identify the running speed associated with the lactate threshold (SLT) and the maximal speed (Smax). Functional tests were performed at D1, D7, and D14. Microglia form, cytokines, p75NTR (pan-neurotrophin receptor p75), potassium–chloride cotransporter type 2, and sodium–potassium–chloride cotransporter type 1 expression were made at D15.
Results—HIT was more effective to improve the endurance performance than MOD and induced a fast recovery of the impaired forelimb grip force. The ionized calcium binding adaptor molecule 1 (Iba-1)–positive cells with amoeboid form and the pro- and anti-inflammatory cytokine expression were lower in HIT group, mainly in the ipsilesional hemisphere. A p75NTR overexpression is observed on the ipsilesional side together with a restored sodium–potassium–chloride cotransporter type 1/potassium–chloride cotransporter type 2 ratio on the contralesional side.
Conclusions—Low-volume HIT based on lactate threshold seems to be more effective after cerebral ischemia than work-matched MOD to improve aerobic fitness and grip strength and might promote cerebral plasticity.
Ischemic stroke remains the leading cause of long-term physical disorders. Poststroke hemiparesis frequently leads to physical deconditioning that strongly reduces the quality of life and represents an important burden on the family and society. Growing evidence from animal and human experiments indicated that aerobic training induced beneficial effects at the cardiovascular, muscular, cerebral, and functional levels after cerebral ischemia.1,2 Moderate-intensity aerobic training (MOD; for recommendations see Marsden et al1) is advised after stroke to improve the locomotor abilities, the peak oxygen uptake (VO2peak), and the maximal running speed (Smax), which are strong indicators of quality of life. Early treadmill training in rodents could also promote functional recovery and cerebral plasticity by upregulating the neurotrophin levels, enhancing synaptogenesis, and limiting microglia-mediated proinflammatory cytokine release in the perilesional zones.3,4
However, beneficial effects of MOD on functional recovery, aerobic fitness, and quality of life remain frequently insufficient and controversial.1,5 It is, thus, crucial to reconsider the current guidelines for exercise by defining a safe/effective dosage of training.5 In this regard, authors recently showed that higher training intensities appeared promising for stroke patients.6 Indeed, high-intensity interval training (HIT), known to be feasible and safe in moderate stroke patients,7 could improve VO2peak, running economy, and functional recovery, but it remains controversial.6,7 No clear evidence indicated whether the HIT effectiveness is more efficient on aerobic fitness and neuroplasticity than MOD.6,8 Given that HIT is a time-efficient strategy, we postulated that it might accentuate functional recovery in the acute phase of cerebral ischemia compared with MOD.
In light of these considerations, the present study was designed to compare the effects of work-matched HIT and MOD programs on functional outcomes and cerebral plasticity during the first 2 weeks after cerebral ischemia in rats. One of the key points of the endurance protocols relies on determining for each animal the training intensity from an underestimated submaximal physiological parameter, that is, the running speed associated with the lactate threshold (SLT), which is relevant to distinguish high from moderate running speeds9,10 and is highly sensitive to assess aerobic fitness.10,11 In addition, the training effects on brain inflammation through microglia activation form was measured, as well as the related expression of pro- (IL [interleukin]-1β and IL-12p40) and anti-inflammatory (IL-10) cytokines, to determine the microglia function, which could be related to neurotrophin actions and synaptic plasticity.9 Therefore, the p75NTR expression (pan-neurotrophin receptor p75), known to strongly influence the neurotrophin functions after cerebral ischemia,12 was also assessed. The training-induced synaptic plasticity was observed through the expression of the potassium–chloride cotransporter (KCC2, a neuronal chloride extruder) and sodium–potassium–chloride cotransporter type 1 (NKCC1, an ubiquitously chloride importer) that are disturbed after cerebral ischemia and leads to alteration in the excitation/inhibition balance in brain.13
Material and Methods
Overall, 108 adult male Sprague–Dawley rats (250–270 g; JANVIER, France) were used, but 70 of them were included (see the online-only Data Supplement). Anesthesia and surgical procedures were performed according to the French law on animal care guidelines. Animal Care Committees of Aix-Marseille Université approved our protocol.
Each animal was randomly assigned to a group, making the impact of individuals less prominent: (1) control (n=15), no surgery was performed that enabled to verify the reliability of measurements on 14 days; (2) SHAM (n=9), animals underwent surgery without cerebral ischemia to ensure that it did not affect measurements; (3) middle cerebral artery occlusion (MCAO) (n=13), animals underwent middle cerebral artery occlusion–reperfusion (MCAO-r) that enable to assess the spontaneous functional recovery and to verify the variance of rat activity level (no training); (4) MCAO-D1 (n=7) in which animals were euthanized 1 day (D1) after MCAO-r to confirm that animals started training with a similar lesion severity; (5) HIT (n=13); and (6) MOD (n=13) in which animals underwent MCAO-r and performed HIT and MOD programs, respectively (see the online-only Data Supplement).
MCAO-r and Behavioral Tests
Rats were subjected to right MCAO-r for 2 hours (see the online-only Data Supplement). The elevated body swing test, the ladder-climbing test, and the forelimb grip force were performed before (PRE) and after the surgery at day 1, 7, and 14 (D1, D7, and D14, respectively) after MCAO-r (Figure I in the online-only Data Supplement).
Incremental tests were performed on 1° inclined treadmill at D1 and D14. These tests started with 5 minutes of warm-up at 9 m/min to reduce stress. Then, running speed was increased by 3 m/min every 3 min until animals could not maintain the imposed speed. Smax was associated with the last reached stage. Each stage was separated by 20-s interval to perform blood sampling (0.2 μL) after partially cutting the distal area of the tail vein to determine SLT (see the online-only Data Supplement).
Work-Matched HIT and MOD Programs on Treadmill
HIT and MOD programs included 10 sessions from D2 to D12 and 2 recovery days (D7 and D13) to reduce fatigue accumulation that may affect the incremental test performance (see the online-only Data Supplement).
Each animal was randomly assigned to either immunostaining analysis or Western blot at D15. Cresyl violet was used to measure the infarct volume and the percentage of tissue loss (% tissue loss). To investigate the changes of p75NTR and microglia form, immunostaining with antibodies against the p75NTR and ionized calcium binding adaptor molecule 1 (Iba-1) were made at D15 (see the online-only Data Supplement).
Western Blot Analysis
To detect IL-10, IL-1β, IL-12p40, p75NTR, KCC2, and NKCC1 expression, the total protein extracted from each frozen hemisphere was used for Western blot (see the online-only Data Supplement).
Statistical analysis was performed using SigmaStat software program (San Jose, CA). All data are presented as mean±SD (see the online-only Data Supplement).
For the overall parameters, no difference was observed between control and SHAM groups from PRE to D14. Likewise, no significant difference was observed at D1 for functional outcomes, infarct volume, and endurance performance between MCAO, MCAO-D1, HIT, and MOD groups, indicating a similar lesion severity prior to training for each animal.
Running speed during MOD was lower than HIT during the first and second training weeks (−28.9% and −31.2%, respectively). Session duration of MOD group was higher than HIT group during the first and second weeks (+49.2%; +59.8%, respectively; Table I in the online-only Data Supplement).
HIT induced a complete grip strength recovery without affecting other functional parameters. Indeed, grip force exerted by the affected forelimb decreased significantly between PRE and D1 for all injured groups (P<0.001) and was significantly lower in MCAO, HIT, and MOD groups than in control and SHAM groups (P<0.001). Grip force remained significantly decreased (P<0.01) at D7 and D14 for both MCAO and MOD groups, while it recovered in HIT group at D7 and D14 (P<0.001). Moreover, no difference was observed between control, SHAM, and HIT groups from D7 to D14 contrary to MCAO (P<0.05) and MOD (P<0.001) groups (Figure 1A). No difference was observed for both forelimbs force and for the nonaffected forelimb force between groups from PRE to D14 (data not shown). The A/N ratio significantly decreased at D1 compared with PRE for MCAO, HIT, and MOD groups (P<0.001) and was lower within all injured groups than control and SHAM groups (P<0.001). A/N ratio completely recovered only for HIT group from D7 to D14 compared with PRE (P<0.001). Likewise, A/N ratio of HIT group was significantly higher than MCAO and MOD groups from D7 to D14 (P<0.01) and remained similar to control and SHAM groups, contrary to MCAO and MOD groups (P<0.01; Figure 1B).
The left swings/total swings ratio (elevated body swing test) significantly increased for MCAO, HIT, and MOD groups at D1, D7, and D14 compared with PRE (P<0.001; Figure 1C).
The successful score (ladder-climbing test) significantly decreased for MCAO, HIT, and MOD groups at D1, D7, and D14 compared with PRE (P<0.001), without difference between groups. Nevertheless, this score was significantly higher for MCAO, HIT, and MOD groups at D1 compared with D7 (P<0.001) and with D14 (P<0.001; P<0.01; and P<0.001, respectively; Figure 1D).
HIT appeared to be more effective to recover aerobic fitness than MOD as indicated by changes in Smax and SLT. The resting blood lactate concentration of MCAO group at D14 (4.4±1.4 mmol/L) was higher (P<0.001) compared with control (2.2±0.6 mmol/L), SHAM (2.0±0.6 mmol/L), MOD (3.1±1.3 mmol/L), and HIT (2.7±1.1 mmol/L) groups (Figure 2A).
SLT of MCAO, HIT, and MOD groups at D1 were significantly lower than the one in control (P<0.001) and SHAM (P<0.001) groups (Figure 2B). SLT significantly increased from D1 to D14 for HIT (20.4±2.4 m/min for D1 and 34.5±3.8 m/min for D14; P<0.001; Figure 2C) and MOD (21.8±3.3 m/min for D1 and 26.7±5.3 m/min for D14; P<0.01) groups contrary to MCAO (22.5±3.9 m/min for D1 and 22.5±4.5 m/min for D14), control (32.8±3.9 m/min for D1 and 30.8±4.1 m/min for D14), and SHAM (34.7±4.3 m/min for D1 and 33.7±3.6 m/min for D14) groups (Figure 2D). However, the SLT of HIT group at D14 was higher than that of MOD and MCAO groups (P<0.001). The SLT of MOD group was higher than the SLT of MCAO group at D14 (P<0.01) but remained significantly lower than control and SHAM groups (P<0.01), contrary to HIT.
Smax at D1 was significantly lower in MCAO, HIT, and MOD groups than control and SHAM groups (P<0.001; Figure 2B). However, Smax significantly increased from D1 to D14 for HIT (26.1±3.5 m/min for D1 and 40.8±5.9 m/min for D14; P<0.001) and MOD (27.0±3.3 m/min for D1 and 35.7±4.5 m/min for D14; P<0.001) groups contrary to MCAO (27.5±4.9 m/min for D1 and 30.0±5.4 m/min for D14), control (40.0±4.5 m/min for D1 and 39.8±4.3 m/min for D14), and SHAM (43.2±4.5 m/min for D1 and 41.3±2.9 m/min for D14) groups (Figure 2D). Moreover, the Smax at D14 of HIT group was significantly higher than that of MOD (P<0.05) and MCAO (P<0.001) groups. The Smax of MOD group was higher than the Smax of MCAO group at D14 (P<0.01) but remained significantly lower than control and SHAM groups (P<0.05).
HIT promoted ramified microglia, p75 increase, restoration of NKCC1/KCC2 ratio, and downregulated pro- and anti-inflammatory cytokine expression, without affecting infarct volume and the percentage of tissue loss. Indeed, the number of amoeboid Iba-1+ cells for HIT group (20.6±5.2%) was significantly lower than that for MOD (74.9±29%; P<0.01) and MCAO (77.1±31.5%; P<0.001) groups within the perilesional site, as well as in the contralesional hemisphere (HIT, 11.3±3.1%; MOD, 54.3±22.2%; P<0.01; and MCAO, 46.9±29.9%; P<0.01; Figure 3).
For qualitative staining, the cells of damaged hemispheres expressed p75NTR proteins in all lesioned groups contrary to SHAM group (Figure 4C).
No difference was observed between lesioned groups for infarct size and percentage of tissue loss (MCAO, −3.3±7.6%; HIT, −5.4±6.1%; MOD, −2.7±7.8% of the contralesional hemisphere).
In the HIT group, IL-10 expression was significantly downregulated in the ipsilesional hemisphere when normalized to the IL-10 expression of MCAO group (0.75±0.09; P<0.01) contrary to MOD (1.03±0.23). IL-12p40 expression was significantly downregulated in the ipsilesional hemisphere after HIT (0.81±0.03; P<0.01). No difference was observed for IL-12p40 expression in MOD group (0.93±0.15). Likewise, no difference was observed for IL-1β between groups (0.85±0.23 for HIT and 0.93±0.12 for MOD).
The relative expression of p75NTR protein within the ipsilesional hemisphere in HIT (4.4±2.7; P<0.01) and MOD (3.1±1.9; P<0.05) groups was significantly higher than that in SHAM, contrary to MCAO (2.8±2.6; Figure 4A). In the contralesional hemisphere, the p75NTR expression was not different between groups (Figure 4B).
In the ipsilesional hemisphere, no difference was observed for NKCC1/KCC2 ratio between MCAO (2.20±1.46), MOD (4.23±5.16), HIT (1.75±0.89), and SHAM (1.05±0.68; data not shown). The NKCC1/KCC2 ratio of HIT group (0.62±0.16) was significantly lower than MCAO (1.16±1.36; P<0.05) and MOD (2.42±1.38; P<0.05) groups in the contralesional hemisphere (Figure 5).
At D1 and D15, grip force of the left forepaw, infarct volume, and immunohistochemistry results were not correlated within groups (data not shown).
For the first time, this study demonstrated that 2 weeks of HIT was more effective than a work-matched MOD program on multiscale measurements after cerebral ischemia. The use of lactate threshold enabled defining high (>25 m/min) and moderate (<20 m/min) running speed in an individualized manner for each animal with cerebral ischemia. Such intensity ranges were not in accordance to previous studies in which exercise intensity between 10 and 13 m/min was considered as intense for MCAO rats.10 The difference might be explained by their use of empirical training intensities or maximal parameters (Smax or VO2peak) that were not highly relevant to distinguish high from moderate intensities.9–11 Indeed, the ability to prescribe the optimal training stimulus might be greater whether intensity was based on submaximal physiological parameter that could be reached by the majority of patients, contrary to maximal parameters.11,14 In our study, rats could begin an early individualized intense program that seemed not to be deleterious for functional recovery and infarct volume.15 It was in accordance to previous studies showing that treadmill training starting during the first 5 days induced beneficial effects on recovery (contrary to training that began within 24 hours postischemia in both rodent2 and human16). It was also found in human that early constraint-induced movement resulted in less motor improvement at 90 days.17 However, our results did not indicate that HIT need to be performed in moderate stroke patients during the first 2 weeks because the initiation of aerobic program should later be feasible and safe (during the subacute phase, ie, the first 3 months).18,19 Moreover, HIT induced rapid physiological adaptations, although its session duration was shorter than work-matched MOD confirming that HIT is time efficient.20 It was important given that lack of time remains a major barrier for patients to exercise participation. Interestingly, HIT could also elicit higher enjoyment than MOD, despite higher ratings of perceived exertion during intense series. After reporting an initial apprehension, the patient confidence progressively increased during HIT.6 It suggested that higher intensities might not be considered as a major barrier for patients.
Our study revealed a maintained decrease of SLT and Smax during the first 14 days after MCAO-r for nontrained animals. It was, thus, strongly argued that spontaneous aerobic fitness recovery was insufficient. It was in accordance with another study in which a decrease of SLT was observed 2 days postischemia.21 The decreased SLT at D1 might be associated with sensorimotor alterations, such as interlimb coordination or strength deficits, given that neuromuscular disorders might disturb metabolic activity. Given that the SLT was influenced by muscular typology composition and atrophy (observed from D7 after cerebral ischemia10), muscle typology changes might partially explain the decrease of SLT at D14 but not at D1.
HIT was more effective than MOD to improve aerobic fitness as indicated by a superior shift of SLT and Smax to higher intensities during incremental test. Other studies indicated that HIT induced higher ventilatory threshold improvements than MOD in cardiovascular patients.22 It, thus, suggested that rats were able to exercise for longer durations at greater percentages of their Smax, reducing fatigue at a given intensity after HIT. Both programs are known to improve the maximal oxidative capacity in humans and animals by increasing VO2peak, contributing to explain Smax improvement.15 However, HIT further improved maximal running performance in our study because it was recently observed in rats with chronic heart failure.23
Our study also revealed a resting hyperlactatemia 14 days after the cerebral ischemia in nontrained rats. High lactate levels during the acute phase of stroke (<3 months) was also observed on 25% of stroke patients and seemed to have deleterious repercussions on functional recovery.24 It might be suggested with caution that increase of cortisol and catecholamine blood levels induced by cerebral ischemia are known to be involved in blood lactate concentration accumulation, but also that anaerobic glycolysis promoted by cerebral hypoperfusion in affected neurons might facilitate an increase in blood lactate.25 Given that HIT and MOD are known to increase the lactate transporter expression, the lactate clearance might be improved by training, preventing hyperlactatemia.
The decrease of affected forelimb grip force and its consequent strength asymmetry persisted until D14. However, HIT induced a rapid recovery of the affected forelimb grip force from D7 without acting on the other behavioral tests. The force improvement might be partially explained by a facilitation of fast motor units recruitment during HIT sessions because running speed was above the lactate threshold (contrary to MOD). The force improvement of forelimb flexor muscles might be possible because they were activated during locomotion (and not only extensor muscles) that might promote muscular changes. Given that only grip force was improved, HIT might be combined with skilled reaching training, known to improve limb function after stroke, to maximize recovery.
However, no functional improvements were observed after MOD, similarly to previous reports.26 In humans, beneficial effects of MOD on motor functions and quality of life were not consistently observed.18 In rodents, treadmill training did not have a significant effect on limb function.2 It was also demonstrated that running at a low intensity (10.5 m/min) was not sufficient to observe a recovery.27 We, thus, postulated that the intensity for MOD (−20% of SLT) was too low to observe benefits.
The amoeboid Iba-1+ cells, known to secrete proinflammatory agents and free radicals, was higher in MCAO and MOD animals within both hemispheres (which was not always observed after MCAO-r28) than in HIT animals. It reflected an inflammatory state, even after MOD. We postulated that MOD exercise intensity might not be sufficient to promote significant changes in microglia morphology, indicating that running intensity under the lactate threshold was unlikely the most effective strategy to reduce inflammation. Conversely, the Iba-1+ cells after HIT mainly showed ramified form within both hemispheres. This was in accordance with previous studies showing that physical training might induce microglia morphological changes.29 However, the ramified form could exert either detrimental activity (M1 phenotype), characterized by proinflammatory cytokine release, or beneficial activity (M2 phenotype) by secreting anti-inflammatory cytokines and neurotrophins. In the present study, HIT, but not MOD, might decrease neuroinflammation in the ipsilesional hemisphere by downregulating both pro- and anti-inflammatory cytokine expression. It was in accordance with findings revealing that aerobic training might be beneficial for neuroprotection by reducing the proinflammatory cytokine expression in healthy animals30 and in mice with neurodegenerative diseases.29 However, our results disagreed with a study on healthy mice in which vigorous training on treadmill enhanced anti-inflammatory cytokine IL-10 release.30 The different training protocols and rodent models (healthy versus cerebral ischemia) might explain controversial findings. Alternatively, it might also be possible that microglia progressively returned to resting state after HIT, resulting in a downregulation of IL-10 level at D15. Therefore, the time course of cytokines during aerobic training remains to be investigated. Nevertheless, the absence of an IL-10 upregulation after MOD was consistent with the observed amoeboid Iba-1+ cells, which was in accordance with a previous study.30 Finally, results should be interpreted with caution because of the sample size for immunohistochemistry analysis of injured groups. However, we observed similar outcomes in each group without excessive variability that might be caused by the individualization of training programs and by a strict control of behavioral outcomes after MCAO-r. Several studies, focused on brain structures with larger sample size, are required to clarify the effects of exercise intensity on cytokine expression and microglia function after cerebral ischemia. It might be possible that a decrease of a proinflammatory state promoted an adaptive neuroplasticity even if the influence of neuroinflammation on neuroplasticity remains to be explored.31
To determine whether HIT or MOD might influence brain plasticity, we quantified the p75NTR level, which strongly contribute to mediate the neurotrophin cellular functions during embryonic development and after central nervous system lesion.12 Our study was the first to measure the effectiveness of different training intensities on p75NTR expression within both hemispheres after severe cerebral ischemia. We found that HIT induced an increase of p75NTR at ipsilesional level. However, p75NTR expression could be associated with beneficial or detrimental functions complicating result interpretation. It was, thus, difficult to establish whether the increase of p75NTR expression was associated with cellular death processes or with beneficial neuroplasticity. Nevertheless, several results allowed us to suggest that the p75NTR expression might be beneficial. First, aerobic training stimulated endogenous BDNF (brain-derived neurotrophin factors)/tropomyosin receptor kinase B expression within both hemispheres that is known to influence p75NTR expression.4,32 Moreover, BDNF suppression, by injecting the ectodomain of tropomyosin receptor kinase B (TrkB) inhibited the increase of p75NTR expression after axotomia, reducing the neuronal survival.33 Only one study found on aged rats that after 8 weeks of endurance training, the p75NTR level increased in parallel with an enhancement of BDNF expression.34 Authors postulated that p75NTR increase might promote survival of damaged neurons, trigger apoptosis for cleaning debris, and induce beneficial environment for axonal regrowth and inflammatory prevention.12 In addition, it was demonstrated in healthy rats that high-intensity training could induce higher cerebral concentrations of BDNF and glial cell-derived neurotrophic factor compared with MOD, which was related to neuroprotection.4,35 To reinforce our hypothesis on the p75NTR role, it was demonstrated that aerobic training could facilitate the conversion of the proBDNF to the mature BDNF in the peri-ischemic regions, which was associated with functional improvements.4,29,36 Indeed, the decrease of proBDNF expression was associated with lower cell death and synaptic depression. On the other hand, the increase of mature BDNF expression could promote synaptic plasticity and rescue neuronal loss.37 In light of these findings, we suggested that the increase of p75NTR expression after HIT might reflect a beneficial role of neurotrophin expression. Moreover, circulating BDNF levels was not measured because it does not mirror brain BDNF levels after stroke. It, thus, complicates interpretation and might be less relevant than brain BDNF.
Finally, to link neurotrophin action to brain plasticity, we studied the Cl− homeostasis through the KCC2 and NKCC1 expression, proteins known to be sensitive to neurotrophin levels. Indeed, BDNF could promote KCC2 expression after central nervous system trauma. To our knowledge, no study determined the role of different aerobic trainings on the Cl− cotransporters after cerebral ischemia, despite their crucial role in the central nervous system function.38 Only 1 study indicated that these chloride cotransporters were sensitive to aerobic training because such physical activity could affect the spinal KCC2 and NKCC1 expression after spinal cord injury, in parallel with functional recovery improvement.39 We, thus, postulated that HIT could influence the Cl− homeostasis by changing the KCC2 and NKCC1 expression after cerebral ischemia that might optimize the equilibrium between excitation/inhibition in brain cells. It also appeared interesting to postulate that the contralesional hemisphere was sensitive to brain plasticity, as indicated by the decrease of NKCC1/KCC2 ratio after HIT, together with changes in reactive gliosis and inflammation.40
This study provided new promising insights into the effectiveness of low-volume HIT on the physiological determinants of aerobic fitness and grip strength. It also seemed that HIT might promote neurotrophin action, synaptic plasticity compared with work-matched MOD. According to human studies, results needed to be interpreted with caution because risk factors and comorbidities were not taken into account in the present study that might change the effects of these endurance programs. This study needed to be considered as an initial assessment of the effects of these training protocols. Nonetheless, HIT is known to be feasible in moderate stroke patients, but its effectiveness compared with MOD needs to be assessed. In animals, it was recommended to deepen the neuroplasticity mechanisms induced by HIT without forgetting its outcomes on functional recovery. It seems now critical to bring evidence on the effects of detraining for the different aerobic programs that remains poorly investigated but important for improving the long-term recovery.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.017962/-/DC1.
- Received March 14, 2017.
- Revision received July 28, 2017.
- Accepted August 1, 2017.
- © 2017 American Heart Association, Inc.
- Schmidt A,
- Wellmann J,
- Schilling M,
- Strecker JK,
- Sommer C,
- Schäbitz WR,
- et al
- Chen MD,
- Rimmer JH
- Boyne P,
- Dunning K,
- Carl D,
- Gerson M,
- Khoury J,
- Rockwell B,
- et al
- Bosch PR,
- Holzapfel S,
- Traustadottir T
- Jaenisch N,
- Witte OW,
- Frahm C
- Marzolini S,
- Oh P,
- McIlroy W,
- Brooks D
- Bernhardt J
- Bernhardt J,
- Churilov L,
- Ellery F,
- Collier J,
- Chamberlain J,
- Langhorne P,
- et al
- McGinley C,
- Bishop DJ
- Holloway TM,
- Bloemberg D,
- da Silva ML,
- Simpson JA,
- Quadrilatero J,
- Spriet LL
- Ito D,
- Tanaka K,
- Suzuki S,
- Dembo T,
- Fukuuchi Y
- Jensen SK,
- Yong VW
- Song C,
- Zhang Y,
- Dong Y
- Shulga A,
- Magalhães AC,
- Autio H,
- Plantman S,
- di Lieto A,
- Nykjær A,
- et al
- Vilela TC,
- Muller AP,
- Damiani AP,
- Macan TP,
- da Silva S,
- Canteiro PB,
- et al
- Côté MP,
- Gandhi S,
- Zambrotta M,
- Houlé JD