Articles in PresS. J Appl Physiol (November 14, 2013). doi:10.1152/japplphysiol.00956.2013

1

Determinants of resting lipid oxidation in response to a prior bout of endurance exercise

2 3

By:

Gregory C. Henderson1,2 and Brandon L. Alderman1

4 5 6

1

Department of Exercise Science and Sport Studies, Rutgers University, New Brunswick, NJ 08901

2

Rutgers Center for Lipid Research, Rutgers University, New Brunswick, NJ 08901

7 8 9 10 11 12

AUTHOR CONTRIBUTIONS G.C.H. conceived of and designed the experiment, performed calculations, interpreted findings, and wrote the manuscript. B.L.A. designed the experiment, performed calculations, analyzed results, interpreted findings, and wrote the manuscript.

13

14 15

Corresponding author:

16

Gregory C. Henderson, Ph.D.

17

Rutgers University

18

70 Lipman Drive

19

Loree Building

20

New Brunswick, NJ 08901

21

Email: [email protected]

22

Phone: 732-932-7564

23

Fax: 732-932-9151

24 25 26

Running Title: Postexercise lipid oxidation

27 28 1

Copyright © 2013 by the American Physiological Society.

29

ABSTRACT

30

A single bout of exercise can alter subsequent resting metabolism for many hours and into the next day.

31

However, differences between men and women, effects of nutritional state, and relative effects of resting

32

metabolic rate (RMR) and respiratory exchange ratio (RER) in controlling the increase in lipid oxidation

33

(Lox) after exercise are not yet clear. Effects of aerobic capacity (VO2peak) and exercise bout

34

parameters (intensity and volume) also remain to be clearly elucidated as does the time course of

35

changes after exercise. We performed a meta-analysis to assess these potential moderators of the impact

36

of endurance exercise (effect sizes, ESs) on subsequent Lox at rest (ES=0.91; 95% CI: 0.69–1.12) on the

37

day of exercise (ES=1.22; 95% CI: 0.89–1.55) and on the following day (ES=0.60; 95% CI: 0.35–0.85).

38

ES for the exercise-related increase in resting Lox was significantly greater in men than women in the

39

postabsorptive state but similar in the postprandial state. The ES for depression of RER after exercise

40

was similar between men and women, while the ES for RMR in the postabsorptive state tended to be

41

higher in men than women. Finally, VO2peak and energy expenditure of exercise (EEE), but not

42

intensity, were predictive of postexercise Lox. The findings indicate importance of EEE and fitness for

43

ability to achieve robust enhancement of Lox after exercise. The results additionally indicate a gender

44

difference in postexercise Lox that is dependent on nutritional state, as the ES for Lox was lower in

45

women only in the postabsorptive state.

46 47 48 49

Key words: Meta-analysis, fuel metabolism, substrate oxidation, sex, sex-based differences

50

2

51

INTRODUCTION

52

Maintenance of body composition over time, or prevention of fat gain, requires that individuals

53

oxidize at least as much fat as they consume. Thus, not surprisingly, a reduced relative contribution of

54

lipid oxidation (Lox) to energy metabolism is associated with increased risk of gaining weight over time

55

(59, 70). Along with the contribution of a positive energy balance, poor ability to oxidize lipids in

56

certain individuals could contribute to their propensity toward gain and retention of body fat. A

57

physically active lifestyle prevents age-related weight gain over time (33), and this effect could be

58

related to energy expenditure and to energy substrate selection.

59

Even in individuals who successfully meet current physical activity guidelines for total exercise

60

volume (33) and for whom exercise energy expenditure (EEE) thus significantly adds to total energy

61

expenditure (TEE), weight management through exercise alone can still be difficult. The non-substantial

62

impact of exercise on body weight may be because most of the TEE is still from the resting metabolic

63

rate (RMR); in individuals characterized as active, likely RMR is in the vicinity of 75% of TEE (33).

64

EEE remains a minor contributor to TEE, and so in attempts to monitor and manipulate energy substrate

65

metabolism, resting metabolism is a logical and important focus. Although energy substrate metabolism

66

is altered during exercise, the majority of the 24-hour period in individuals in Western society is spent at

67

rest. Exercise, however, can be a strategy to alter resting metabolism, as the effects of each individual

68

exercise bout on energy substrate utilization continue many hours into subsequent rest, with Lox

69

remaining elevated above that observed under sedentary conditions for many hours (25). That is to say,

70

resting lipid metabolism is altered by a recent exercise bout.

71

Previous work has modeled the relationship between exercise intensity and fuel partitioning

72

during exercise (7, 9, 10), but to our knowledge no model exists for the postexercise recovery period.

73

Extensive work has led to the conclusion that women rely upon lipid as a fuel to a relatively greater 3

74

extent than do men during an endurance exercise bout (13, 15, 16, 21, 22, 25, 26, 30, 56, 60, 62), and

75

this conclusion was reinforced by a meta-analysis on the topic (61). Additionally, some initial

76

observations indicate that there may be differences between men and women during the postexercise

77

recovery period (25, 30). As resting metabolism makes up the vast majority of TEE in most people, even

78

in those engaged in formal exercise training programs, we undertook an attempt to further explore the

79

determinants of postexercise Lox, including the effects of gender, physical fitness, nutritional state,

80

timing of assessment (immediately after, next day), exercise intensity, and EEE. Additionally, we

81

explored the components of the enhancement of Lox after exercise which can include an increase in

82

RMR and a relative shift away from carbohydrate and toward lipid for energy (i.e., reduced respiratory

83

exchange ratio, RER).

84

A variety of study designs have been used to investigate the effects of exercise on subsequent

85

substrate oxidation at rest. Investigations have been performed on both men and women, employing a

86

variety of exercise volumes and intensities, and subjects of a wide variety of fitness levels have been

87

studied after exercise in the postabsorptive (fasted) and postprandial (fed) states. Each individual study

88

may have some degree of statistical power limitation, and each pool of study participants represent only

89

a subset of variations in phenotype within society. Based on the limited statistical power inherent to any

90

one individual study, the ability to study important moderators of Lox or substrate utilization during the

91

postexercise window is limited. Therefore, we set out to harness these variabilities in study designs and

92

to provide a statistically powerful test beyond that of any individual study by employing meta-analytic

93

and meta-regression approaches to identify the determinants of postexercise Lox.

94 95 96

METHODS 4

97

Study aim. We performed a systematic quantitative analysis of published results on energy substrate

98

utilization data derived from pulmonary gas exchange. Specifically, we explored the impact of each

99

single endurance exercise bout in humans, and the moderating influence of gender, physical fitness,

100

nutritional state, timing of assessment (immediately after, next day), exercise intensity, and EEE.

101 102

Identification of data for the analysis. Relevant papers were identified by searching the PubMed

103

database using various combinations of the following terms: prior exercise, postexercise, post-exercise,

104

lipid oxidation, fat oxidation, substrate oxidation, substrate utilization, RER, and RQ. Additional papers

105

that might have been relevant for inclusion were identified from reference lists from all articles

106

identified from the PubMed search. Studies in which endurance exercise was performed at a steady

107

intensity were included (i.e., not interval exercise or resistance exercise). Because energy substrate

108

metabolism changes with the time of day and time elapsed since the prior meal (25, 30, 53), we only

109

included studies that utilized a time-matched sedentary control condition performed on a separate

110

occasion from that of exercise in the same subjects (i.e., cross-over design). In studies where

111

postexercise nutrition was provided, only those that fed the same energy and macronutrient content

112

under both exercise and sedentary conditions were included. There were not a sufficient number of

113

studies to allow for a separate meta-analysis of those in which energy intake was different between

114

exercise and sedentary conditions, but these individual studies are discussed separately from the main

115

findings in this report. The minimum duration of postexercise recovery was chosen to be two hours in

116

order to observe persistent changes in resting metabolism. Based upon knowledge that men and women

117

can respond differently to exercise (25), and to address a primary aim of investigating gender differences

118

in substrate oxidation, we excluded any papers that pooled men and women into one single group. In

119

the one study in which women were studied in both the follicular and luteal phases of the menstrual 5

120

cycle (20), data from the follicular phase were used in the analysis because this phase is often chosen

121

when standardizing cycle phase in exercise and substrate utilization research. Mean and variance values

122

for postexercise Lox and for Lox in sedentary control trials were obtained from published papers. When

123

data were displayed graphically but not numerically, authors were contacted to request the mean and

124

variance values, and these data were included in the analysis when authors agreed to provide the results.

125

Mean and variance for RMR and RER, because they are the components in derivation of Lox values,

126

were also obtained as available. Data were categorized as those collected on the day of exercise (on the

127

same calendar day), or on the day after exercise (after an overnight period had passed after exercise).

128

Data were also categorized into those collected in the postprandial (a meal taken immediately before

129

indirect calorimetry assessments) or the postabsorptive state.

130 131

Calculations. Exercise duration and relative exercise intensity (% VO2peak) were recorded for each

132

study, as well as energy expenditure of exercise (EEE). When EEE was not reported, it was estimated

133

from published oxygen consumption (VO2) rate and exercise duration, estimating 5 kcal energy per liter

134

of VO2 (8). If exercise VO2 was not reported, from the published mechanical power output (watts), the

135

energy requirement of exercise was estimated from a published table to convert power output to

136

metabolic equivalents (METS) (1). As needed, percentage of energy from fat and carbohydrate

137

(assuming negligible protein oxidation) was calculated from published equations (43). When not

138

provided, RMR was calculated using these percentages of fat and carbohydrate, published VO2, and the

139

assumption of 5.05 kcal/LO2 for carbohydrate and 4.7 kcal/LO2 for fat (8). When standard error (SE)

140

but not standard deviation (SD) were available, SD was calculated by multiplying SE by the square root

141

of the reported sample size. Effect sizes (ESs) for postexercise Lox were calculated as followed:

142

[(postexercise mean) – (sedentary mean)] / (sedentary SD)]. 6

143 144 145

Statistical analysis. Descriptive statistics were generated using SPSS v. 19 and are expressed as mean ± SE. ESs

146

were calculated and meta-analyses were conducted using Comprehensive Meta-Analysis v. 2.0. Cohen’s

147

d was used as the measure of the overall effect and evidence suggests that Type I error rates for tests of

148

heterogeneity are well controlled using this metric (32). Given that sample size impacts the precision of

149

the ES estimate, each ES was weighted by the inverse of its variance prior to conducting further

150

analyses. A random effects model was used to pool effects and compare results across studies to address

151

potential concerns regarding the lack of independence of data points when multiple effects are derived

152

from a single study (41). Overall ESs for Lox, RER, and for RMR were conducted independently to

153

examine subject and exercise characteristics on these outcomes.

154

Heterogeneity in meta-analysis refers to the variation in outcomes between studies included in

155

the review. The extent of heterogeneity partly determines the difficulty in drawing overall conclusions

156

from the literature. Understanding the cause of heterogeneity, in part through the use of moderating

157

variable analyses, can increase both the scientific and clinical value of the meta-analysis (64). Detecting

158

heterogeneity among ESs supports the necessity of follow-up moderating analyses. Heterogeneity was

159

examined using Cochran’s Q and I2 statistics. The traditional approach for assessing heterogeneity in

160

meta-analyses has been through the use of Cochran’s Q statistic (32, 55). Cochran’s Q is computed by

161

summing the squared deviation of each study ES from the overall ES estimate, weighting the

162

contribution of each ES by the inverse of its variance. The alpha value for statistical significance for Q

163

was set at P ≤ 0.10 because this statistic tends to suffer from low power (24, 34). More recently

164

Cochran's Q has been shown to be heavily influenced by the number of studies in a review such that it

165

has low power with a small number of studies and excessive power with a large number of studies (28). 7

166

Thus, the Q index provides a test of the presence of heterogeneity among ESs along with a

167

corresponding P statistic, but not the exact extent of heterogeneity within the meta-analysis (28). The I2

168

statistic, which expresses the ratio of the observed variance between outcomes to the total observed

169

variance in ESs (32), has been recommended along with Q to describe the extent of between-studies

170

variability, and was calculated as (Q - df)/Q x 100, where df equals the degrees of freedom. The I2

171

statistic was interpreted as small (25% to 0.10. Women

217

in the postabsorptive state exhibited an ES of 0.50±0.20 while in the postprandial state the ES for Lox

218

was 1.04±.36, a difference that was not statistically significant, Q(1)=1.8, P = 0.18. Additionally, a

219

general time course was observed in which a greater ES for Lox occurred immediately following

220

exercise (1.22±.17) when compared to that observed the day following exercise (0.60±0.13), Q(1)=8.51,

221

P < 0.05.

222

Respiratory exchange ratio. The overall ES for RER was 0.82±0.10, which was significantly

223

different from zero. The Q and I2 statistics revealed a lack of heterogeneity across effects, Q(27)=23.96,

224

I2=0, P > 0.10. Despite this lack of heterogeneity among the ESs, we examined potential moderators

225

based on a priori study hypotheses and to determine the relative role of RER in postexercise substrate

226

oxidation. No significant difference in RER was observed between men (ES = 0.83±0.12) and women

227

(ES = 0.80±0.16). There was also no significant difference in ESs as a function of

228

postprandial/postabsorptive state of the sample. RER effects for men did not differ from women in either

229

the postabsorptive (ES = 0.84 ±.13 vs 0.82±0.21) or postprandial (ES = 0.81±0.43 vs 0.85±0.35) state,

230

respectively. Notably, postprandial state RER data were only attained for men (20, 66) and for women

231

(14, 20) in two studies each, whereas a far more extensive RER data set was analyzed for the

232

postabsorptive state. A trend emerged indicating larger RER effects in the immediate recovery period

10

233

following exercise (ES = 0.99±0.15) relative to those observed the next day (ES = 0.65±0.15),

234

Q(1)=2.54, P = 0.11. Resting metabolic rate. ESs for metabolic rate were derived from 12 studies. The overall

235 236

observed ES was 0.26±0.10, which was significantly different from zero. The Q and I2 statistics revealed

237

a lack of heterogeneity across effects, Q(22)=12.13, I2=0, P > 0.10. However, similar to RER, follow-up

238

tests were conducted to assess the relative role of RMR in postexercise substrate utilization. Women had

239

less accentuation of metabolic rate postexercise (ES = 0.06±0.16) relative to men (ES = 0.38±0.13), and

240

this trend approached but did not reach statistical significance, Q(1)=2.51, P = 0.11. In terms of the

241

effects of feeding, overall average ES for the postabsorptive state was 0.29±0.11 while in the

242

postprandial state the impact of exercise on RMR was smaller with an ES of 0.05±0.26, Q(1)=0.73, P >

243

0.10. In the postabsorptive state, men exhibited a significantly larger ES for RMR (ES = 0.51±0.15) than

244

women (ES = 0.05±0.17), Q(1)=3.94, P < 0.10. No significant sex difference was found in the

245

postprandial state, Q(1)=0.02, P > 0.10. However, only three effects emerged from samples in the

246

postprandial state with one of the observations taken immediately after exercise in overweight women

247

(14) while the other two were assessed on the following day in men (23, 66). When analyzing RMR

248

data with postprandial and postabsorptive states pooled together immediately postexercise, men

249

expressed a larger effect for metabolic rate (0.59±0.19) than women (0.11±0.19), Q(1)=3.0, P < 0.10.

250

This effect was not statistically different the following day, although the trend remained with men

251

(0.29±0.20) retaining higher rates than women (-0.07±0.29), Q(1)=1.0, P > 0.10. No statistically

252

significant effects were observed for exercise on RMR during the immediate postexercise recovery

253

period (ES = 0.39±0.15) relative to values measured the following day (ES = 0.17±0.16), Q(1)=0.96, P

254

> 0.10.

255

Regression Analyses 11

256

Separate regression analyses were conducted using Lox ESs as the dependent variable with

257

exercise intensity (% VO2peak) and EEE (kcal) as predictors in meta-regression analyses to determine

258

the impact of these exercise characteristics on postexercise Lox. No evidence of ES moderation was

259

found for exercise intensity (QR= 0.79; P>0.05; R2 = 0.05; QE = 33.47; P>0.05). However, the overall

260

meta-regression model was found to be significant for EEE (QR= 8.89; P0.05). Thus, overall EEE (β = 0.42; z = 2.98; P < 0.01) but not exercise intensity (β = 0.22; z = 0.89;

262

P>0.05), accounted for significant variation in the overall effect of endurance exercise on subsequent

263

Lox. We also ran these meta-regression analyses by gender to determine whether this moderation of

264

EEE on postexercise Lox exists for both men and women. EEE was a significant predictor of

265

postexercise Lox for men (QR= 6.20; P0.05) but not for women (QR=

266

0.32; P>0.05; R2 = 0.001; QE = 16.12; P>0.05). Similar to the overall model, exercise intensity did not

267

significantly moderate postexercise Lox in men (QR= 1.51; P>0.05; R2 = 0.14; QE = 19.82; P>0.05) or

268

women (QR= 0.01; P>0.05; R2 = 0.01; QE = 11.38; P>0.05). However, these gender-specific analyses

269

were underpowered and thus results should be interpreted with caution.

270

Separate meta-regression analyses were conducted using Lox ESs as the dependent variable with

271

VO2peak as a predictor to determine whether postexercise Lox is greater among more aerobically fit

272

subjects. For total VO2peak, expressed in L/min, the test for linear regression with Lox was significant,

273

QR= 3.86; P

Determinants of resting lipid oxidation in response to a prior bout of endurance exercise.

A single bout of exercise can alter subsequent resting metabolism for many hours and into the next day. However, differences between men and women, ef...
247KB Sizes 0 Downloads 0 Views