Adrenaline during mental stress in relation to fitness, metabolic risk factors and cardiovascular responses in young men

Abstract

We studied plasma adrenaline (A) in relation to physical fitness, metabolic cardiovascular risk factors and cardiovascular responses. Men (age 21–24 years) with high and normal (both n = 19) screening blood pressure (BP) were studied cross‐sectionally. We measured peak oxygen uptake (VO_2peak) (treadmill exercise), and plasma catecholamines, heart rate (HR), finger systolic (SBP) and diastolic (DBP) BP, and insulin‐adjusted glucose disposal rate (GDR/I) during a hyperinsulinaemic glucose clamp (rest) and mental arithmetic stress test (MST). By multiple regression, A at rest (A_rest) (β = 0.37, p<0.05) and during MST (A_mst) (β = 0.40, p<0.01) were associated with high screening BP. In the respective models, A_rest was negatively related to body mass index (BMI) (β = −0.56, p<0.001) and A_mst positively to VO_2peak (β = 0.54, p<0.001). BP and HR responses correlated positively with VO_2peak, but were determined by A_mst in multiple regression models. Independently of BMI and VO_2peak, serum high‐density lipoprotein cholesterol was positively related to A levels, whereas GDR/I was independently related only to VO_2peak. Increased adrenaline secretion may be related to high BP, but may at the same time be associated with a beneficial metabolic profile.

• Adrenaline
• blood pressure
• heart rate
• metabolic syndrome X
• physical fitness
• psychological stress

Introduction

Circulating adrenaline (A) is mainly derived from the adrenal medulla [1]. The release of A during stress is thought to improve mental and physical performance, and to be critical for the “fight‐or‐flight” response [2]. Plasma A correlates with blood pressure (BP) [3] and it is often increased in hypertension [4], apparently due to increased secretion [5], but its pathophysiological role is unclear [6]. Mental stress also increases plasma noradrenaline (NA) in young men [7–11], reflecting sympathetic neural activation [1]. According to the “adrenaline hypothesis” [12], A contributes to hypertension by increasing sympathetic transmission, and evidence suggests a link between stress, adrenal medullary activation and high BP [4]. We have repeatedly found correlations between plasma A and BP responses to mental stress in young men [7], [8], [11]. This may be of interest, since increased cardiovascular reactivity may be related to the development of hypertension [13].

Hypertension, obesity and insulin resistance frequently co‐exist, and they are all associated with increased sympathetic activity [14]. The latter partly explains why obesity predisposes to hypertension [15]. The risk of hypertension is also increased by low physical activity [16] and poor fitness [17], regardless of body mass index (BMI). However, increased fitness per se may not reduce sympathetic nerve activity [18], and the reduction of resting heart rate (HR) with exercise training is mainly due to increased vagal cardiac control [19]. Adrenal medullary activity may in fact increase with higher physical fitness [20]. It is important to note that sympathetic neural and adrenal medullary activities can be affected independently by distinct stimuli [2], [21]. Thus, in contrast to sympathetic neural activity, A secretion apparently decreases with increasing obesity [3], [10], [22–28] and is negatively associated with features of the metabolic syndrome [14], [27], [28]. Accordingly, and based on known adrenergic effects on metabolism, Reaven et al. [14] postulated low adrenal medullary activity as a contributor to the lipid abnormalities associated with insulin resistance.

Increased cardiovascular reactivity may be associated with poor fitness [29], but this is not a consistent finding [30]. Similarly, fitness is not consistently related to A responses induced by mental stress [30], [31]. Already published data [11] from the present groups of young men showed significant correlations between BP and plasma A responses to mental stress. When analysing pooled data from several subsets of young men, including the present groups, we found a negative relationship between BMI and plasma A [10]. However, physical fitness may have been a confounder, since higher BMI is consistently associated with lower physical activity [32]. We have shown that fitness is related to insulin sensitivity and vagal cardiac control, assessed from short‐term HR variability [33]. However, we have not addressed potential relationships between A levels and physical fitness or metabolic factors, or the relationship between fitness and cardiovascular responses, in these groups. Therefore, we aimed to compare A levels and responses in men with high and normal screening BP in multivariate models with adjustment for fitness. Based on the background summarized above, we hypothesized that the adrenal medullary response to mental stress is positively related to fitness. With the same reasoning, we also hypothesized that higher plasma A levels are associated with a beneficial lipid profile and insulin sensitivity. Finally, we assessed the relationships of fitness and plasma A with the plasma NA, HR and BP responses.

Methods

Participants

As described previously [11], [33], we studied men born in 1977–1978 who had systolic (SBP)⩾140 mmHg and diastolic BP (DBP)⩾90 mmHg (n = 20) or SBP⩽115 mmHg and DBP⩽75 mmHg (n = 21) at the Oslo military draft examination in 1996. The present data were obtained from the 19 subjects in each group who successfully completed a treadmill exercise test. They were healthy, used no regular medication, and fasted and refrained from smoking for the preceding 10 h and abstained from alcohol for the preceding 24 h. Examinations were conducted at constant room temperature in a quiet room. There were six and seven smokers among men with normal and high screening BP, respectively. In the respective groups, 11 and 10 subjects were weekly engaged in aerobic exercise (average 4 h/week), but this was not significantly related to fitness [33]. As described previously [11], we randomized nine additional men with high screening BP (BMI 24 kg/m², BP 129/79 mmHg at examination) to a saline infusion study (time control group). We obtained informed consent from each participant and approval by the regional medical research ethics committee. Sitting BP, waist circumference and BMI were measured as described previously [33] and are given in Table I. Correlations between fitness and insulin sensitivity [33] have been published, as have the catecholamine and BP responses to mental stress for the complete groups [11].

Table I. Group characteristics.

Variable	Screening BP		p
	⩾140/90 mmHg (n = 19)	⩽115/75 mmHg (n = 19)
Age (years)	22.2 (21–23)	22.6 (22–24)
Body mass index (kg/m^2)	25.8 (4.8)	23.0 (2.6)	0.03
Waist circumference (cm)	92.4 (11.5)	85.1 (7.6)	0.03
Systolic BP (mm Hg)	130 (10)	115 (8)	<0.0001
Diastolic BP (mm Hg)	80 (8)	68 (10)	0.0003
Pulse rate (beats/min)	67 (11)	63 (9)	0.16
Total cholesterol (mmol/l)	4.2 (0.7)	4.2 (0.7)	0.67
HDL (mmol/l)	1.2 (0.3)	1.2 (0.2)	0.71
Triglycerides (mmol/l)	1.6 (2.4)	0.8 (0.4)	0.22
Glucose (mmol/l)	5.0 (0.4)	4.8 (0.3)	0.14
Insulin (pmol/l)	50.4 (42)	34.6 (11)	0.42
GDR/I (AU)	1.9 (1.2)	2.3 (0.8)	0.20
Peak oxygen uptake (ml/min/kg)	45.8 (8.5)	48.3 (5.2)	0.29

Mean values (SD). Age not compared statistically (range in parenthesis).

BP, blood pressure; HDL, serum high‐density lipoprotein cholesterol; GDR/I, insulin‐adjusted glucose disposal rate; AU, arbitrary units.

Hyperinsulinaemic isoglycaemic glucose clamp and saline control infusion

We applied a 90‐min hyperinsulinaemic isoglycaemic glucose clamp [33]. After baseline blood sampling, insulin infusion (0.86 mU/min/kg body weight) was started and blood glucose was clamped at the fasting level by glucose infusion (200 mg/ml). Insulin sensitivity was expressed as glucose disposal rate (GDR) (final 20‐min glucose infusion rate/weight) divided by serum insulin at 90 min (GDR/I)×100 (arbitrary units) [8]. Clamp infusions were substituted with matched volumes of 0.9% saline in the blinded time control group.

Mental arithmetic stress test

The hyperinsulinaemic glucose clamp does not affect BP, HR or catecholamine responses to the mental arithmetic stress test (MST) [9]. Thus, for practical reasons, MST was carried out immediately after completion of the 90 min required for the glucose clamp, but with infusions maintained constant and participants supine. The participants were first briefly informed of the test. After 2 min of anticipation, they were then instructed to verbally subtract 13 serially from 1079 for 5 min. A metronome was used for distraction, and miscalculations were mentioned. The total stress period thus lasted for 7 min, and was followed by a 15‐min recovery period.

Cardiovascular measurements

R‐R intervals were measured with a Medilog FD4 ECG recorder (Oxford Instruments, Oxon, UK), as described previously [11]. HR was calculated from mean R‐R intervals during the entire pre‐stress period (30 min at baseline before infusions and the glucose clamp), and during MST and the recovery period. SBP and DBP were recorded on the right third finger with a finger BP monitor (Finapres, Ohmeda 2300, Englewood, CO, USA), as described previously [11]. We obtained 15‐min recordings at baseline and during the first, second and third 30‐min period of the clamp, and during MST. The Finapres allows reliable assessment of short‐term BP changes, but may be less accurate for absolute levels [34]. Thus, only BP changes (Δ) are presented.

Exercise test

Fitness was assessed as peak oxygen uptake (VO_2peak) during treadmill (Jaeger LE 300, Jaeger, Höchberg, Germany) exercise by a continuous incremental protocol, as described previously [33]. After 4 min of walking at 2.7 km/h, the treadmill was elevated by 1% and the speed increased by 0.6 km/h every 2 min until exhaustion. Respiratory gas exchange was measured breath‐by‐breath using an external volume sensor and the Oxycon Champion gas analyser (Jaeger, Höchberg, Germany). Peak gas exchange and HR values were defined as the highest 30‐s averages obtained. The test was considered satisfactory if respiratory exchange ratio⩾1.05 and HR⩾90% of age‐predicted maximum (208−0.7×age) [35].

Blood sampling and biochemical measurements

Baseline blood samples were drawn through an indwelling i.v. catheter after 45 min of rest. Further samples for catecholamines were drawn after 30, 60 and 90 min during the glucose clamp, within 1 min after announcement of MST, during minutes 1, 2 and 3 of the task, and the last minute of recovery. Serum glucose, triglyceride, total cholesterol and high‐density lipoprotein cholesterol (HDL) concentrations were measured with a Cobas Integra (Roche, Basel, Switzerland), blood glucose during clamp with an Accutrend® sensor (Boehringer Mannheim, USA), serum insulin with an enzyme‐linked two‐site immunoassay (DAKO Diagnostics Ltd., Cambridgeshire, UK), and plasma catecholamines with a radioenzymatic method [36], as described previously [37].

Statistical methods

We used SPSS 11.0 (SPSS Inc., Chicago, IL, USA). Catecholamine levels before (A_rest and NA_rest) and during MST (A_mst and NA_mst) were averaged over the 4 samples drawn in each situation. We also noted the baseline values, individual maximal A during MST (A_max), and recovery values. Changes (Δ) in all variables were calculated as the difference between mean values at rest and during MST. Non‐parametric tests were used unless normal distribution was achieved by natural logarithmic (ln) transformation. Groups were compared with Student's t‐test or the Mann–Whitney test. Stress responses were assessed by repeated measures analysis of variance. We used Pearson's (r), or when appropriate, Spearman's (r_s) correlation coefficients, and standardized regression coefficients (β) from forward stepwise multiple regression models, all adjusted for high screening BP (0 or 1), BMI, waist circumference, smoking status (0 or 1), and VO_2peak. Null hypotheses were rejected if p<0.05. Data are given as mean (SD).

Results

Adrenaline, fitness and measures of obesity

A levels did not differ significantly between the groups (Table II). A levels and ΔA correlated positively with VO_2peak in the combined groups (Table III) and within the high screening BP group (Figure 1). Correlations of BMI (r = −0.59, p<0.001) and waist circumference (r = −0.62, p<0.0001) with VO_2peak were paralleled by negative correlations with A levels (Table III). BMI correlated significantly with A only in the high screening BP group (Figure 2), but when the two men without satisfactory exercise tests were included, correlations with A_rest (r = −0.55, p<0.05) and A_mst (r = −0.54, p<0.05) were also significant in the normal group. As the dependent variable in multiple regression, A_rest was related to BMI and high screening BP, whereas A_mst and ΔA were related to VO_2peak and high screening BP (Table IV).

Table II. Plasma catecholamine and cardiovascular responses to mental stress.

	Baseline and glucose clamp		Mental stress test		Recovery
	H	N	H	N	H	N
Adrenaline (pmol/l)	224 (183)	163 (78)	376 (303)^d	228 (128)^b	225 (158)^c	173 (106)^a
Noradrenaline (nmol/l)	1.17 (0.41)	0.94 (0.32)	1.57 (0.66)^c	1.14 (0.40)^c	1.41 (0.57)^a	1.21 (0.49)
Heart rate (ms)	61 (9)	59 (8)	78 (10)^d	70 (12)^d	66 (9)^d	63 (9)^d
ΔSBP (mm Hg)			15.5 (15.8)	8.6 (11)
ΔDBP (mm Hg)			10.4 (10.8)	3.7 (6.1)

Mean values (SD).

H, high screening blood pressure; N, normal screening blood pressure; SBP, systolic blood pressure; DBP, diastolic blood pressure.

^ap<0.05, ^bp<0.01, ^cp<0.001, ^dp<0.0001 vs. previous measurement. p<0.0001 for effects by repeated measures analysis, except adrenaline (p<0.01), SBP (p<0.001), and DBP (p<0.05) in N. p = 0. 06 for group difference in Δadrenaline and stress adrenaline level; Δnoradrenaline, stress noradrenaline level and heart rate, and ΔDBP significantly higher in H (all p<0.05).

Table III. Correlations between plasma adrenaline and peak oxygen uptake and measures of obesity in the combined groups.

Variable	Plasma adrenaline
	Abas		ln Arest		ln Amst		ln Amax		ΔA		ln Arec
	rs	p	r	p	R	p	r	p	rs	p	r	p
VO2peak	0.18	NS	0.40	<0.05	0.47	<0.01	0.49	<0.01	0.38	<0.05	0.29	NS
ln BMI	−0.38	<0.05	−0.44	<0.01	−0.36	<0.05	−0.36	<0.05	−0.20	NS	−0.32	NS
ln WC	−0.33	<0.05	−0.39	<0.05	−0.37	<0.05	−0.38	<0.05	−0.29	NS	−0.27	NS

A, adrenaline; A_bas, baseline (before glucose clamp); A_rest, mean resting value; A_mst mean value during mental stress test; A_max, maximal value (individual); ΔA, change from mean resting value; A_rec, recovery; ln, natural logarithm; r_s, Spearman's rank correlation coefficient; VO_2peak, peak oxygen uptake; BMI, body mass index; WC, waist circumference.

Table IV. Stepwise multiple regression analyses with plasma adrenaline as the dependent variable.

Explanatory variables	Plasma adrenaline
	Rest		Mental stress		Response (Δ)
	β	p	β	p	β	p
High screening blood pressure	0.37	<0.05	0.40	<0.01	0.38	<0.05
Body mass index	−0.56	<0.001		NS		NS
Peak oxygen uptake		NS	0.54	<0.001	0.46	<0.01

β, standardized regression coefficient.

All models adjusted for waist circumference and smoking status; the response also adjusted for resting value (all non‐significant).

Figure 1. Correlations between peak oxygen consumption and mean plasma adrenaline at rest (top left) and during mental stress test (MST) (top right), maximal level during MST (bottom left), and the response (bottom right). Filled circles and solid regression line: men with high (H) screening blood pressure; open circles and dashed regression line: men with normal (N) screening blood pressure. r_s, Spearman's rank correlation coefficient. Correlations in H still significant after exclusion of the subject with the lowest peak oxygen consumption for mean (r = 0.59, p<0.05) and maximal (r = 0.66, p<0.01) adrenaline level during MST and for the response (r_s = 0.57, p<0.05).

Figure 2. Correlations between body mass index and mean plasma adrenaline at rest (top) and during mental stress test (MST) (bottom). Filled circles and solid regression lines: men with high (H) screening blood pressure; open circles and dashed regression lines: men with normal (N) screening blood pressure.

Serum lipids and glucose metabolism in relation to fitness and adrenaline

Serum triglyceride (r_s = −0.37, p<0.05) and HDL (r = 0.44, p<0.01) levels correlated with VO_2peak. In parallel, A levels correlated negatively with triglycerides and positively with HDL (Table V, Figure 3). In multiple regression models adjusted for A levels, BMI was the only determinant of total cholesterol (β = 0.50, p<0.01) and triglyceride (β = 0.66, p<0.0001) levels. By contrast, HDL was predicted by waist circumference (β = −0.31, p<0.05) and A_rest (β = 0.36, p<0.05), or, when added, by A_mst alone (β = 0.55, p<0.001). Serum insulin correlated with VO_2peak (r_s = −0.43, p<0.01). It also correlated negatively with A_mst (r_s = −0.49, p<0.05) in the high screening BP group, but was independently related only to waist circumference (β = 0.78, p<0.0001) in this group, as in the overall regression analysis (β = 0.65, p<0.0001). Similarly, although GDR/I correlated with positively with A_mst (r = 0.52, p<0.05) and ΔA (r_s = 0.51, p<0.05), in men with high screening BP, it was predicted only by VO_2peak in this group (β = 0.75, p<0.001), as in the overall analysis (β = 0.70, p<0.0001).

Table V. Correlations between plasma adrenaline and serum lipids and variables of glucose metabolism in the combined groups.

Variable	Plasma adrenaline
	Abas		ln Arest		ln Amst		ln Amax		ΔA		ln Arec
	rs	p	r/rs	p	r/rs	p	r/rs	p	rs	p	r/rs	p
Total cholesterol	−0.46	<0.01	−0.38	<0.05	−0.26	NS	−0.28	NS	−0.05	NS	−0.32	<0.05
HDL	0.39	<0.05	0.48	<0.01	0.55	<0.001	0.52	<0.001	0.32	<0.05	0.39	<0.05
Triglycerides^a	−0.38	<0.05	−0.30	NS	−0.34	<0.05	−0.39	<0.05	−0.21	NS	−0.09	NS
Glucose	−0.31	NS	−0.28	NS	−0.17	NS	−0.16	NS	0.03	NS	−0.18	NS
Insulin^a	−0.18	NS	−0.26	NS	−0.23	NS	−0.27	NS	−0.13	NS	−0.16	NS
GDR/I	0.11	NS	0.21	NS	0.25	NS	0.30	NS	0.19	NS	0.01	NS

A, adrenaline; A_bas, baseline (before infusion), A_rest, mean resting value, A_mst mean value during mental stress test, A_max, maximal value (individual); ΔA, change from mean resting value; A_rec, recovery; HDL, serum high‐density lipoprotein cholesterol; GDR/I, insulin‐adjusted glucose disposal rate.

^aSpearman's rank correlation coefficients (r_s).

Figure 3. Correlations between mean plasma adrenaline level at rest (left) and during mental stress test (MST) (right) and serum high‐density lipoprotein cholesterol (HDL). Filled circles and solid regression line: men with high (H) screening blood pressure; open circles and dashed regression line: men with normal (N) screening blood pressure.

Sympathetic neural and cardiovascular responses in relation to fitness and adrenaline

NA_mst and ΔNA were not significantly related to VO_2peak. NA_mst correlated with A_mst (r = 0.54, p<0.001) and ΔNA with ΔA (r_s = 0.71, p<0.0001), and these relationships remained significant (both β = 0.54, p<0.001) in multiple regression analysis. ΔHR correlated positively with VO_2peak in the high screening BP group (r = 0.72, p<0.001). It correlated with ΔA in men with high (r_s = 0.77, p<0.001) and normal screening BP (r_s = 0.54, p<0.05). As the dependent variable in multiple regression, ΔHR was unrelated to VO_2peak, whereas A_mst (β = 0.70, p<0.0001), or, when added, ΔA (β = 0.76, p<0.0001) were the only independent predictors. ΔSBP correlated positively with VO_2peak in men with high (r = 0.47, p<0.05) and normal (r = 0.61, p<0.01) screening BP. ΔDBP also correlated with VO_2peak in the normal screening BP group (r = 0.59, p<0.01), but not significantly in the high screening BP group (r = 0.42, p = 0.08). As dependent variables in regression models, both ΔSBP (β = 0.54, p<0.001) and ΔDBP (β = 0.57, p<0.001) were independently related only to A_mst.

Adrenaline levels during insulin and saline infusions versus baseline

In the 36 subjects with complete sets of blood samples during glucose clamp, plasma A fell from baseline to 30 min [232 (309) vs 158 (77) pmol/l, p<0.05] but then remained stable [181 (128) and 158 (87) pmol/l at 60 and 90 min, respectively]. The same occurred in the time control group, with plasma A levels 287 (90), 184 (90), 192 (92), and 186 (106) pmol/l at baseline, and 30, 60, and 90 min of saline infusion, respectively. Accordingly, there was no treatment×time interaction (p = 0.98).

Discussion

In these young men, high screening BP was associated with increased plasma A levels after adjustment for BMI and physical fitness. At the same time, resting A was negatively related to BMI, whereas the A level and response during mental stress were positively related to physical fitness, independently of BP status. Further, we found independent positive relationships between plasma A levels and serum HDL, but not with insulin sensitivity. The cardiovascular responses to mental stress were positively related to physical fitness, apparently as a result of higher A levels in more fit individuals.

To our knowledge, and as previously reviewed [31], there has been little evidence that A responses to mental stress are related to physical fitness. Some studies [31], [38], [39] have shown trends suggesting higher A during stress with higher fitness, but we are aware of only one report [40] of a significant positive correlation. However, human and animal studies have suggested that the capacity to secrete A may increase as an adaptive response to physical training [20]. For example, plasma A increases more in physically trained than in sedentary individuals at identical relative exercise intensities and in response to various non‐exercise stimuli [20]. Also, athletes have been shown to have higher night‐time as well as 24‐h integrated plasma A levels than untrained subjects [41]. Furthermore, the A content of the adrenal gland and the volume of the adrenal medulla are increased in endurance‐trained rats compared to sedentary control rats [42]. Although our study did not involve physical training, the present observations suggest a similar effect on the response to psychological stress.

Circulating A mobilises substrates for energy metabolism by stimulating lipolysis, inhibiting insulin secretion, and increasing metabolic rate and thermogenesis [43]. Several studies [23], [24], [26–28] have suggested higher sympathetic neural activity, but lower adrenal medullary activity, with increasing measures of obesity. Also, subjects with more components of the metabolic syndrome were found to have progressively higher urinary NA excretion, but lower A excretion [28]. The present observations support our previous findings [3], [10], [25] of a negative association between BMI and A secretion at rest, but suggest that greater stress responses in lean subjects [10] may at least partly be due to higher fitness. The correlations between A and lipid levels agree with findings from larger‐scale studies [27], [28] measuring 24‐h urinary A excretion. In particular, HDL was positively related to A excretion after adjustment for covariates, including BMI, serum insulin and physical activity [27]. β‐adrenergic receptor blocking drugs decrease serum HDL and increase serum triglyceride levels, and these effects are lower for drugs with intrinsic sympathomimetic activity [44]. Accordingly, low A secretion has been discussed as a possible contributor to the high triglyceride and low HDL levels associated with obesity and insulin resistance [14], [27], [28]. However, we cannot exclude the possibility that the relationship reflects the known association between fitness and HDL [17] through other mechanisms. There are previous reports [25], [27], [28] in support of the negative correlation between plasma A and serum insulin in the men with high screening BP. On the other hand, the positive correlation between GDR/I and A in this group is not consistent with our previous findings [8], [45]. However, it was not significant when adjusted for VO_2peak, suggesting poor fitness, and not adrenal medullary activity, as a main determinant of insulin resistance.

Studies of the relationship between fitness and plasma NA and cardiovascular responses to mental stress have shown inconsistent results [30], [31]. There are, however, some reports [46–48] showing greater cardiovascular reactivity with higher fitness. In view of the present findings, it is tempting to suggest adrenal medullary activation as a link between fitness and cardiovascular reactivity. However, this interpretation relies on the assumption that endogenous A contributes significantly to such responses, which is not obvious [1]. HR and BP reactivity are both dependent on cardiac β‐adrenergic responsiveness [49], which was not assessed in the present study. Although A infusion increases plasma NA and HR [50], [51], the NA and HR responses to mental stress are not affected [51], so the relationships between A levels and these variables do not necessarily imply direct causality. The BP response to mental stress is largely caused by an increase in cardiac output [52]. Infusion of A increases stroke volume, ejection fraction and cardiac output [50]. It induces a dose‐dependent [50] and sustained [53] increase in SBP, and increases the SBP response to mental stress [51]. Thus, indirect evidence suggests a role for circulating A. However, positive correlations between VO_2peak and BP responses might also be explained by increased adrenergic responsiveness with higher fitness. The inotropic response to β‐adrenergic stimulation is enhanced in endurance athletes [54] and increased by exercise training [55]. Accordingly, higher fitness has been associated with greater SBP responses to A [56].

Some limitations of the present study should be mentioned. First, the cross‐sectional design does not allow for definite mechanistic conclusions. The data should also be viewed with some caution due to the relatively small sample size, and the large number of statistical tests increasing the risk of chance findings. We did not recruit men with screening BP between 115/75 and 140/90 mmHg, who might be more representative of young men in general. However, correlations which were rather similar in our two groups (e.g. between A and HDL and BMI) might not differ radically in men with intermediate BP levels. Limitations to the use of venous plasma catecholamine concentrations include disproportionate reflection of local sympathetic nerve activity by NA and lower A concentrations than in arterial blood (i.e. those to which tissues are exposed) [1]. Finally, as discussed previously [11], blood sampling during hyperinsulinaemia is a potential limitation of the present work. However, in our studies [9], [57], the glucose clamp has not affected resting plasma NA or HR more than saline infusion, nor has it affected cardiovascular or catecholamine responses to MST [9]. In agreement with our previous observations [9], [57], plasma A was not affected by insulin compared to saline infusion. Rather, the reduction during both infusions suggests a similar adaptation to supine rest. Thus, the correlations with metabolic variables were similar for plasma A at baseline and the average resting level.

As discussed previously [7], high screening BP during military enlistment may partly reflect increased BP reactivity to stress. The present findings add to our previous studies [7–11], [45] by suggesting that cardiovascular reactivity in young men can be affected by physical fitness through differences in adrenal medullary activity. It may seem like a paradox that increased A secretion has potentially harmful [2] and pro‐hypertensive [4] cardiovascular effects, but is associated with metabolic characteristics known to protect against hypertension and/or its consequences. However, not all hypertensive patients have metabolic abnormalities [14]. The assumed pro‐hypertensive effect of high A secretion may be difficult to prove if it is counterbalanced by a beneficial metabolic profile and efficient vagal cardiac control.

In conclusion, our findings suggest a positive relationship between physical fitness and adrenal medullary activity during mental stress in young men. This may contribute to the proposed relationship between low A secretion and low HDL cholesterol in obesity [14], which is also supported by our findings. However, plasma A was independently associated with high screening BP and closely related to plasma NA and cardiovascular stress responses. Increased adrenaline secretion may be related to high BP, but may at the same time be associated with high physical fitness and a beneficial metabolic profile.

Acknowledgements

We thank the Norwegian Council on Cardiovascular Diseases for financial support, and the Joint Norwegian Medical Services, Head Quarters of the Defence Command, Norway, for their generous cooperation.