Use-dependent corticospinal excitability is associated with resilience and physical performance during simulated military operational stress

Abstract

Simulated military operational stress (SMOS) provides a useful model to better understand resilience in humans as the stress associated with caloric restriction, sleep deficits, and fatiguing exertion degrades physical and cognitive performance. Habitual physical activity may confer resilience against these stressors by promoting favorable use-dependent neuroplasticity, but it is unclear how physical activity, resilience, and corticospinal excitability (CSE) relate during SMOS. To examine associations between corticospinal excitability, physical activity, and physical performance during SMOS. Fifty-three service members (age: 26 ± 5 yr, 13 women) completed a 5-day and -night intervention composed of familiarization, baseline, SMOS (2 nights/days), and recovery days. During SMOS, participants performed rigorous physical and cognitive activities while receiving half of normal sleep (two 2-h blocks) and caloric requirements. Lower and upper limb CSE were determined with transcranial magnetic stimulation (TMS) stimulus-response curves. Self-reported resilience, physical activity, military-specific physical performance (TMT), and endocrine factors were compared in individuals with high (HIGH) and low CSE based on a median split of lower limb CSE at baseline. HIGH had greater physical activity and better TMT performance throughout SMOS. Both groups maintained physical performance despite substantial psychophysiological stress. Physical activity, resilience, and TMT performance were directly associated with lower limb CSE. Individual differences in physical activity coincide with lower (but not upper) limb CSE. Such use-dependent corticospinal excitability directly relates to resilience and physical performance during SMOS. Future studies may use noninvasive neuromodulation to clarify the interplay among CSE, physical activity, and resilience and improve physical and cognitive performance.

NEW & NOTEWORTHY We demonstrate that individual differences in physical activity levels coincide with lower limb corticospinal excitability. Such use-dependent corticospinal excitability directly relates to resilience and physical performance during a 5-day simulation of military operational stress with caloric restriction, sleep restriction and disruption, and heavy physical and cognitive exertion.

INTRODUCTION

Resilience can be defined as the ability to maintain physical performance in the presence of physical or psychological stress (1, 2). Efforts to study resilience have traditionally emphasized pathophysiological aspects of stress-related conditions such as posttraumatic stress disorder and depression (3). Yet rather than the absence of dysfunction, there is growing evidence that resilience also involves the presence of active processes that mitigate deleterious stress responses or facilitate recovery thereafter (2, 4, 5). Military operations provide a useful model to better understand resilience in humans, as the collective stress of exhaustive physical and cognitive exertion, caloric restriction, and sleep disruption/restriction can substantially degrade performance (6–8). Although many of the consequences of military operational stress are well-established, we are only beginning to understand the factors that determine resilience (9).

In addition to its well-established cardiovascular and musculoskeletal benefits, physical activity promotes brain adaptations that are associated with resilience (2, 10). During acute exercise for example, the release of brain-derived neurotrophic factor (BDNF) (11, 12), cortisol (13), and other neuroendocrine factors (e.g., growth hormone) (14, 15) helps maintain homeostasis (16–18). When performed habitually, physical activity is associated with increased gray matter volume (19) and an improved capacity for neuroplasticity (10, 20), which may be particularly important when allostasis is considered as a central aspect of resilience (2, 21). As a noninvasive brain stimulation technique, transcranial magnetic stimulation (TMS) is often used to assess neuroplasticity in humans based on changes in corticospinal excitability – the balance between excitatory and inhibitory activity within the corticospinal tract (22, 23). Corticospinal excitability is sensitive to stress-related hormones (24) and behavioral events, including disruptions in sleep (25–27), sympathetic activity (28), as well as acute and habitual physical activity (29–31). Causal evidence from repetitive TMS also indicates that learning is accompanied by increases in corticospinal excitability (32, 34). Thus, individual differences in corticospinal excitability may provide an objective means to assess interactions between use-dependent plasticity and resilience during military operational stress (33).

Given that physical activity induces neuroplastic changes in brain structure and function, differences in chronic physical activity are likely to result in distinct corticospinal excitability patterns. Use-dependent corticospinal excitability patterns may also coincide with individual differences in physical performance (i.e., resilience) during military operational stress. We thus examined the interplay between corticospinal excitability, physical activity, and resilience during a five-day simulated military operational stress (SMOS) protocol with an ecologically valid combination of stressors including exhaustive physical and cognitive exertion, caloric restriction, and sleep restriction and disruption. Corticospinal excitability was assessed in the legs, which were heavily stressed during SMOS, as well as a commonly studied hand muscle that was not targeted during the intervention. Based on previous work describing use-dependent increases in corticospinal excitability as a central component of motor learning (32, 34), we hypothesized that individuals with greater habitual physical activity levels would display greater corticospinal excitability and that such differences would directly relate to resilience during SMOS.

METHODS

Participants and Experimental Design

Fifty-three healthy adults (n = 53, 13 women) participated in this study (DoD: W81XWH-17-2-0070). Inclusion criteria included an age of 18–41 yr and active or recent (within 2 yr) military service (Army, Marines, Navy, Air Force or Reserve Officer Training Corps). Exclusion criteria included contraindications to TMS (35), history of seizure, substance abuse or disorder, recent musculoskeletal injury or weight gain (>10% since discharge), past or current psychotic or bipolar disorder (or significant subthreshold symptoms thereof), untreated or unstable medical conditions, hormone replacement therapies other than birth control, and severe psychiatric distress associated with marked impairments in functioning. Caffeine intake was prohibited throughout the study to reduce any potential influence on corticospinal excitability (36) or physical performance (7). Potential participants were recruited using in-person meetings, electronic correspondence, or flyers and contacted the study staff to determine eligibility. If eligible, written informed consent was provided. All study procedures were approved by the University of Pittsburgh Institutional Review Board and in alignment with the Declaration of Helsinki.

Experimental Approach

Participants completed a SMOS protocol over a five-day and -night period, during which mood, physical performance, endocrine factors, and corticospinal excitability were assessed as summarized in Fig. 1A. Comprehensive descriptions of the study design are provided elsewhere (37, 38). Briefly, questionnaires were completed during intake and all testing procedures were practiced during a familiarization day (D0; not shown in Fig. 1). During the baseline and recovery days (D1 and D4), participants were allotted 8 h of sleep and 100% total daily energy expenditure (TDEE) via standardized meals, whereas during the intervention (D2 and D3) participants were allotted 50% TDEE and two 2-h blocks of sleep (4 h total) each night (Fig. 1B). Sleep timing and duration were monitored via polysomnography in accordance with American Academy of Sleep Medicine guidelines (39).

Figure 1.

Experimental procedures. A: all participants completed five consecutive days of simulated military operational stress (SMOS). Participants were familiarized with the study procedures (Day 0) and then completed baseline testing (Day 1), 2 days of SMOS (Days 2 and 3), and a recovery day (Day 4). Every morning, we assessed mood. Throughout SMOS, we assessed physical performance, endocrine factors, and corticospinal excitability. B: during each day of SMOS, participants received 50% of their total energy expenditure estimate and 4 h sleep (in two 2-h blocks). During baseline and recovery testing, participants received 8 h sleep and individualized meals to meet their estimated total daily energy expenditure requirement. Caloric intake on Day 4 does not reflect total daily energy expenditure (participants discharged in the late afternoon). Solid lines show time asleep, dashed lines show caloric intake; N = 53. Data are means ± SD. No statistical analysis was performed; descriptive statistics only.

Questionnaires

After consent, participants completed the Connor-Davidson Resilience Scale (CD-RISC) (40) and provided information about their military service history. The profile of mood states (POMS) (41) was used to determine mood and was reassessed every morning throughout SMOS. Scores were first determined in six domains (anger, confusion, depression, fatigue, tension, and vigor), after which total mood disturbance was quantified as the sum of all negative POMS scores plus the nonpositive remainders of the vigor scale [Anger + Confusion + Depression + Fatigue + Tension + (maximum possible Vigor – actual Vigor)] (42).

Physical Activity

Participants self-reported weekly session frequency, duration per session, and ratings of perceived exertion (Borg scale 6–20; RPE) (43) for each of the following physical activity types during the past 3 mo (when regularly performed): bicycling, calisthenics, circuit training, moderate sports (e.g., volleyball, golf, walking carrying clubs), rowing, running, swimming, vigorous sports (e.g., basketball, racquetball, soccer and tennis), walking with and walking without carrying a load, yoga, weight training, and wrestling. For each activity, weekly physical activity minutes were derived by multiplying the weekly frequency and duration per session. Activity-specific metabolic equivalents (MET) were calculated by multiplying weekly physical activity minutes with MET weighting factors derived from the 2011 Compendium of Physical Activities (44–46). Total MET minutes per week were then expressed as the sum of all activity-specific MET minutes.

Body Composition and Aerobic Fitness

During familiarization, body composition and TDEE were determined via air-displacement plethysmography (BodPod Body Composition Systems, Life Measurement Instruments, Concord, CA). Peak oxygen consumption (V̇o_2peak) was determined via indirect calorimetry (Parvo TrueOne, Salt Lake City, UT) during a standard incremental Bruce treadmill protocol (Woodway, Waukesha, WI).

Physical Performance

Every day at approximately noon, participants performed a strenuous tactical mobility test (TMT) reflective of common military occupational tasks. The TMT included the following independently scored subtests: water can carry, fire and movement drill, casualty drag, loaded and unloaded 300 m shuttle run, and a 4-mi ruck march (Supplemental Fig. S1; see https://doi.org/10.6084/m9.figshare.17062028). Each event was performed in series with minimal rest between. A comprehensive description of the TMT is provided here (38). To generate a composite TMT score, each subtest score (time to complete subtest) was converted into a z-score and summed except for the water can carry (velocity during the subtest), for which z-scores were first inverted to allow for consistent directional coding. Negative TMT composite scores indicate better overall performance.

Endocrine Factors

Blood was drawn before and after the TMT from an antecubital vein (6 mL serum and EDTA plasma tubes, Becton, Dickinson and Company, Franklin Lakes, NJ) for the analysis of cortisol, BDNF, human growth hormone (GH), and insulin-like growth factor I (IGF-I) via standard enzyme-linked immunoassays. Serum (cortisol, GH, IGF-I) was allowed to clot for 30 min at ambient temperature and plasma (BDNF) was placed on ice before being transported for processing. Blood samples were centrifuged at 1,500 g and 4°C for 15 min within 30–45 min of collection. The serum/plasma supernatant was aliquoted into tubes and frozen at −80°C until analysis. All assays were performed in duplicate with intra-assay coefficients of variation of 10% or less. Full results on the endocrine markers were previously reported in a larger (n = 69) study investigating sex-specific differences in neuromuscular and hormonal responses to military operational stress (38).

Corticospinal Excitability

CSE was assessed each day between 14:00 and 21:00 h based on motor-evoked potential (MEP) stimulus-response curves obtained from the dominant vastus lateralis (VL) – a quadriceps muscle involved in all physical performance assessments – and the first dorsal interosseus (FDI) – a commonly studied hand muscle that was not targeted in the intervention. Although time-of-day varied between participants, assessment timing was standardized within-subject (±2 h). Hand- and foot dominance were determined via the Edinburgh Handedness Inventory (47) and the Waterloo Footedness Questionnaire (48), respectively. Three (n = 3) participants were left-handed and seven (n = 7) were left-footed. The procedures are summarized below, with full details and reliability statistics reported here (49, 50).

Wireless active Ag differential parallel-bar electromyography sensors (bandwidth: 20–45 0Hz, interelectrode spacing: 10 mm; Trigno Avanti, Delsys Inc., Natick, MA) were placed over the VL and FDI in accordance with SENIAM guidelines (51). To ensure optimal signal to noise ratios, skin preparation included shaving hair, removing dry skin with adhesive tape, and isopropyl alcohol application. Before testing, adequate signal quality was verified and sensor location marked with indelible ink for consistency across days. Signals were digitized at 2,000 Hz and a gain of 500.

Hotspots for the motor cortex (M1) representations of the dominant FDI and VL were determined or reconfirmed (D2–D4) sequentially at 75% (FDI) and 65% (VL) of maximum stimulator output. Coil position was adjusted in 1–2 cm increments (anterior-posterior and medio-lateral) until the scalp location that consistently produced the largest peak-to-peak MEP amplitude was found. For the FDI, stimulation was applied with a D70² figure-of-eight coil oriented at a 45° angle to the longitudinal fissure. For the VL, a 110 mm double-cone coil (both The Magstim Company Ltd, Carmarthernshire, UK) was placed parallel to the longitudinal fissure. Each coil was thus positioned to induce a posterior-anterior/anterior-posterior current along the precentral gyrus (35). Coil positioning was monitored and recorded using frameless stereotaxy (Brainsight v2.4, Rogue Research, Inc., Montreal, Quebec, Canada), with each coil and participant referenced into Montreal Neurological Institute (MNI) space using cranial landmarks (nasion, left- and right preauricular areas) and reflective markers.

Next, a series of four 3–5 s maximum voluntary contractions (MVC) were completed during bilateral isometric index finger abductions (FDI) and knee extensions (VL) against load cells (FDI: MB-100, VL: SSM-AJ-500, Interface Inc, Scottsdale, AZ). Participants then performed bilateral isometric contractions at 15% MVC for eight ∼25 s sets (30 s rest between) with visual feedback to maintain consistent force levels. Five single biphasic TMS pulses were delivered each set (total number of pulses = 40; Super Rapid² Stimulator, The Magstim Company Ltd, Carmarthernshire, UK) with an interstimulus interval of ∼0.2 Hz. Stimulation was delivered in random 5% increments from 5 to 100% of maximum stimulator output (Fig. 2B). MEPs were determined as the peak-to-peak EMG amplitude during the 15–65 ms poststimulus interval (Fig. 2C) and least-squares fitted to a Boltzmann sigmoidal curve using nonlinear regression. Corticospinal excitability was then determined as the plateau of the Boltzmann sigmoid (MEP_MAX) and extracted for further analysis. The selected recruitment curve procedure provides efficient and reliable (ICC = 0.92 for MEP_MAX) characterization of corticospinal input-output properties (49, 52).

Figure 2.

Experimental protocol to derive corticospinal excitability stimulus-response curves and median split. A: corticospinal excitability was assessed in the dominant first dorsal interosseus and vastus lateralis in n = 53 individuals following a series of maximum voluntary contractions (MVCs). B: for each muscle, a total of 40 pulses were delivered during eight 25 s sets of isometric contractions at 15%MVCs, with five pulses applied at random 5% increments of stimulator output (range: 5%–100%) per set. C: representative motor-evoked potentials (MEPs) at select percentages of stimulator output were (D and E) fitted to a Boltzmann sigmoidal curve to derive muscle-specific stimulus-response-curves for each day. Corticospinal excitability was then determined as the plateau (MEP_MAX) of the stimulus-response-curve. F: neither lower nor upper limb corticospinal excitability changed throughout the intervention, but a persistent bimodal distribution of lower limb MEPs was evident. As a result, we divided the sample into individuals with HIGH (n = 27) and LOW (n = 26) corticospinal excitability of the VL using a median split of baseline (Day 1) MEP_MAX.

A two-way repeated measures analysis of variance (ANOVA; 2 muscles × 4 days) indicated that neither VL nor FDI corticospinal excitability (MEP_MAX) differed during SMOS (F_3,101.2 =0.6, P = 0.65; Fig. 2, D and E), but a consistent bimodal distribution was evident for the VL. We thus performed a median split of baseline (D1) VL MEP_MAX to separate participants into groups with HIGH (n = 26) and LOW (n = 27) corticospinal excitability (Fig. 2F). To confirm the stability of group assignments throughout SMOS, we repeated median splits for each day and compared groupings. Only four participants switched groups after D1 (Supplemental Fig. S2; see https://doi.org/10.6084/m9.figshare.16587164).

Data and Statistical Analysis

All data analysis and visualization were performed in R (53) and SPSS (Version 25; IBM, Armonk, NY). Missing data were imputed via multivariate imputation by chained equations using random forest with 100 iterations repeated five times (54) (see Supplemental Table S1; see https://doi.org/10.6084/m9.figshare.16587335). After data imputation, outliers (6%) were removed based on the interquartile rule (> 1.5 × IQR above the third or below the first quartile) and visual inspection of boxplots. Physical activity levels were further correlated with maximum oxygen consumption to cross-validate the accuracy of self-reports (r = 0.42, P < 0.01). After inspection of the scatter plot, one individual was removed due to a mismatch between (high) self-reported physical activity and (low) peak oxygen consumption (Supplemental Fig. S3; see https://doi.org/10.6084/m9.figshare.16587182).

Chi-square tests (χ²) were used to determine the association between group assignment and sex. Participant demographics were compared using independent t tests. Physical activity was compared between groups using a mixed-model ANOVA (2 groups × 13 activity types). Data acquired during SMOS were compared with one between- (2 groups) by one within-factor (4 days) repeated-measures ANOVA (e.g., performance and mood) or one between- (2 groups) by two within-factor (4 days × 2 muscles or 2 timepoints) repeated-measures ANOVA (e.g., MEPs, force, EMG, biomarkers). To reduce the potential influence of subcutaneous fat on EMG outcomes, all comparisons involving EMG (e.g., corticospinal excitability and muscle activity) included body fat percentage as a covariate. For all ANOVAs, Greenhouse–Geisser corrections were used when Mauchley’s test indicated nonsphericity. When significant, interactions were followed up with Bonferroni-corrected pairwise comparisons. Partial eta squared (${\eta }_p^2$) values are shown for significant main and interaction effects. Levene’s test was used to assess homogeneity of variances and normality was assessed via Shapiro–Wilk tests. Nonnormality was evident for MEP_MAX, GH, and BDNF. All variables were normally distributed after logarithmic transformation. Pearson correlations were used to determine the interplay between physical activity, corticospinal excitability, physical performance, and resilience. For all tests, the level of significance was set at P < 0.05.

RESULTS

Demographics

Group characteristics are shown in Table 1. Briefly, HIGH were older and had more military experience, higher resilience scores, greater physical activity levels, and lower body fat.

Table 1.

Participant characteristics for individuals with high and low corticospinal excitability

	High (n = 26)		Low (n = 27)
	N/Mean	SD	N/Mean	SD	P value
Women	4		9		0.129
Age, yr*	27.4	5.3	24.0	4.5	0.015
Years of service, yr*	7.8	5.2	4.5	3.7	0.009
Resiliency score (CD-RISC)*	86.8	9.7	80.5	10.6	0.028
RMR, kcal	1,668.8	226.5	1,669.8	323.2	0.990
Fat percentage, %*	19.7	4.6	24.3	6.6	0.005
Fat mass, kg*	14.9	5.2	20.0	6.9	0.004
Fat free mass, kg	62.5	8.6	60.4	14.1	0.519
Body mass, kg	77.5	11.2	82.0	14.3	0.207
V̇O2, mL·kg−1·min−1	48.1	7.0	45.1	6.3	0.108

*Between-group difference based on independent t test at P < 0.05. For sex comparisons, a Chi-squared test was used.

CD-RISC, Connor-Davidson Resilience Scale; N, frequency; RMR, resting metabolic rate; V̇O₂, peak oxygen consumption.

Corticospinal Excitability

As expected, after a median split based on VL corticospinal excitability, there was a group × muscle interaction (F_1,41 = 9.4, P < 0.01, ${\eta }_p^2$= 0.19), with greater VL MEP_MAX in HIGH compared with LOW (P < 0.01) but similar values in the FDI (P = 0.24). Regardless of muscle, corticospinal excitability did not change during SMOS (F_3,123 = 0.55, P = 0.65; Fig. 3).

Figure 3.

Lower and upper limb corticospinal excitability during simulated military operational stress. A and B: lower (but not upper) limb corticospinal excitability was greater in HIGH (n = 26; blue) than LOW (n = 27; yellow), but neither were affected by simulated military operational stress overall. Data were logarithmically transformed and analyzed using a three-way repeated-measures ANCOVA (2 groups × 2 muscles × 4 days; covariate = body fat percentage). Asterisks (*) and vertical bars indicate a main effect of group (P < 0.05). Gray-shaded areas indicate the timing of the intervention. Data are means ± SD. FDI, first dorsal interosseous; MEP, motor-evoked potential; VL, vastus lateralis.

Physical Activity

For the weekly time spent physically active, mixed-model ANOVA revealed a main effect of activity (F_4.4,215.8 = 17.9, P < 0.01, ${\eta }_p^2$= 0.26; Fig. 4A) and a main effect of group (F_1,51 = 5.80, P = 0.02, ${\eta }_p^2$= 0.10; Fig. 4B). HIGH reported greater physical activity with walking, running, weight training, and loaded walking listed as the most frequent activities. Accordingly, HIGH displayed more MET minutes per week (t_40.5 = 2.43, P = 0.02; Fig. 4C). Ratings of perceived exertion (LOW: 15.6 ± 1.7, HIGH: 14.8 ± 3.8, t_35.8 = 0.94, P = 0.35) and the relative contribution of each activity to weekly physical activity did not differ between groups (Fig. 4D).

Figure 4.

Physical activity profile of individuals with high (HIGH) and low (LOW) corticospinal excitability. A: weekly time spent physically active differed between activity types, with weight training, walking, loaded walking, and running most frequently reported. B: LOW reported lower levels of physical activity, and this difference persisted after activity-specific differences in metabolic equivalents were considered (C). D: the relative contribution of each activity type to weekly physical activity did not differ. Boxplots in A demonstrate the median, 25th, and 75th percentile. Stacked bar charts in B and D show the mean physical activity and percentage contribution per activity type, respectively. Raincloud plot in C demonstrates the probability density distribution, boxplot, and individual datapoints for group-specific metabolic equivalent (MET) minutes per week. HIGH (n = 25) is shown in blue, whereas LOW (n = 27) in yellow in C. Data were compared using an independent t test (C) or a mixed-model ANOVA (2 groups × 13 physical activity subtypes). MMA, mixed-martial-arts. Asterisk (*) indicates significant main effect of physical activity type (A) and group (B) or a significant between-group difference (C).

Mood, Physical Performance, and Biomarkers during SMOS

Regardless of group, mood worsened (F_{4,188 = 21}.0, P < 0.01, ${\eta }_p^2$ = 0.29, Fig. 5A) throughout SMOS (F_1,47 = 3.0, P = 0.09, ${\eta }_p^2$ = 0.06) as a result of increases in tension, fatigue, and confusion (tension: F_2.63 = 5.8, P < 0.01, ${\eta }_p^2$ = 0.10; fatigue: F_3.3,169.5 = 26.5, P < 0.01, ${\eta }_p^2$= 0.34, confusion: F_4,204 = 12.9, P < 0.01, ${\eta }_p^2$= 0.20) and reduced vigor (F_3.46,176.55 = 29.1, P < 0.01, ${\eta }_p^2$= 0.36; Supplemental Fig. S4; see https://doi.org/10.6084/m9.figshare.16587248). Compared with LOW, HIGH displayed greater vigor (F_1,51 = 4.5, P = 0.04, ${\eta }_p^2$ = 0.08) and less confusion (F_1,51 = 7.1, P = 0.01; ${\eta }_p^2$= 0.12) but similar tension, fatigue, depression, and anger (tension: F_1,51 = 1.6, P = 0.21; fatigue: F_1,51 = 1.7, P = 0.19, depression: F_1,51 = 0.1, P = 0.75; anger: F_1,51 = 0.37, P = 0.55).

Figure 5.

Influence of simulated military operational stress on mood, military-specific physical performance, strength, and endocrine factors. A: mood (n = 53) worsened throughout simulated military operational stress (SMOS). B: HIGH performed better on the physical performance tests (TMT, n = 39), despite similar maximal isometric strength in the FDI (D) and VL (C) (n = 46). Neither strength nor physical performance declined during SMOS. E: nevertheless, canonical exercise-induced endocrine responses were evident from pre- to post-TMT Data were analyzed using a repeated-measures ANOVA with one between-group factor (2 groups) and one (3- or 4 days; mood, TMT, strength) or two (3- or 4 days × 2 timepoints; cortisol, IGF-I, BDNF, GH) within-group factors. GH and BDNF were log-transformed before analysis. The gray-shaded areas correspond to the SMOS intervention. HIGH (n = 26) are shown in blue, whereas LOW (n = 27) are shown in yellow. Asterisks (*) and horizontal/vertical bars indicate a main effect of day/group, whereas asterisks and brackets ( [ ) indicate main effects of time (pre- to post-TMT). Hashtags (#) indicate significant pairwise difference after Bonferroni correction. Data show means ± SD. For all tests P < 0.05. BDNF, brain-derived neurotrophic factor (n = 34), cortisol (n = 45); FDI, first dorsal interosseus; GH, growth hormone (n = 38); IGF-I, insulin-like growth factor I (n = 52); TMT, tactical mobility test; VL, vastus lateralis.

Each group maintained physical performance and strength throughout SMOS (TMT: F_2.0,72.6 = 1.6, P = 0.21, Fig. 5B; Strength: F_2.2,98.6 = 2.5, P = 0.08 Fig. 5, C and D), but HIGH outperformed LOW on the TMT (F_1,37 = 7.5, P < 0.01, ${\eta }_p^2$= 0.17) despite producing similar maximal force (F_1,46 =0.1, P = 0.83) in the FDI and VL. The TMT produced characteristic endocrine stress responses (Fig. 5E): circulating cortisol increased from pre to post exercise (F_1,44 = 73.3, P < 0.01, ${\eta }_p^2$= 0.63), but decreased from D1 to D4 overall (F_2,88 = 5.7, P < 0.01, ${\eta }_p^2$= 0.11). BDNF displayed a time × day interaction (F_1.5,47.2 = 6.7, P < 0.01, ${\eta }_p^2$= 0.17); pairwise comparisons confirmed an exercise-induced increase on D1 (P < 0.01) and D3 (P = 0.02) but not D4 (P = 0.84). GH increased from pre to post exercise (F_1,36 = 154.6, P < 0.01, ${\eta }_p^2$= 0.81) and in contrast to cortisol, increased throughout SMOS (F_2,72 = 4.4, P = 0.02, ${\eta }_p^2$= 0.11). In addition, GH was generally higher in LOW compared with HIGH (F_1,36 = 5.2, P = 0.03, ${\eta }_p^2$ = 0.13). Finally, IGF-I decreased throughout SMOS (F_2,100 = 33.7, P < 0.01, ${\eta }_p^2$ = 0.40) and was generally greater in LOW (F_1,50 = 4.9, P = 0.03, ${\eta }_p^2$ = 0.09).

Interplay between Corticospinal Excitability, Physical Activity, and Resilience

Pearson correlations indicated weak-to-moderate associations between VL CSE and physical activity (r = 0.29, P = 0.04), resilience (r = 0.38, P < 0.01), physical performance (r = −0.30, P = 0.04), military experience (r = 0.33, P = 0.01) and aerobic fitness (r = 0.32, P = 0.02) (Fig. 6, A–C). No such associations were detected for the FDI (physical activity: r = 0.10, P = 0.49; resilience: r = 0.05, P = 0.75; physical performance: r = −0.18, P = 0.24; experience, r = −0.01, P = 0.99; aerobic fitness: r = 0.05, P = 0.71).

Figure 6.

Physical activity, resilience, performance, and corticospinal excitability. A–C: higher corticospinal excitability of the vastus lateralis (VL) was associated with greater physical activity (n = 52), resilience (n = 52), and tactical mobility test (TMT) performance (n = 49). Associations were examined using Pearson correlation. CD-RISC, Connor-Davidson Resilience Scale; FDI, first dorsal interosseous; MEP, motor-evoked potential; MET, metabolic equivalent. Negative scores on the TMT performance test indicate better performance.

DISCUSSION

The primary aim of this study was to examine the association between corticospinal excitability (CSE), physical activity, and resilience during five days of simulated military operational stress (SMOS). Using a median split of vastus lateralis CSE at baseline, we classified individuals as HIGH or LOW. Individuals in HIGH were more resilient and physically active than LOW and consistently outperformed LOW on an ecological tactical mobility test (TMT). Each group maintained strength and physical performance despite substantial exercise-induced endocrine stress responses and worsening of mood. Higher CSE of the VL (but not FDI) was weakly to moderately but consistently associated with greater physical activity, resilience, and TMT performance. Together, these findings suggest that individual differences in corticospinal excitability coincide with habitual physical activity levels and that such use-dependent plasticity is directly related to resilience and physical performance during SMOS.

One novel finding of this study is that individuals with greater lower limb (VL) CSE (HIGH) performed better throughout SMOS and that differences in CSE were directly related to physical activity levels. A plethora of work has detailed the use-dependent properties of the human motor cortex (55–60), whereby use (or disuse) is accompanied by task-specific changes in cortical representations (61), gray matter volume (62), and corticomotor excitability (63). Accordingly and in contrast to our observation of a weak positive relationship between CSE and physical activity, recent work suggests that individuals with lower physical activity habits display greater corticospinal excitability (30, 31). This discrepancy may in part reflect differences between self-reported and accelerometry-based physical activity quantifications (31). Thus, although accelerometry-based physical activity measurements reduce issues with recall bias, the approach used in this study provided information on physical activity intensity (RPE) and mode (e.g., weight training, running, walking), which cannot be examined based on step counts alone. For example, weightlifting was one of the most frequently reported activities (not captured via step counts) and weekly physical activity minutes differed across activity types (Fig. 4). Moreover, physical activity levels are commonly assessed with questionnaires (10, 20) and the positive association between physical activity and peak oxygen consumption in this investigation broadly supports the validity of the self-reported physical activity estimates. Since there was no group × activity type interaction for physical activity, and self-reported RPEs did not differ between HIGH and LOW, our findings suggest that corticospinal excitability may more closely reflect the amount, rather than modality or intensity of activity per se. Future studies could clarify this idea by selectively recruiting participants with distinct physical activity habits or intensities, as the intricacies of motor tasks likely play an important role in the temporal dynamics of corticospinal activity (64).

Another salient finding is that in contrast to the vastus lateralis (VL), there was no association between CSE in the first dorsal interosseus (FDI) and physical activity or physical performance. Muscle-specific associations between physical activity and CSE were reported previously (31) and there is growing evidence that CSE is task- and muscle-specific (49). In this study, physical exercise stress consisted of a series of military-specific tests that emphasized lower body muscles, including the fire and movement drill, casualty drag, loaded- and unloaded 300 m shuttle run as well as a 4-mile ruck march (38). Although the VL was heavily involved in all physical performance assessments (and likely habitual physical activity behavior), the FDI was not directly targeted by the intervention but is a commonly studied hand muscle that produces internally valid corticospinal excitability estimates. Thus, muscle-specific associations between TMT performance and corticospinal excitability may reflect differences in use before and throughout SMOS. Although differences in CSE are often conceptualized in global terms (65), our results support the contention that at least some aspects of relations between CSE and behavior are use-dependent (31). These results also support the notion that nondeliberate (i.e., convenience-based) task selections for TMS assessments can obfuscate the behavioral implications of CSE (49).

Contrary to our hypothesis, SMOS did not influence maximum strength or performance despite caloric restriction, sleep restriction and disruption, and objective indications of substantial psychophysiological stress. Given that military operational stress can degrade physical performance (8), the lack of change could reflect effective coping or an insufficiently stressful intervention (15, 66). However, stress-related changes in endocrine factors (cortisol, GH and BDNF) and the worsening of mood throughout SMOS (Fig. 5) suggest that the former possibility is more likely. Both groups had resilience scores well above those associated with stress-related disorders (40, 67): LOW resembled the general U.S. population (40) and HIGH was comparable to long- and ultra-distance runners (68, 69) and other military populations and trainees (70, 71) (for additional comparisons see: http://www.connordavidson-resiliencescale.com/user-guide.php). As such, it appears that participants were largely resilient to the stressful yet relatively brief (48 h) SMOS intervention. More severe disruptions in mood are evident with prolonged stress exposure (6), so it remains uncertain if similar reductions in performance would occur with a longer SMOS intervention. Finally, an offsetting combination of learning-related improvements and stress-related decrements could explain the maintenance (but not improvement) of physical performance throughout the intervention. In this case, increased CSE in HIGH may reflect preexisting differences in physical activity that facilitate superior performance throughout SMOS.

Similar to performance, CSE did not change throughout SMOS regardless of group. This was surprising, as previous work found that CSE is sensitive to exercise (11), motor improvements (32), stress (28), and sleep deprivation (25). However, short-term (≤10 wk) changes in performance can occur in the absence of CSE changes (72–74) and the combination of SMOS-related decreases and motor improvement-related increases could again explain the absence of any net changes. Alternatively, due to logistical challenges associated with the execution of an ecological performance test, CSE was not assessed immediately after completion of the TMT; if the effects of the TMT on CSE were relatively transient (29), CSE may have returned to baseline levels by the time of assessment. As such, the chronic, day-to-day fluctuations of CSE, influenced by a blend of stressors in this study, likely differs from the corticospinal responses assessed immediately at the end of sleep deprivation (26) or after a bout of exercise (11).

At baseline, greater self-reported resilience was associated with increased VL CSE (Fig. 6) and HIGH generally displayed better mood (more vigor and less confusion) than LOW (Supplemental Fig. S2). Similar to regular physical activity, positivity is considered a nonpharmacological facilitator of resilience (2) and may promote brain health and performance during stress exposure. Thus, it is also possible that the elevated CSE in HIGH reflects a confluence of mood and physical activity, which would more closely mimic common resilience phenotypes (2). Indeed, even though both groups maintained their respective performance levels throughout SMOS, HIGH could be considered more resilient than LOW. For example, GH concentrations (indicative of physical stress) throughout SMOS were greater in LOW (Fig. 5). Also, HIGH self-reported greater resilience, had more military experience, and maintained superior performance throughout SMOS. Given that most TMT tasks were based on the participant’s best-effort, this latter finding is not insignificant. Nevertheless, because mood worsened in the absence of changes in military performance or corticospinal excitability, it is difficult to ascertain how interactions among corticospinal excitability, mood, and physical activity contributed to the overall maintenance of performance during SMOS. To clarify this notion, future studies could modulate corticospinal excitability of the lower limbs during stress exposure, include a control group or expose individuals with predetermined differences in resilience (e.g., stress-related conditions) to the same stressors.

Limitations

To increase ecological validity, military operational stress was simulated using a protocol with caloric and sleep restriction as well as rigorous physical and cognitive exertion. The combination of stress factors limits the ability to make strong inferences about the role of any individual factor. Given the nature of the study, the duration of SMOS was known to participants, which may not generalize to real-world scenarios, where a high degree of uncertainty may significantly increase or alter stress responses. Participants were tested in cohorts of up to four, which could introduce confounds related to social dynamics (e.g., competition or social facilitation). Time from TMT to CSE assessments varied between participants, but repeated measures were standardized to the extent possible (±2 h) and physical fatigue-induced changes in CSE typically recover quickly (29), thus minimizing the risk of nonstandardized brain states. Participants were separated solely based on corticospinal excitability of the VL at baseline, even though CSE was assessed throughout SMOS. Nevertheless, sensitivity analyses confirmed the stability of the groupings regardless of the day the median split was performed. To delineate supraspinal from spinal and peripheral contributions to CSE, MEPs are sometimes considered in tandem with muscle compound action- and cervicomedullary evoked potentials. Neither measure was assessed in this study, but we considered the influence of peripheral factors by incorporating body fat percentage as a covariate in all analyses. Physical activity levels were self-reported, which could introduce recall bias, but allowed for the analysis of physical activity subtypes, which would not be possible on the basis of step counts alone. Moreover, the association between physical activity levels and aerobic fitness supports the validity of this approach. Given the performance enhancing and sleep-altering effects of caffeine, the consumption of caffeine was prohibited. Withdrawal effects are possible, but the opposite possibility (i.e., performance bias toward caffeine users) was considered more problematic. Finally, future work may add a control group without stress exposure or noninvasively alter corticospinal excitability using repetitive TMS to clarify the nature of adaptive responses to SMOS and provide causal evidence for the association between corticospinal excitability and resilience.

Conclusions

Lower (but not upper) limb corticospinal excitability is higher in individuals who perform more physical activity, and this association directly translates to physical performance during SMOS. Because individuals with greater corticospinal excitability consistently outperformed those with lower corticospinal excitability and appeared more resilient, these findings suggests that habitual physical activity may confer resilience against stress by promoting favorable use-dependent neuroplasticity. Future work can determine how the modulation of corticospinal excitability in the lower limbs influences physical performance during stress exposure and thereby provide causal information on the interplay between use-dependent neuroplasticity and resilience.

SUPPLEMENTAL DATA

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.16587335.

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.17062028.

Supplemental Fig. S2: https://doi.org/10.6084/m9.figshare.16587164.

Supplemental Fig. S3: https://doi.org/10.6084/m9.figshare.16587182.

Supplemental Fig. S4: https://doi.org/10.6084/m9.figshare.16587248.

DATA AVAILABILITY

The raw data and material supporting the conclusions of this article will be made available by the authors upon request.

GRANTS

Funding for this study was provided by the Department of Defense (Award No.: W81XWH-17-2-0070). P.G.R. was supported in part by KBR’s Human Health and Performance Contract NNJ15HK11B through the National Aeronautics and Space Administration.

DISCLAIMERS

The authors of this report are entirely responsible for its content. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors, are not to be construed as official, and do not necessarily reflect those of the US Government, the Department of Defense, the Department of the Army, the National Aeronautics and Space Administration, the Department of the Navy, KBR, or Leidos. This material has been reviewed by the Walter Reed Army Institute of Research. There is no objection to its presentation and/or publication. The investigators have adhered to the policies for protection of human subjects as prescribed in AR 70-25.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

P.G.R., M.N.D., Q.M., A.G., C.C., B.C.N., and S.D.F. conceived and designed research; F.P., M.C.C., M.E.B., W.R.C., A.D.L., A.M.S., S.R.E., B.J.M., and A.J.S. performed experiments; F.P., M.C.C., M.E.B., W.R.C., A.D.L., A.M.S., S.R.E., B.J.M., and A.J.S. analyzed data; F.P., M.C.C., F.F., and S.D.F. interpreted results of experiments; F.P. and M.C.C. prepared figures; F.P. and M.C.C. drafted manuscript; F.P., M.C.C., M.E.B., W.R.C., A.D.L., A.M.S., S.R.E., B.J.M., A.J.S, P.G.R., M.N.D., Q.M., F.F., A.G., C.C., B.C.N., and S.D.F. edited and revised manuscript; F.P., M.C.C., M.E.B., W.R.C., A.D.L., A.M.S., S.R.E., B.J.M., A.J.S., P.G.R., M.N.D., Q.M., F.F., A.G., C.C., B.C.N., and S.D.F. approved final version of manuscript.