Aerobic training enhances muscle deoxygenation in early post-myocardial infarction

Inclusion criteria of the subjects were: (1) fewer than 6 weeks after onset of their first heart attack, (2) age between 40 and 75 years old, and (3) ability to perform exercise testing. All participants had received a percutaneous coronary stent as part of their post-MI clinical treatment. Initially, 23 MI patients were recruited between June 2010 and March 2012 from the cardiac rehabilitation center at Tokyo Medical University Hospital. Patients with peripheral vascular disease, post-infarction angina, critical arrhythmia, (unstable) heart failure, or lung disease were excluded. As a result, 16 early post-MI patients (21 ± 8 days after the first MI) were chosen to participate and invited to engage in exercise training. However, 6 of the 16 subjects were unable to participate in exercise training due to their work schedules. The authors decided to exclude them from the training group, and consequently, the 16 MI subjects were divided into an aerobic exercise training (TR, n = 10) group and a non-training (CON, n = 6) group (Table 1). Mean left ventricular ejection fraction (LVEF) was determined by echocardiography. Peak creatine kinase was measured during the course of their disease. Hemoglobin concentration was measured before exercise testing by blood sampling from the antecubital vein. All medication dosages taken by the MI patients remained stable during the study. Before this study, all patients participated in approximately five sessions of exercise training as part of cardiac rehabilitation in the hospital. A session consisted of cycling exercise for 30 min at the individual’s three metabolic equivalents. The study protocol was approved by the Medical Research Ethics Committee of Tokyo Medical University (#1405) and was conducted in accordance with the Declaration of Helsinki. All subjects were informed of the purpose of the study, and written informed consent was obtained.

Training frequency was set to two sessions per week for 12 weeks. Each session mainly consisted of cycling exercise for 30 min at the individual’s estimated lactate threshold (LT) minus 10 W. Before and after the main cycling exercise, subjects performed stretching for 10 min and cycling at a half of LT minus 10 W for 1 min, as a warm up and cool down. The intensity of cycling training was increased gradually (5 W) over the training period if the heart rate (HR) of estimated LT level attained during baseline exercise testing was not attained during training. All training was monitored by physicians. In CON, none of the subjects were involved in any type of exercise training over the 12 weeks.

In both groups, exercise testing was administered before and after 12 weeks to evaluate the effects of training. After a 3-min warm up at 10 W, the subjects performed 10 W min⁻¹ ramp cycling exercise until exhaustion (Strength Ergo 8, Fukuda-Denshi, Japan). During exercise, pedal frequency of 50 rpm was maintained by keeping time with a metronome, as used in previous studies (Takagi et al. 2014). Subjects were supervised by the investigators, and verbal encouragement was given when needed. All subjects exercised until they could no longer sustain the required cadence. The HR was monitored continuously during the exercise by a 12-lead electrocardiogram (ML-9000 Stress Test System; Fukuda-Denshi, Japan), and pulmonary O₂ uptake (VO₂) and carbon dioxide production (VCO₂) were assessed breath-by-breath (AE310S; Minato Medical Science, Japan) to determine peak VO₂ and estimated LT, using the V-slope method and substantiated in the profiles of the end-tidal partial pressures of O₂ and CO₂, the ventilatory equivalents for VO₂ and VCO₂, and respiratory exchange ratio (RER) (Beaver et al. 1986; Takagi et al. 2014; Wasserman et al. 1973). Immediately after maximal exercise, all subjects cycled at 10 W for 3 min. The seat and handlebar height remained constant for each subject in exercise testing before and after training.

The ∆VO₂/∆WR slope during ramp exercise was calculated by the linear regression of VO₂ with work rate, from 4 min after the beginning of the ramp loading up to the point that represented 80 % of the total duration of ramp exercise (Boone et al. 2008; Day et al. 2003). Ideally, the sub-LT region of the response should have been analyzed, however, since the fit was based on only a very small amount of data, it would have become unstable in the ∆VO₂/∆WR slope. The mean response time (MRT) in VO₂ during ramp exercise was calculated as the time from onset of the ramp workload to the point of intersection between the baseline VO₂ during warm up and a backward extrapolation of the VO₂ vs time slope (Boone et al. 2008; Takagi et al. 2014). The baseline VO₂ was defined as the mean of VO₂ during the last 2 min of warm up. For some subjects, workloads 4 min after onset of ramp loading were higher than workloads at estimated LT. Therefore, the whole slope of ∆VO₂/∆WR and the MRT were calculated using data from six MI patients in TR and four MI patients in CON.

Relative changes from rest in oxygenated hemoglobin/myoglobin concentration (∆oxy-Hb/Mb), deoxygenated hemoglobin/myoglobin concentration (∆deoxy-Hb/Mb) and total hemoglobin/myoglobin concentration (∆total-Hb/Mb), and muscle O₂ saturation (SmO₂) were continuously assessed at the belly of the vastus lateralis (VL) muscle in the left leg by near-infrared spatial resolved spectroscopy (NIR_SRS). The measurement site in VL was defined as 30 % of the length between the patella and the greater trochanter, above the patella (Takagi et al. 2013, 2014). The measured signals were defined as the values averaged over the last 10 s at rest, every 20 W, and peak exercise. During the recovery time of re-saturation, the half-recovery time of SmO₂ (T1/2reoxy) was determined as the time for 50 % reoxygenation of SmO₂ from the exhaustion level to the peak level (Ichimura et al. 2006). During rest periods without pedaling, subjects sat on the bicycle with the left foot on the pedal, at the lowest position.

In this study, a two-wavelength (770 and 830 nm) light-emitting diode NIR_SRS (Astem Co, Japan) was employed. The probe consisted of one light source and two photodiode detectors, and the optode distances were 20 and 30 mm, as were used in previous studies for measuring muscle deoxygenation (Kime et al. 2010; Takagi et al. 2014). The data sampling rate was 1 Hz. Although a previous study demonstrated that fat layer thickness influences NIRS data due to light scattering, the effects of fat layer thickness can be corrected in relative changes in Hb/Mb (Niwayama et al. 2000) or SmO₂ (Niwayama et al. 2007). The specifications of correction for the influence of fat layer thickness have been fully described (Niwayama et al. 2000, 2007). In this study, fat layer thickness at the measurement site of the VL muscles was measured individually before and after intervention, using an ultrasound device (LogiQ3, GE-Yokokawa Medical Systems, Japan). Subsequently, we utilized an NIRS device which had built-in fat-correction software (Astem Co, Japan) and calculated relative changes in Hb/Mb and SmO₂ with correction for the influence of fat layer thickness, based on individual value of fat layer thickness before and after intervention. Even though an upper limit of fat layer thickness was designated as 10 mm to correct for the light-scattering effects in this study, fat layer thickness at the measurement site was within ~10 mm in all subjects before and after intervention (Table 1).

All data are given as mean ± standard deviation. Differences in NIRS, power output, cardiorespiratory variables at peak exercise and estimated LT, VO₂ response, T1/2reoxy, and demographics variables were analyzed by 2-way repeated-measures analysis of variance (ANOVA) with group and intervention (before–after) as factors. To compare changes in variables during submaximal exercise between before and after intervention over power output and percent of peak VO₂, 2-way ANOVA was used, with intervention (before–after) and power output and percent of peak VO₂ as factors. Because one subject could not exercise at more than 62 W before training and 65 W after training, repeated measures between groups were limited to rest, 20, 40, and 60 W compared as a function of power output. As a function of percent of peak VO₂, differences between groups were determined at rest, 20, 40, 60, and 80 % of peak VO₂. Similarly, 2-way ANOVA was also adapted to test the baseline condition between groups. Where appropriate, the Bonferroni post hoc test was performed to determine specific significant differences. Pearson’s correlation coefficient was employed to determine the relationship between variables. The ANOVA and correlation coefficient were analyzed by SPSS version 17.0 J (SPSS Inc., Chicago, IL). For all statistical analyses, significance was accepted at p < 0.05.

Although training frequency was set at two exercise sessions/week for 12 weeks, the subjects exercised 20 ± 3 sessions during the 12-week training period, as their schedules permitted. Initial training intensity (i.e. estimated LT minus 10 W) determined by exercise testing before training was 42 ± 11 W, and the final training intensity was 51 ± 9 W.

There were no significant interactions or significant main effects in any physical or clinical variables between before and after the intervention.

During submaximal exercise, in TR, there were significant intervention (before–after) × power output interactions for changes in VO₂ (p < 0.05, Fig. 1a) and O₂ pulse (p < 0.05, Fig. 1c). Even though no significant interactions were observed, a significant main effect of intervention was observed in HR (p < 0.05, Fig. 1b) and RER (p < 0.05, Fig. 1d). VO₂ was significantly higher after training at 20–60 W, and O₂ pulse was significantly greater after training at rest and 20–60 W. When expressed as a function of peak VO₂ (i.e. percent of peak VO₂), there were significant interactions for changes in VO₂ (p < 0.01) and O₂ pulse (p < 0.01) in TR. Significantly higher VO₂ and O₂ pulse were also observed during submaximal exercise after intervention in TR. In contrast, there were no significant interactions or significant main effects in HR and RER during submaximal exercise. In CON, both expressed as a function of power output and a function of peak VO₂, no significant changes were found after intervention in any cardiorespiratory variables. Moreover, at baseline, no significant difference was observed between groups in any cardiorespiratory variables during submaximal exercise.

At peak exercise, there were significant group × intervention (before–after) interactions for VO₂ (p < 0.01), O₂ pulse (p < 0.05), and power output (p < 0.05). In TR, peak VO₂ (p < 0.01), O₂ pulse at peak exercise (p < 0.01), and peak power output (p < 0.01) were significantly increased by training compared to before training (Fig. 1a, c; Table 2), and peak VO₂ was significantly higher in TR than CON after training (p < 0.05, Table 2). In contrast, there were no significant interactions or main effects in HR and RER at peak exercise. At estimated LT, there were significant group × intervention (before–after) interactions for VO₂ (p < 0.05) and power output (p < 0.05). In TR, both VO₂ and power output were significantly increased after training (p < 0.01) at estimated LT (Table 2).

There was a significant group × intervention (before–after) interaction for MRT (p < 0.05) but not for ∆VO₂/∆WR slope (p = 0.80). The MRT was significantly lower after training than before in TR (p < 0.01), and the MRT tended to be lower in TR than CON after training (p = 0.09, Table 2). No significant main effects were found in ∆VO₂/∆WR slope.

Figure 2 shows an example trace of the changes in NIRS variables in real time as the exercise proceeds, with and without fat correction. With fat correction, responses in NIRS variables were somewhat enhanced. Even though SmO₂ was increased by fat correction, SmO₂ responses were similar between with and without fat correction in this study.

During submaximal exercise, in TR, there were significant intervention (before–after) × power output interactions for change in SmO₂ (p < 0.01, Fig. 3a) and ∆deoxy-Hb/Mb (p < 0.01, Fig. 3c), but not in ∆oxy-Hb/Mb (p = 0.21, Fig. 3b) or ∆total-Hb/Mb (p = 0.32, Fig. 3d). Moreover, no significant main effect for training was observed in ∆oxy-Hb/Mb (p = 0.18) or ∆total-Hb/Mb (p = 0.54) in TR. After training, SmO₂ was significantly decreased at 40–60 W (p < 0.01) compared to before training. Similarly, ∆deoxy-Hb/Mb was significantly increased after training compared to before at 40–60 W (p < 0.01). We also found significant interactions for changes in SmO₂ (p < 0.01) and ∆deoxy-Hb/Mb (p < 0.01) in TR when NIRS data was expressed as a function of peak VO₂, and significantly lower SmO₂ and higher deoxy-Hb/Mb were obtained during submaximal (40–80 % of peak VO₂) exercise after intervention in TR. There were no significant interactions or main effects of intervention in ∆oxy-Hb/Mb and ∆total-Hb/Mb as a function of peak VO₂. In CON, both expressed as a function of power output and a function of peak VO₂, there were no significant changes after 12 weeks in any NIRS variables. At baseline, no significant differences were found in these variables between TR and CON during submaximal exercise.

At peak exercise, there were significant group × intervention (before–after) interactions for SmO₂ (p < 0.05) and ∆deoxy-Hb/Mb (p < 0.05), while no significant interactions or main effect were observed in ∆oxy-Hb/Mb and ∆total-Hb/Mb. SmO₂ at peak exercise was significantly decreased after training than before training in TR (p < 0.01, Fig. 3a), and consequently, lower SmO₂ was found in TR than CON after training (p < 0.05). Similarly, ∆deoxy-Hb/Mb at peak exercise was significantly higher after training than before in TR (p < 0.01, Fig. 3c) and after in CON (p < 0.05).

A significant group × intervention (before–after) interaction for ∆SmO₂ (SmO₂ at peak exercise minus SmO₂ at rest) was found. ∆SmO₂ was significantly decreased after training in TR (before: 2.4 ± 5.7 %, after: −7.0 ± 3.4 %, p < 0.01) and after intervention in CON (TR: −7.0 ± 3.4 %, CON: 1.1 ± 3.6 %, p < 0.01). There was no significant interaction or main effect in T1/2reoxy (Table 2).

When NIRS variables are expressed without fat correction, similar changes were obtained in all variables. In TR, there were significant interactions for changes in SmO₂ (p < 0.01) and ∆deoxy-Hb/Mb (p < 0.01), and significantly lower SmO₂ and higher deoxy-Hb/Mb were observed during submaximal exercise (40 and 60 W) after training. There were no significant interactions or main effects of intervention in oxy-Hb/Mb and total-Hb/Mb. In CON, no significant interactions or main effects of intervention were found in any NIRS variables without fat correction.

In all subjects (TR and CON), significant negative relationships were observed between change in ∆SmO₂ by training and increase in peak VO₂ (r = − 0.85, p < 0.01; Fig. 4a) and estimated LT (r = − 0.77, p < 0.01; Fig. 4b). Similarly, there were positive relationships between changes in ∆deoxy-Hb/Mb at peak exercise by training and increase in peak VO₂ (r = 0.75, p < 0.01; Fig. 4c) or increase in estimated LT (r = 0.76, p < 0.05; Fig. 4d). In TR, change in ∆SmO₂ by training was also significantly negatively correlated with an increase in peak VO₂ (r = − 0.65, p < 0.05) and estimated LT (r = − 0.73, p < 0.05). Similarly, significant positive relationships were observed between changes in ∆deoxy-Hb/Mb at peak exercise by training and increase in peak VO₂ (r = 0.64, p < 0.05) or estimated LT (r = 0.78, p < 0.01).

When the influence of light scattering on NIRS variables was not corrected, the relationship was robust. Changes in ∆SmO₂ without fat correction were significantly negatively correlated with an increase in peak VO₂ (r = − 0.62, p < 0.05) and estimated LT (r = − 0.81, p < 0.01). There were significant positive relationships between changes in ∆deoxy-Hb/Mb at peak exercise by training and enhancement of peak VO₂ (r = 0.70, p < 0.01) or estimated LT (r = 0.83, p < 0.01).

Workloads of training intensity (W) at end of training (r = 0.57, p = 0.08), improvement of training intensity (r = 0.71, p < 0.05), and training frequency (r = 0.71, p < 0.05) were related to the increase in peak VO₂ by training.

After exercise training for 12 weeks in early post-MI patients, skeletal muscle deoxygenation during submaximal and peak exercise was enhanced, while no significant changes were observed in any variables in CON. Because deoxygenation also differed significantly at a given absolute power output, enhancement of deoxygenation is not solely attributed to the increase in maximal power output by training. Aerobic exercise training is confirmed to enhance deoxygenation measured by NIRS technique in healthy subjects (evaluated without a non-training group) (Kime et al. 2010; Prieur and Mucci 2013) and CHF patients (evaluated with a non-training group) (Mezzani et al. 2013). One possible explanation for the enhancement of muscle deoxygenation is the enhancement of muscle O₂ consumption by aerobic training. Aerobic training increased oxidative enzyme activity in skeletal muscle of MI rats (Moreira et al. 2013). Belardinelli et al. (1995) reported that aerobic cycling training increased mitochondrial density at VL without improvement of cardiac output in CHF patients. In the present study, SmO₂ was significantly lower and deoxy-Hb/Mb was significantly higher after training compared to before training, while total-Hb/Mb, which is an indicator of blood volume, was not significantly changed during exercise. Moreover, oxy-Hb/Mb values were decreased after training, even though changes in oxy-Hb/Mb did not reach statistical significance. Although NIRS technique measured the muscle O₂ balance between supply and utilization, muscle deoxygenation after training in MI may be partly explained by the increase in muscle O₂ consumption of exercising muscle. Another possibility for enhanced muscle deoxygenation during submaximal exercise is reduced muscle blood flow caused by exercise training, as seen in previous studies of healthy subjects (Saito et al. 1980) and post-MI patients (Clausen and Trap-Jensen 1970). However, VO₂ during submaximal exercise was significantly higher after training in this study (as described below). Due to the generally positive relationship between muscle blood flow and VO₂, we suspect that a reduction in muscle blood flow results in higher VO₂. Taken together, we presume that the enhancement of muscle deoxygenation can be most likely be explained by muscle O₂ utilization change, rather than flow change, during submaximal exercise.

T1/2reoxy is an index of muscle oxidative capacity, and it also reflects the balance between muscle O₂ supply and utilization. In this study, no significant changes were observed both in TR and CON. In contrast, in the TR group, peak VO₂ was significantly increased after intervention. These results suggest that muscle VO₂ after peak exercise was higher after intervention compared with before in TR, and lead us to speculate that muscle blood flow at peak exercise is also increased by training. However, we should measure muscle blood flow directly in future studies.

In early post-MI patients, the effects of aerobic training on peripheral factors and its relation to peak VO₂ have not been established. Previous studies partly demonstrated that an aerobic training intervention affected a cross-sectional area of thigh muscle (Murabayashi et al. 1997) and muscle oxidative capacity by nuclear magnetic resonance spectroscopy (Cottin et al. 1996), and the changes were related to peak VO₂ in early post-MI patients. However, these studies did not have a non-training control group. In the TR group of the present study, changes in ∆SmO₂ and ∆deoxy-Hb were significantly correlated with changes in peak VO₂ and estimated LT, while no significant difference was observed in either cardiovascular or muscle deoxygenation variables in CON. An increase in aerobic capacity reduces cardiovascular-associated morbidity and mortality (Merz et al. 2009). Therefore, our findings partly support the idea that exercise training may be important for early post-MI patients to improve peripheral functions and aerobic capacity.

VO₂ is the product of cardiac output (convective O₂ supply) and O₂ extraction, and in a strict sense, muscle O₂ extraction can be estimated by deoxy-Hb/Mb or SmO₂ when total-Hb/Mb is stable (Ferrari et al. 1997). In this study, total-Hb/Mb increased as power output increased, however, total-Hb/Mb was not different between before and after 12 weeks of intervention. Therefore, the change in deoxy-Hb/Mb and SmO₂ between before and after training may be partly explained by increasing muscle O₂ extraction. Okita et al. (2013) suggested that skeletal muscle dysfunction, rather than cardiac dysfunction, limits exercise tolerance in most patients with CHF. Hence, the strong relationship between change in muscle deoxygenation and increase in peak aerobic capacity in the present study may be attributed to the idea that increasing peak aerobic capacity was mainly caused by increasing muscle O₂ extraction, rather than cardiac output, during training in early post-MI patients.

In the present study, training intensity in last session and training frequency were positively related to improvement of peak aerobic capacity. Moreover, peak VO₂ was certainly improved by aerobic training in TR, even though both the training intensity and the frequency are low. These results indicated that both training intensity and frequency are important factors in improving peak aerobic capacity in early post-MI. Recent previous studies demonstrated that high-intensity interval training may be effective to improve aerobic capacity in coronary heart disease patients (Rognmo et al. 2012), and the risk may be the same compared with moderate intensity aerobic training in heart failure patients (Wisløff et al. 2007). However, Blumenthal et al. (1988) pointed out that “many patients experience subjective discomfort during exercise at high intensity and are therefore dissuaded from pursuing further training, despite its potential benefits”. Moreover, to our knowledge, a risk of high-intensity interval training is unknown in early post-MI patients. Therefore, we presume that frequent aerobic training seems to be one of the best options to improve aerobic capacity in early post-MI patients for now, even if the benefits are not ideal.

We recognize the absence of direct measurement, such as muscle blood flow, cardiac output during exercise, or muscle biopsy, to support our findings. Secondarily, the number of subjects is quite low, especially women. Our findings should be confirmed in a larger group of subjects. Moreover, the subjects in this study were not divided randomly into TR or CON. Thus, some sampling bias cannot be ruled out. In addition, our subjects included patients treated with medications. However, our findings cannot be solely attributed to these medications, because the dosages remained stable throughout the study. Furthermore, there are some general limitations of NIRS technique (Hamaoka et al. 2007). NIRS data reflects tissue oxygenation in arterioles, capillaries, and venules of localized exercising muscle, especially in the superficial region of the muscle. The deoxygenation measured by NIRS was heterogeneous between muscles and also within a single muscle (Takagi et al. 2013). Hence, the muscle deoxygenation in early post-MI patients may be restricted to the superficial region of the distal site of the VL muscle. Because muscle recruitment affects the changes in NIRS variables during ramp cycling exercise (Chin et al. 2011), it may be that central attenuation or inhibition of muscle activity in the superficial region of the distal site of the VL muscle is responsible for the changes observed in NIRS dynamics. In particular, post-MI patients in this study had substantially preserved LVEF even before training, and the effect of training on deoxygenation may be influenced by the previously preserved LVEF. Furthermore, in early post-MI patients, aerobic capacity may be increased by very light intensity exercise such as physical activity (Dressendorfer et al. 1995). Though none of the subjects in CON were involved in training over the 12 weeks, we did not regulate their usual physical activity. Future studies are needed which include measurement of physical activity.