A moderate response to stress is adapting for human survival. The activation of the hypothalamic-pituitary-adrenocortical (HPA) axis, inducing the secretion of glucocorticoids, plays a central role in the stress response, as well as interacting with other systems involved in this response: endocannabinoids; the immune response (cytokines have been implicated in response to long-term stress); the release of catecholamines by means of adrenergic neurotransmission [1].

The activation of these pathways leads to a fast physical and psychological response and, in adaptive conditions, produces long-term positive modifications in the brain [2]. These long-term adaptive mechanisms allow humans to effectively deal again with similar requests. Thus, a moderate level of stress exerts a favorable effect on coping mechanisms. By contrast, exposure to severe, traumatic, or chronic stress might have the opposite effect and may result in memory loss, cognitive impairment and stress-related psychopathologies such as anxiety or depressive disorders and post-traumatic stress disorder (PTSD) [2, 3]. Great effort has been devoted to understanding the negative effects of chronic stress. Nevertheless, little has been done to identify how stress promotes coping mechanisms in the brain. These features will provide critical information to improve adjustment strategies in both normal and pathological conditions.

The level of blood cortisol is a milestone in monitoring the stress response. However, the relationship among stress, well-being, and cortisol levels is very complex [4].

The importance of understanding the mechanisms underlying well-being in the elderly has currently been emphasized; nevertheless, little is known about the relationship between well-being and blood cortisol in older people.

Severe stress was shown to be associated with a significant increase in the cortisol/dehydroepiandrosterone (DHEA) ratio in elderly people [5]. During exposure to severe stress, exercise may protect the elderly from an increased cortisol/DHEA ratio [5]. However, in elderly people, low cortisol levels during the day were related with worst performance on the Berg Balance Scale and lower handgrip strength, showing a correlation between the levels of cortisol and physical functioning during the aging [6]. Presumably, according to Seyle’s original theories about eustress and distress [7], if a strong stress can induce excessively high levels of cortisol, moderate stress can be adaptive, stimulating a moderate but not excessive cortisol level.

Quality of Life (QoL) is a subjective concept of well-being depending on several factors, such as concomitance of illness, income and stress. The subjective perception of QoL is now considered of great relevance in measuring the level of appreciation of daily life in elderly patients with or without health problems, particularly in those with chronic conditions [8, 9]. It has become central in evaluating the effectiveness of treatment as well [10].

The perception of QoL improves in women with depression following a fitness program [11, 12]. In a recent study, we showed that a high QoL was associated with exercise in the elderly, and a vigorous physical activity program might be associated with better maintenance of results over time as compared to a postural gymnastic program [13].

The present paper sought to determine whether, in the same sample used in the above-cited study [13], cortisol levels taken before, during and after the trial were related to the perception of QoL.

The hypothesis is that the positive changes in the perception of QoL induced by exercise in a sample of older people could be related to exercise-induced changes in blood cortisol.

The study was carried out as part of the “A Chent’Annos in Salude” (a typical Sardinian saying, which could be translated as “I hope you’ll live a hundred years in good health”) trial, a prospective population-based study aiming to verify the physiological modifications induced by exercise in the elderly, carried out by the Italian Olympic Committee (CONI) Sardinia, approved by CONI Sardinia and conducted at the Italian Olympic Committee in Cagliari, Italy. The methods were described in a previous paper [13]; briefly, the study was a 12-week randomized controlled trial, in which participants were volunteers aged ≥65 years, recruited by announcements in a local newspapers, general practitioners, flyers in pharmacies. About 350 people manifested their interest to participate, and they were contacted by phone by the research team for further information about the study; more or less 180 people agreed to participate and underwent a detailed medical evaluation trough the list of inclusion/exclusion criteria.

Inclusion required at least one criterion between sedentariness (defined according the Physical Activity Guidelines for Older Adults as less than 30 minutes/day of moderate physical activity performed on five or fewer days/week [14]) and dysmetabolism (defined according to the American Hearth Association criteria [15]). Exclusion criteria were the presence of: any cardiovascular, osteoarticular, neurological or oncological disease that could be a contraindication to perform a stress test or that could be incompatible with exercise; moderate to severe anemia (Hb<10 mg/dl); obesity with BMI >35. Moreover, only subjects without difficult to reach the gym where the study was carried out were included.

After signing a written informed consent, all participants underwent a thorough medical examination, that included: anthropometric measurements; electrocardiogram, systolic/diastolic blood pressure measurement, spirometry at rest and under stress test; blood sampling for complete blood count, glycaemia, total cholesterol, high-density and low-density lipoproteins, triglycerides, azotemia, creatinine, cortisol. Following this check-up, participants were randomly assigned in a 1:1 fashion to two different groups: VAG (Vigorous Activity Group) and PGG (Postural Gymnastic Group).

The study started in April 2013 (T0) and ended in July 2013 (T1). Twelve weeks later, participants were submitted to follow-up and assessed again (T2).

21 subjects were assigned to treatment with the VAG (12 males, 9 females), mean age 69.24±3.9; 21 subjects were assigned to treatment with the PGG (14 males and 7 females, mean age 70.0±5.3).

In the VAG, 3 subjects reported at least one previous depressive episode, one of them being treated with antidepressant drugs. Two other subjects were taking benzodiazepines. In the PGG, one person reported at least one previous depressive episode. Another subject was taking antidepressants and benzodiazepines. One participant in the PGG was undergoing chronic therapy with corticosteroids; nevertheless, as this participant’s cortisol level was in the normal range, we did not remove it.

In the VAG, 11 subjects were receiving pharmacological treatment, in 7 cases multiple-drug treatment. In the PGG, 15 subjects were taking medication, in 11 cases multiple drugs. In the VAG, concomitant organic diseases were: hypertension (8 subjects), hypercholesterolemia (3 subjects) and osteoporosis (2 subjects); in the PGG, 8 subjects were suffering from hypertension, 5 from diabetes mellitus (type 1 in one case), 4 from hypercholesterolemia, 4 from non-severe heart disease (heart failure, atrial fibrillation, aorta-coronary bypass), 3 from thyroid diseases (post-surgical hypothyroidism, Hashimoto's thyroiditis) and 2 from musculoskeletal diseases (arthritis, osteoporosis).

Two subjects in the VAG group and one in the PGG group dropped out of the study.

Table 1 shows blood cortisol levels during the trial. In both groups, cortisol levels rose at the end of the trial as compared to the baseline, with statistically significant differences. Cortisol in both groups decreased slightly 12 weeks after the end of the trial (T2); however, the difference from the initial level in the VAG sample remained statistically significant, while the cortisol level in the PGG group did not differ from the initial level.

Table 2 shows SF-12 scores at T1 and T2 for the VAG and PGG groups as compared to the normative Italian sample. At the end of the study (T1), the level of QoL measured by means of SF-12 scores was better in both the VAG and PGG groups than in the normative standardized sample. SF-12 scores for the two groups decreased slightly 12 weeks after the end of the trial; however, the difference between the VAG and the normative sample remained statistically significant (ANOVA 1 way: F=5.52; p=0.019; df 1,703,704), while the QoL level in the PGG group did not differ from the normative sample (ANOVA 1 way: F=3.10; p=0.079; df 1,703,704).

Table 3a and 3b respectively show the distribution of subjects at T1 and T2 with an SF-12 score higher than average, plus one standard deviation from the distribution on the national normative sample standardized by sex and age [16], between those with cortisol levels ranging from 200 to 300 mg/ml and those with cortisol levels<2 00 or >300 mg/ml. At T1, subjects in both groups (N=20) with cortisol level ≥200 mg/ml showed no difference in SF-12 scores as compared with subjects with less than 200 mg/ml (N=19) (40.5±4.6 vs. 39.0±5.4, df 1,37,38, F=0.875, p=0.356). At T2, subjects with cortisol level ≥200 mg/ml in both groups (N=20) showed higher levels on SF-12 scores than subjects with less than 200 mg/ml (N=19) (41.2±4.8 vs. 38.1±4.6, df 1,37,38, F=4.424, p=0.047). Subjects with more than 200 mg/ml of cortisol numbered 7 (36.8%) at T0; 17 (43.6%) at T1 (T0 vs. T1 chi square with Yates’ correction 4.875, 1df, p=0.0027); 17(43.6%) at T2 (T0 vs. T2 chi square with Yates’ correction 4.875, 1df, p=0.0027).

No differences were found in the distribution of subjects with more than 200 mg/ml of blood cortisol levels according to group at the three observation times [T0: PGG=4 (20%), VAG=3 (15.8%): Yates' χ²=0.006, p=0.938; T1: PGG=11 (55%), VAG=6 (31.6%): Yates' χ²=0.132, p=0.249; T2: PGG=9 (45%), VAG=8 (42.1%): Yates' χ²=0.002, p=0.887] (Table 4a).

Subjects with cortisol levels between 200 and 300 mg/ml numbered 7 (36.8%) at T0, 13 (26.5%) at T1 (T0 vs. T1 chi square with Yates’ correction 2.421, 1df, p=0.119) and 16 (41.0%) at T2 (T0 vs. T2 chi square with Yates’ correction 4.994, 1df, p=0.025). The distribution of subjects with cortisol levels between 200 and 300 mg/ml according to group showed no differences at the three observation times [T0: PGG=4 (20%), VAG=3 (15.8%): Yates' χ²=0.006, p=0.938; T1: PGG=7 (35%), VAG=6 (31.6%): Yates' χ²=0.013, p=0.909; T2: PGG=9 (45%), VAG=7 (36.8%): Yates' χ²=0.604, p= 0.847] (Table 4b).

The study shows that both at the end and 12 weeks after a trial proposing different exercise levels, a sample of older people increased their blood levels of cortisol, reaching, in most subjects, a higher but not pathological level. These changes appear to be related to changes in the perception of the Quality of Life.

In the group of elderly subjects undergoing vigorous exercise, at the end of the trial, blood cortisol levels were higher than those prior to beginning the trial, along with a better perception of the Quality of Life than the score for a normative national sample. Twelve weeks after the end of the trial, both cortisol levels and scores regarding the Quality of Life decreased, but remained higher, with statistically-significant differences, when compared, respectively, both to basal cortisol levels and to the level of the Quality of Life for the normative sample.

In the group of elderly subjects undergoing moderate exercise, blood cortisol levels were higher at the end of the trial than prior to its start, just as the Quality of Life score was higher than the score for the normative national sample, with statistically significant differences in both cases. Twelve weeks after the end of the trial, both cortisol levels and Quality of Life scores decreased, reaching levels and scores with no statistically-significant differences compared, respectively, both to doses of basal cortisol as well as to the level of the Quality of Life for the normative sample.

Subjects with cortisol levels ≥200 mg/ml were higher at T1 and T2 than at T0.

Subjects with cortisol levels ≥200 mg/ml in both groups showed no difference in SF-12 scores as compared to subjects with less than 200 mg/ml at T1, but they reported Quality of Life scores higher than those with less than 200 mg/ml at T2.

This seemingly contradictory result must be assessed in light of the fact that some individuals, including those with cortisol levels of more than 200 mg/ml, showed a blood cortisol level above 300, which can be judged to be in the pathological range.

At both T1 and T2, a better perception of the Quality of Life (SF-12 with a score of one standard deviation higher than the normalized national average) was in fact more common among those who had a cortisol levels between 200 and 300 mg/ml.

Subjects with cortisol levels between 200 and 300 mg/ml increased from T0 to T1, but only at T2 was the difference higher at a significant statistical level as compared to T0.

It could therefore be hypothesized that exercise, in some specific cases, could excessively increase the cortisol level and, for this reason, results might be better in these cases 12 weeks after the end of the survey. However, it could alternatively be speculated that it was not exercise that directly caused the increase in cortisol between T0 and T2, if the increase at T0, T1 and T2 was linear. However, other papers have confirmed that exercise is one factor that can determine an increase in cortisol, but that it interacts with other determinants of cortisol levels, such as stressful events [5]. Older adults with higher levels of stress demonstrated a significantly higher cortisol/DHEA ratio. Individuals with higher stress incidence who did not train aerobic exercise had a significantly higher cortisol/DHEA ratio and a flatter DHEA diurnal rhythm as compared with those who regularly did so [5].

One could assume at least two distinct cortisol-action mechanisms in the perception of well-being in the elderly: first, the role of cortisol in the formation of steroid hormones deriving from progesterone, and secondly, the direct effect of cortisol on the attenuation of stress, also related to the endocannabinoid system.

It has been hypothesized that neurosteroids deriving from progesterone may play a role as endo-mood stabilizers [8], and high levels of 3alpha-hydroxy-5alpha-pregnan-20-one (3alpha,5alpha-THPROG) and 3alpha,21-dihydroxy-5alpha-pregnan-20-one have been found in women with bipolar disorder, clinically compensated for and in symptomatic remission [20]. Not much is known about how cortisol might influence the level of neuroactive steroid hormones deriving from progesterone, or rather their analogues sulfates, or testosterone, which apparently have a different effect on mood [21]. Indeed, a recent study found that in the elderly, the perception of Quality of Life was not affected by testosterone levels [22]. Among adolescents, cortisol has been found to be positively associated with DHEA slopes, and negatively associated with testosterone [23]. As the level of testosterone decreases with age, it is therefore possible that an increase in cortisol in the elderly can lead to a more marked increase in progesterone derivatives.

A direct role of cortisol in modulating stress was also shown; in particular, cortisol and endocannabinoid systems seem to interact in stimulating "positive memories" after stressful events [1]. Several studies showed that administering cortisol to seriously ill or injured subjects produces a significant reduction of PTSD symptoms after recovery, without affecting the number of trauma memories [1].

A recent paper reviewed the role of glucocorticoids and glucocorticoid receptors (GRs) in memory formation and modulation [2], highlighting the evidence on the molecular events at the basis of GRs activation effect during moderate stress, promoting long-duration memory. Such effect could mainly involve the brain-derived neurotrophic factor (BDNF), because, after stress-induced learning, the activation of hippocampal GRs might determine the activation of the BDNF/cAMP response element-binding protein (CREB) pathway, the general mechanism promoting the consolidation of long-term memory [2]. According to this hypothesis, exercise could moderately increase cortisol levels, which might improve coping with daily stress in the elderly, and might protect against the effects of chronic stress by counteracting a stress-related decrease in hippocampal BDNF mRNA.

The sample was small; the perception of Quality of Life was not measured at the beginning of the trial, as the “A Chent’annos in Salude” trial was not specifically designed to address the issue of how exercise might induce both an increase in blood cortisol levels and a better subjective Quality of Life. Thus, the association among exercise, the perception of Quality of Life and the moderate increase in cortisol levels in the elderly suggested by this research must be taken into account as only a heuristic hypothesis. Nevertheless, if confirmed, these results could represent a first step in better understanding the relationship between physical activity and well-being in elderly people.

The results of this study appear to demonstrate that exercise moderately increases blood cortisol levels in the elderly, and that this increase is related to the perception of a high Quality of Life. More sustained exercise extends both the increase in cortisol levels ​​and improvement in the perception of the Quality of Life. Future research using larger samples and stronger methodologies is needed to confirm the relationship among cortisol, exercise and perception of the Quality of Life suggested by this study.

The authors confirm that this article content has no conflict of interest.