- Email: [email protected]
The influence of high-normal testosterone levels on risk-taking in healthy males in a 1-week letrozole administration study
Psychoneuroendocrinology (2010) 35, 1416—1421
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
SHORT COMMUNICATION
The influence of high-normal testosterone levels on risk-taking in healthy males in a 1-week letrozole administration study Anna E. Goudriaan a,*, Bruno Lapauw b, Johannes Ruige b, Els Feyen b, Jean-Marc Kaufman b, Matthias Brand c, Guy Vingerhoets d a
Department of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam Institute for Addiction Research, P.O. Box 22660, 1100 DD Amsterdam, The Netherlands b Department of Endocrinology and Andrology, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium c Department of Cognitive Psychology, University of Duisburg-Essen, Germany d Laboratory for Neuropsychology, Department of Neurology, Ghent University, De Pintelaan 185, B-9000 Ghent, Belgium Received 24 November 2009; received in revised form 13 April 2010; accepted 13 April 2010
KEYWORDS Testosterone; Estradiol; Aromatase inhibitor; Risk-taking; Neurocognition
Summary Human studies on the relation between testosterone levels and risk-taking behaviour are scarce. Related functions, like aggression, have been related to higher testosterone levels more consistently, especially in the animal literature. Estradiol affects several neurotransmitter systems that play a role in behaviour regulation. Existing human studies on neurocognitive functions and testosterone levels have largely ignored the interrelatedness of testosterone levels and estradiol levels. Therefore, in this study, the effects of a 1-week combined testosterone and estradiol intervention on risk-taking behaviours were investigated. Twenty-one healthy men, with a normal body mass index, were treated for 7 days with an aromatase inhibitor (letrozole 2.5 mg), resulting in high-normal levels of testosterone and lownormal levels of estradiol, or with a combination of an aromatase inhibitor and estradiol (75 mg/ 24 h), resulting in low-normal levels of testosterone and high-normal levels of estradiol. A randomized experimenter and participant-blind controlled design was applied. Neurocognitive measures of risk-taking and reward and punishment sensitivity were assessed before starting with the medication and after 7 days of drug administration: Balloon Analogue Risk Task (BART), Game of Dice Task (GDT), and Iowa Gambling Task (IGT). A group by time effect was present for the BART, indicating that the high-normal testosterone group showed an increase in risk-taking on the BART, from the first drug-naive BART performance, to the second BART performance (aromatase inhibitor), whereas such an increase was not present in the low-normal testosterone/high estradiol group. No group by time interactions were present in GDT or IGT performance. These
* Corresponding author. Tel.: +31 020 8913761; fax: +31 020 8913701. E-mail addresses: [email protected], [email protected] (A.E. Goudriaan). 0306-4530/$ — see front matter # 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.psyneuen.2010.04.005
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
1417
results implicate that testosterone levels in healthy men are associated with increased risk-taking under conditions of unknown probabilities, but not in conditions of known probabilities (GDT) or of strategic decision making (IGT). # 2010 Elsevier Ltd. All rights reserved.
1. Introduction Animal studies have linked increases in testosterone levels to increased aggression and reduced fear, and more complex testosterone/estrogen interactions have been implicated in the expression of aggression in the animal literature (Haug et al., 1986; Trainor et al., 2006). In humans, elevated aggression levels have been linked to supraphysiological levels (pharmacological enhancement) more consistently than testosterone administration within the normal physiological range (Kouri et al., 1995; Perry et al., 2003). Although higher testosterone levels have been related to impulsive responding and reduced fear in animals, research into impulsivity and risk-taking in humans is limited. A recent study on real life risk-taking behaviour in day traders indicates that higher — naturally present — morning testosterone levels are associated with higher post-morning profits in day traders (Coates and Herbert, 2008). High testosterone levels in pubertal boys have been associated with more self-reported risk-taking behaviour (Vermeersch et al., 2008). In a laboratory study, pharmacologically heightened testosterone levels in women resulted in disadvantageous decision making in the Iowa Gambling Task (IGT, van Honk et al., 2004). In another study, physiological testosterone levels correlated negatively with IGT performance in young men, but not in older men (Reavis and Overman, 2001). Several aspects of reward processing and risk-taking influence IGT performance, and knowledge regarding risks seems to influence testosterone’s role differently (e.g., a positive effect in the day trader’s study–—where probabilities of winning or losing remain unknown, a negative effect in the IGT studies where probabilities become known during performance of the task). In addition, non-linear relationships between testosterone levels and risk-attitudes have been reported (Sapienza et al., 2009). Therefore, this study focuses on the effects of an intervention that moderately increases testosterone levels — within or near normal-physiological boundaries — on risk-taking under known and unknown probabilities. In males, testosterone is converted through aromatase (P450a) and 5a reductase in estradiol and dihydrotestosterone (DHT). When testosterone is administered, production of estradiol is changed by alterations in substrate concentration, and possibly through changed aromatase activity. Testosterone and estradiol are neuroactive steroids that exert their influence on brain regions involved in emotion and motivation (Zheng, 2009). Estradiol influences cognition in women (Spencer et al., 2008), but very little is known on the role of estradiol in males’ cognitive functioning. Therefore, in this study, both testosterone and estradiol levels were manipulated, in order to control the influence of high-normal testosterone/low estradiol and low-normal testosterone/high estradiol on risk-taking. An aromatase inhibitor in combination with estradiol results in an absolute decline in testosterone levels by 44%. In men, an increase in peripheral estradiol reflects the inhibitory tone exerted by estrogens on gonadotropin
release and is a major determinant of peripheral testosterone. A 7-day medication trial was implemented, consisting of administration of (1) an aromatase inhibitor, resulting in high-normal testosterone and low-normal estradiol, or (2) an aromatase inhibitor in combination with estradiol, resulting in low-normal testosterone, and high-normal estradiol (Raven et al., 2006). Given the literature on the enhancing effects of testosterone on risk-taking and its role in fear reduction, we hypothesize that testosterone leads to enhanced risk-taking in neurocognitive gambling tasks. In particular, stronger effects are expected on tasks with unknown or partially known/implicit risks (Balloon Analogue Risk Task, IGT) compared to tasks with known risks (Game of Dice Task).
2. Methods 2.1. Participants and test procedure Twenty-one healthy male subjects were recruited. Participants had a body mass index (BMI) < 27, normal medical history, and normal biochemical measures of haematological, hepatic, renal, gonadal and metabolic function. Exclusion criteria included smoking, performing extreme sporting activities, excessive alcohol/substance use, major psychiatric disorders, use of psychotropic drugs or drugs with a potential influence on sex hormones (e.g., glucocorticoids), recent weight loss or gain (>10%), shift work, and prostate-specific antigen levels exceeding 4 ng/mL. This study was approved by the Ethical Review Board of Ghent University Hospital, conducted in accordance with the Declaration of Helsinki, and written informed consent was obtained. Participants were randomized to receive either an aromatase inhibitor (letrozole 2.5 mg/day, orally taken on awakening, testosterone group), or letrozole plus an E2 patch (75 mg/day; estradiol group), for 7 days. Hormonal blood samples were performed before and after treatment in similar conditions at 08:00 h a.m. after an overnight fast and 10-min bed rest at the hospital. Repeated hormonal measurements, e.g., 2 days before and 2 days after a 1-week treatment were not regarded feasible as all subjects had a fulltime job. Commercial immunoassays were used to determine serum E2 (Incstar, Stillwater, MN, USA; adapted protocol with use of double amount of serum), T and SHBG (Orion Diagnostica, Espoo, Finland). Tests were administered by a trained research assistant who was blind to medication group. One person was excluded because of lack of fasting at blood sampling and one person was excluded due to low medication adherence.
2.2. Risk-taking tasks 2.2.1. Balloon Analogue Risk Task (BART) The BART assesses risk-taking behaviour under unknown probabilities (Lejuez et al., 2002; Hunt et al., 2005). In this
1418
A.E. Goudriaan et al.
computerized task, a balloon appears on the computer screen, together with a button to inflate the balloon, and a button to accumulate money. With each pump, money is accrued (5 cents per pump: this is not displayed), but when the balloon explodes all gains for that balloon are lost. Participants are instructed to earn as much money as possible by inflating the balloon, and accumulating money. Total amount of money earned in the task, and amount earned with the last balloon are displayed. Participants are instructed that at some point each balloon will explode, as early as the first pump, or as late as the point at which the balloon would fill the entire computer screen (average breakpoint 64 pumps). However, no information is given on the probability of balloon explosions. The task ends after 30 balloons are completed. The dependent measures of the BART are the adjusted total number of pumps (excluding exploded balloon pumps), total money earned, and total number of balloon explosions. 2.2.2. Game of Dice Task (GDT) The GDT assesses risk-taking under known probabilities (Brand et al., 2008). In this computerized task, participants are instructed to maximize a fictitious starting capitol of 1000s, within 18 throws of a virtual die. Before each throw, subjects are required to guess what number(s) presented on the screen will include the number that will be thrown next. With increasing numbers chosen (i.e., higher winning probability), potential wins decrease, but potential losses also decrease, e.g., choosing one number would result in winning 1000s if the number chosen is thrown, and in losing 1000s if another number is thrown, whereas choosing 3 out of 6 numbers as winning numbers would result in winning or losing 200s. After each trial, gains or losses, total money earned and numbers of trials left are indicated. The dependent measure is the number of risky choices, i.e., the frequency of choosing one or two numbers, where winning probability is less than 34%, resulting in high frequent losses in the long run (Brand et al., 2008). 2.2.3. Iowa Gambling Task (IGT) The IGT measures decision making under ambiguous conditions. Probabilities of winning and losing have to be discovered during performance of the task by using feedback
Table 1
from previous trials. A computerized IGT was used (Bechara et al., 2000). Participants choose between four decks of cards. Unknown to the subject, two decks give high rewards, but also result in high losses, and are disadvantageous in the long run. The other decks give lower rewards, but also lower losses, and result in a net gain in the long run. Participants have to learn to avoid choosing the disadvantageous decks and to choose cards from the advantageous decks. The task ends after 100 trials. Total number of cards selected from the advantageous decks is taken as the dependent measure.
2.3. Statistical analyses T-tests were performed to investigate differences in demographic and clinical variables. Separate MANOVAs were performed for each task. Group (testosterone vs. estradiol) was entered as between subjects factor, and session (baseline vs. session 2, after 1-week hormone treatment) was included as within subject factor. When multiple dependent measures were included, post hoc comparisons were made for each dependent measure separately. In addition, correlations of the dependent measures with testosterone or estradiol were examined.
3. Results 3.1. Demographics and hormone levels The two groups did not differ significantly with respect to age (t = 1.11, p = 0.28) or BMI (t = 0.94, p = 0.36). The group by time interactions were significant for hormone levels in both groups, showing an increase in testosterone levels from baseline to session 2 in the testosterone group, and a decrease in testosterone levels in the estradiol group, F(1,18) = 132.3, p < 0.0001; all testosterone levels were (slightly above) high-normal in the testosterone group, and (slightly below) low-normal in the estradiol group. Reversely, estradiol levels increased in the estradiol group but decreased in the testosterone group, F(1,18) = 26.15, p < 0.0001, resulting in (slightly below) low-normal levels in the testosterone group, and high-normal levels in the estradiol group (see Table 1).
Demographics and hormone levels. Estradiol group (n = 9) Baseline session
Age (years) Body mass index Total testosterone (ng/dL) Free testosterone (ng/dL) Total estradiol (pg/mL) Free estradiol (pg/mL)
Mean
SD
30.6 23.4 425.36 9.51 17.09 0.33
4.7 2.24 136.87 2.25 2.64 0.05
Testosterone group (n = 10) Test session 2 (aromatase 2.5 mg + estradiol 50 mg) Mean SD
Baseline session
Test session 2 (aromatase 2.5 mg)
Mean
SD
Mean
SD
245.72 5.16 25.26 0.46
33.5 24.4 495 8.8 20.5 0.37
6.7 2.40 137 2.0 3.54 0.07
988 21.5 8.9 0.18
137 4.9 0.93 0.02
126.83 2.64 13.61 0.24
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
3.2. Risk-taking tasks Dependent measures for groups and test sessions are presented in Table 2. 3.2.1.1. BART A significant time effect was present, F(3,15) = 4.26, p = 0.023, indicating a change in performance from baseline to the second session. A trend for a group effect was present, F(3,15) = 3.05, p = 0.061. A significant group time effect, F(3,15) = 4.06, p = 0.027, indicates that the testosterone group exhibited a higher increase in risk-taking behaviour during the second test session compared to the estradiol group. Post hoc comparisons revealed a non-significant higher increase in adjusted number of pumps, from baseline to the second test session, F(1,17) = 1.97, p = 0.18, the amount of money earned increased more in the testosterone group compared to the estradiol group, F(1,17) = 4.82, p = 0.04, and the number of explosions was non-significantly higher from baseline to the second session in the estradiol group compared to the testosterone group, F(1,17) = 1.63, p = 0.22. 3.2.1.2. IGT A trend for a time effect was present, indicating better performance in the second test session, compared to baseline, F(1,17) = 3.91, p = 0.064. Neither the group effect, F(1,17) = 0.19, p = 0.67, nor the group time interaction, F(1,17) = 0.05, p = 0.82, was significant 3.2.1.3. GDT A time effect was present, F(1,17) = 5.62, p = 0.03, indicating more advantageous decisions in the second session compared to baseline. Again, neither the group effect, F(1,17) = 0.06, p = 0.81, nor the group time interaction, F(1,17) = 0.66, p = 0.43, reached significance. 3.2.2. Correlations risk-taking and hormone levels A significant correlation between total testosterone and amount of money earned in the BART (r = 0.50, p < 0.05) was present in the second session, but not in the baseline Table 2
1419
session. IGT performance tended to correlate with levels of free estradiol (r = 0.41, p < 0.10) in the second session, but not in the baseline session. As can be seen in Table 3, most of the BART measures correlated significantly during session 1, but less correlation between the BART measures was present during session 2. In addition, IGT performance correlated significantly with BART number of pumps during session 1, but not during session 2, whereas the Game of Dice Task did not correlate significantly with BART or IGT measures during either of the sessions.
4. Discussion In this study, we tested the hypothesis that a group with highnormal levels of testosterone/low-normal estradiol (after a 7-day administration of an aromatase inhibitor) in contrast to a group with low-normal testosterone/high-normal estradiol levels (after a 7-day administration of an aromatase inhibitor and estradiol), would lead to higher risk-taking behaviour in neuropsychological decision-making tasks, especially in tasks with unknown probabilities. In accordance with this hypothesis, we found higher risk-taking behaviour in the BART (unknown probabilities) in the high-normal testosterone/ low-normal estradiol group, compared to the high-normal estradiol/low-normal testosterone group. These effects were not found for both IGT (in which knowledge about probabilities typically increases over task duration) and GDT (known probabilities from the beginning of the task). The specific link between high-normal testosterone/lownormal estradiol and increased risk-taking in the BART, but not in the IGT or GDT, implies that risk-taking is affected by fluctuations of testosterone within or close to normal levels, particularly when no information about the actual risk is provided. Other studies have shown that elevation of testosterone levels to supraphysiological levels shows more consistent effects compared to studies employing lower levels of testosterone. Thus, the positive effects of testosterone on risk-taking as seen in this study, are consistent with the positive effects of natural testosterone in the day traders study (Coates and Herbert, 2008). Our results are inconsistent, however, with the two previous studies which have
Balloon Analogue Risk Task, Iowa Gambling Task, and Game of Dice Task measures in the first and second test session. Estradiol Group (n = 9)
Testosterone group (n = 10)
Baseline session
Test session 2 (aromatase 2.5 mg + estradiol 50 mg)
Baseline session
Test session 2 (aromatase 2.5 mg)
Mean
SD
Mean
SD
Mean
SD
Mean
SD
32.33 3268.9 8.0
15.00 1174.9 4.58
36.72 3551.5 9.5
12.35 845.9 4.38
28.09 3390.0 8.5
10.55 558.5 4.77
41.32 4262.5 8.5
14.83 156.1 4.77
IGT N advantageous choices
45.89
11.84
55.11
18.12
44.60
11.77
51.90
15.01
GDT N advantageous throws
13.78
4.84
15.78
2.68
12.30
6.92
16.40
2.66
BART Adjusted average pumps Total money earned Number of explosions
1420 Table 3
A.E. Goudriaan et al. Pearson’s correlations between risk-taking dependent measures. Session 1 Adjusted average pumps
Total money earned
Number of explosions
N advantageous choices
BART Adjusted average pumps Total money earned Number of explosions
1.000 0.83 *** 0.63 **
1.000 0.63 **
1.000
IGT N advantageous choices
0.67 **
0.411
0.406
1.000
GDT N advantageous throws
0.07
0.01
0.17
0.11
N advantageous throws
1.000
Session 2 Adjusted average pumps
Total money earned
Number of explosions
N advantageous choices
BART Adjusted average pumps Total money earned Number of explosions
1.000 0.36 0.89 ***
1.000 0.30
1.000
IGT N advantageous choices
0.02
0.14
0.07
1.000
GDT N advantageous throws
0.34
0.08
0.43
0.08
N advantageous throws
1.000
*p < 0.05, **p < 0.01, ***p < 0.001.
indicated a negative effect of testosterone on IGT performance (Reavis and Overman, 2001; van Honk et al., 2004). One explanation for the divergent findings may be that the less extreme elevations of testosterone play a smaller role in IGT performance, especially in older men, as suggested by the study of Reavis and Overman (2001). In addition, the nonlinear relationship between testosterone and risk-taking, may explain the differential findings of the reported studies (Takahashi et al., 2006; Sapienza et al., 2009). However, it may also be that we failed to show such effects because of the relatively small sample size and a high variance in IGT performance, which are clear limitations of our study. In addition, our findings may be related to the simultaneous variation in estradiol. An animal study showed that estradiol receptors are related to the influence of femaleassociated cues on male risk taking (Kavaliers et al., 2008), but human studies are scarce. Given our preliminary findings, the mutual influence of testosterone and estradiol on risk-taking in humans has to be examined in more detail. Finally, this study was limited to adult men, and a recent null-finding in post-menopausal women shows that testosterone administration in this population did not change risk-attitudes (Zethraeus et al., 2009). Future studies therefore have to replicate these findings in larger samples, including both sexes, and both young adults and older adults. The role of stress hormones such as cortisol in decision making should also be investigated in future studies, because stress may also influence risk-taking (Starcke et al., 2008). Different levels of testosterone (supraphysiological, normal to high-normal range) seem to affect
risk-taking and impulsivity differentially. Therefore, administering several levels of testosterone could increase our knowledge on the link between testosterone levels and risk-taking and impulsivity. In conclusion, results from this study imply that an increase of testosterone in the normal to high-normal range can increase risk-taking behaviour and improve task performance on a specific risk-taking task with unknown risks.
Role of funding source This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector. Anna E. Goudriaan is supported by a new researcher grant as provided by the Dutch Scientific Organisation (NWO ZonMw, #91676084, 2007-2011).
Conflict of interest All authors have no conflicts of interest to disclose.
Acknowledgements We are indebted to Kaatje Toye and Bea Vervinckt for the meticulous realization of the study protocol and to Inge Bocquart and Kathelijne Mertens for performing the immunoassays. We thank all volunteers who participated as study subjects.
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.psyneuen. 2010.04.005.
References Bechara, A., Tranel, D., Damasio, H., 2000. Characterization of the decision-making deficit of patients with ventromedial prefrontal cortex lesions. Brain 123, 2189—2202. Brand, M., Heinze, K., Labudda, K., Markowitsch, H.J., 2008. The role of strategies in deciding advantageously in ambiguous and risky situations. Cogn. Process. 9, 159—173. Coates, J.M., Herbert, J., 2008. Endogenous steroids and financial risk taking on a London trading floor. Proc. Natl. Acad. Sci. U.S.A. 105, 6167—6172. Haug, M., Brain, P.F., Kamis, A.B., 1986. A brief review comparing the effects of sex steroids on two forms of aggression in laboratory mice. Neurosci. Biobehav. Rev. 10, 463—468. Hunt, M.K., Hopko, D.R., Bare, R., Lejuez, C.W., Robinson, E.V., 2005. Construct validity of the Balloon Analog Risk Task (BART): associations with psychopathy and impulsivity. Assessment 12, 416—428. Kavaliers, M., Devidze, N., Choleris, E., Fudge, M., Gustafsson, J.A., Korach, K.S., Pfaff, D.W., Ogawa, S., 2008. Estrogen receptors alpha and beta mediate different aspects of the facilitatory effects of female cues on male risk taking. Psychoneuroendocrinology 33, 634—642. Kouri, E.M., Lukas, S.E., Pope Jr., H.G., Oliva, P.S., 1995. Increased aggressive responding in male volunteers following the administration of gradually increasing doses of testosterone cypionate. Drug Alcohol Depend. 40, 73—79. Lejuez, C.W., Read, J.P., Kahler, C.W., Richards, J.B., Ramsey, S.E., Stuart, G.L., Strong, D.R., Brown, R.A., 2002. Evaluation of a behavioral measure of risk taking: the Balloon Analogue Risk Task (BART). J. Exp. Psychol. Appl. 8, 75—84. Perry, P.J., Kutscher, E.C., Lund, B.C., Yates, W.R., Holman, T.L., Demers, L., 2003. Measures of aggression and mood changes in
1421
male weightlifters with and without androgenic anabolic steroid use. J. Forensic Sci. 48, 646—651. Raven, G., de Jong, F.H., Kaufman, J.M., de, R.W., 2006. In men, peripheral estradiol levels directly reflect the action of estrogens at the hypothalamo-pituitary level to inhibit gonadotropin secretion. J. Clin. Endocrinol. Metab. 91, 3324—3328. Reavis, R., Overman, W.H., 2001. Adult sex differences on a decisionmaking task previously shown to depend on the orbital prefrontal cortex. Behav. Neurosci. 115, 196—206. Sapienza, P., Zingales, L., Maestripieri, D., 2009. Gender differences in financial risk aversion and career choices are affected by testosterone. Proc. Natl. Acad. Sci. U.S.A. 106, 15268—15273. Spencer, J.L., Waters, E.M., Romeo, R.D., Wood, G.E., Milner, T.A., McEwen, B.S., 2008. Uncovering the mechanisms of estrogen effects on hippocampal function. Front. Neuroendocrinol. 29, 219—237. Starcke, K., Wolf, O.T., Markowitsch, H.J., Brand, M., 2008. Anticipatory stress influences decision making under explicit risk conditions. Behav. Neurosci. 122, 1352—1360. Takahashi, T., Sakaguchi, K., Oki, M., Homma, S., Hasegawa, T., 2006. Testosterone levels and discounting delayed monetary gains and losses in male humans. Neuro. Endocrinol. Lett. 27, 439—444. Trainor, B.C., Kyomen, H.H., Marler, C.A., 2006. Estrogenic encounters: how interactions between aromatase and the environment modulate aggression. Front. Neuroendocrinol. 27, 170—179. van Honk, J., Schutter, D.J., Hermans, E.J., Putman, P., Tuiten, A., Koppeschaar, H., 2004. Testosterone shifts the balance between sensitivity for punishment and reward in healthy young women. Psychoneuroendocrinology 29, 937—943. Vermeersch, H., T’Sjoen, G., Kaufman, J.M., Vincke, J., 2008. The role of testosterone in aggressive and non-aggressive risk-taking in adolescent boys. Horm. Behav. 53, 463—471. Zethraeus, N., Kocoska-Maras, L., Ellingsen, T., von, S.B., Hirschberg, A.L., Johannesson, M., 2009. A randomized trial of the effect of estrogen and testosterone on economic behavior. Proc. Natl. Acad. Sci. U.S.A. 106, 6535—6538. Zheng, P., 2009. Neuroactive steroid regulation of neurotransmitter release in the CNS: action, mechanism and possible significance. Prog. Neurobiol. 89, 134—152.
a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m
j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / p s y n e u e n
SHORT COMMUNICATION
The influence of high-normal testosterone levels on risk-taking in healthy males in a 1-week letrozole administration study Anna E. Goudriaan a,*, Bruno Lapauw b, Johannes Ruige b, Els Feyen b, Jean-Marc Kaufman b, Matthias Brand c, Guy Vingerhoets d a
Department of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam Institute for Addiction Research, P.O. Box 22660, 1100 DD Amsterdam, The Netherlands b Department of Endocrinology and Andrology, Ghent University Hospital, De Pintelaan 185, B-9000 Ghent, Belgium c Department of Cognitive Psychology, University of Duisburg-Essen, Germany d Laboratory for Neuropsychology, Department of Neurology, Ghent University, De Pintelaan 185, B-9000 Ghent, Belgium Received 24 November 2009; received in revised form 13 April 2010; accepted 13 April 2010
KEYWORDS Testosterone; Estradiol; Aromatase inhibitor; Risk-taking; Neurocognition
Summary Human studies on the relation between testosterone levels and risk-taking behaviour are scarce. Related functions, like aggression, have been related to higher testosterone levels more consistently, especially in the animal literature. Estradiol affects several neurotransmitter systems that play a role in behaviour regulation. Existing human studies on neurocognitive functions and testosterone levels have largely ignored the interrelatedness of testosterone levels and estradiol levels. Therefore, in this study, the effects of a 1-week combined testosterone and estradiol intervention on risk-taking behaviours were investigated. Twenty-one healthy men, with a normal body mass index, were treated for 7 days with an aromatase inhibitor (letrozole 2.5 mg), resulting in high-normal levels of testosterone and lownormal levels of estradiol, or with a combination of an aromatase inhibitor and estradiol (75 mg/ 24 h), resulting in low-normal levels of testosterone and high-normal levels of estradiol. A randomized experimenter and participant-blind controlled design was applied. Neurocognitive measures of risk-taking and reward and punishment sensitivity were assessed before starting with the medication and after 7 days of drug administration: Balloon Analogue Risk Task (BART), Game of Dice Task (GDT), and Iowa Gambling Task (IGT). A group by time effect was present for the BART, indicating that the high-normal testosterone group showed an increase in risk-taking on the BART, from the first drug-naive BART performance, to the second BART performance (aromatase inhibitor), whereas such an increase was not present in the low-normal testosterone/high estradiol group. No group by time interactions were present in GDT or IGT performance. These
* Corresponding author. Tel.: +31 020 8913761; fax: +31 020 8913701. E-mail addresses: [email protected], [email protected] (A.E. Goudriaan). 0306-4530/$ — see front matter # 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.psyneuen.2010.04.005
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
1417
results implicate that testosterone levels in healthy men are associated with increased risk-taking under conditions of unknown probabilities, but not in conditions of known probabilities (GDT) or of strategic decision making (IGT). # 2010 Elsevier Ltd. All rights reserved.
1. Introduction Animal studies have linked increases in testosterone levels to increased aggression and reduced fear, and more complex testosterone/estrogen interactions have been implicated in the expression of aggression in the animal literature (Haug et al., 1986; Trainor et al., 2006). In humans, elevated aggression levels have been linked to supraphysiological levels (pharmacological enhancement) more consistently than testosterone administration within the normal physiological range (Kouri et al., 1995; Perry et al., 2003). Although higher testosterone levels have been related to impulsive responding and reduced fear in animals, research into impulsivity and risk-taking in humans is limited. A recent study on real life risk-taking behaviour in day traders indicates that higher — naturally present — morning testosterone levels are associated with higher post-morning profits in day traders (Coates and Herbert, 2008). High testosterone levels in pubertal boys have been associated with more self-reported risk-taking behaviour (Vermeersch et al., 2008). In a laboratory study, pharmacologically heightened testosterone levels in women resulted in disadvantageous decision making in the Iowa Gambling Task (IGT, van Honk et al., 2004). In another study, physiological testosterone levels correlated negatively with IGT performance in young men, but not in older men (Reavis and Overman, 2001). Several aspects of reward processing and risk-taking influence IGT performance, and knowledge regarding risks seems to influence testosterone’s role differently (e.g., a positive effect in the day trader’s study–—where probabilities of winning or losing remain unknown, a negative effect in the IGT studies where probabilities become known during performance of the task). In addition, non-linear relationships between testosterone levels and risk-attitudes have been reported (Sapienza et al., 2009). Therefore, this study focuses on the effects of an intervention that moderately increases testosterone levels — within or near normal-physiological boundaries — on risk-taking under known and unknown probabilities. In males, testosterone is converted through aromatase (P450a) and 5a reductase in estradiol and dihydrotestosterone (DHT). When testosterone is administered, production of estradiol is changed by alterations in substrate concentration, and possibly through changed aromatase activity. Testosterone and estradiol are neuroactive steroids that exert their influence on brain regions involved in emotion and motivation (Zheng, 2009). Estradiol influences cognition in women (Spencer et al., 2008), but very little is known on the role of estradiol in males’ cognitive functioning. Therefore, in this study, both testosterone and estradiol levels were manipulated, in order to control the influence of high-normal testosterone/low estradiol and low-normal testosterone/high estradiol on risk-taking. An aromatase inhibitor in combination with estradiol results in an absolute decline in testosterone levels by 44%. In men, an increase in peripheral estradiol reflects the inhibitory tone exerted by estrogens on gonadotropin
release and is a major determinant of peripheral testosterone. A 7-day medication trial was implemented, consisting of administration of (1) an aromatase inhibitor, resulting in high-normal testosterone and low-normal estradiol, or (2) an aromatase inhibitor in combination with estradiol, resulting in low-normal testosterone, and high-normal estradiol (Raven et al., 2006). Given the literature on the enhancing effects of testosterone on risk-taking and its role in fear reduction, we hypothesize that testosterone leads to enhanced risk-taking in neurocognitive gambling tasks. In particular, stronger effects are expected on tasks with unknown or partially known/implicit risks (Balloon Analogue Risk Task, IGT) compared to tasks with known risks (Game of Dice Task).
2. Methods 2.1. Participants and test procedure Twenty-one healthy male subjects were recruited. Participants had a body mass index (BMI) < 27, normal medical history, and normal biochemical measures of haematological, hepatic, renal, gonadal and metabolic function. Exclusion criteria included smoking, performing extreme sporting activities, excessive alcohol/substance use, major psychiatric disorders, use of psychotropic drugs or drugs with a potential influence on sex hormones (e.g., glucocorticoids), recent weight loss or gain (>10%), shift work, and prostate-specific antigen levels exceeding 4 ng/mL. This study was approved by the Ethical Review Board of Ghent University Hospital, conducted in accordance with the Declaration of Helsinki, and written informed consent was obtained. Participants were randomized to receive either an aromatase inhibitor (letrozole 2.5 mg/day, orally taken on awakening, testosterone group), or letrozole plus an E2 patch (75 mg/day; estradiol group), for 7 days. Hormonal blood samples were performed before and after treatment in similar conditions at 08:00 h a.m. after an overnight fast and 10-min bed rest at the hospital. Repeated hormonal measurements, e.g., 2 days before and 2 days after a 1-week treatment were not regarded feasible as all subjects had a fulltime job. Commercial immunoassays were used to determine serum E2 (Incstar, Stillwater, MN, USA; adapted protocol with use of double amount of serum), T and SHBG (Orion Diagnostica, Espoo, Finland). Tests were administered by a trained research assistant who was blind to medication group. One person was excluded because of lack of fasting at blood sampling and one person was excluded due to low medication adherence.
2.2. Risk-taking tasks 2.2.1. Balloon Analogue Risk Task (BART) The BART assesses risk-taking behaviour under unknown probabilities (Lejuez et al., 2002; Hunt et al., 2005). In this
1418
A.E. Goudriaan et al.
computerized task, a balloon appears on the computer screen, together with a button to inflate the balloon, and a button to accumulate money. With each pump, money is accrued (5 cents per pump: this is not displayed), but when the balloon explodes all gains for that balloon are lost. Participants are instructed to earn as much money as possible by inflating the balloon, and accumulating money. Total amount of money earned in the task, and amount earned with the last balloon are displayed. Participants are instructed that at some point each balloon will explode, as early as the first pump, or as late as the point at which the balloon would fill the entire computer screen (average breakpoint 64 pumps). However, no information is given on the probability of balloon explosions. The task ends after 30 balloons are completed. The dependent measures of the BART are the adjusted total number of pumps (excluding exploded balloon pumps), total money earned, and total number of balloon explosions. 2.2.2. Game of Dice Task (GDT) The GDT assesses risk-taking under known probabilities (Brand et al., 2008). In this computerized task, participants are instructed to maximize a fictitious starting capitol of 1000s, within 18 throws of a virtual die. Before each throw, subjects are required to guess what number(s) presented on the screen will include the number that will be thrown next. With increasing numbers chosen (i.e., higher winning probability), potential wins decrease, but potential losses also decrease, e.g., choosing one number would result in winning 1000s if the number chosen is thrown, and in losing 1000s if another number is thrown, whereas choosing 3 out of 6 numbers as winning numbers would result in winning or losing 200s. After each trial, gains or losses, total money earned and numbers of trials left are indicated. The dependent measure is the number of risky choices, i.e., the frequency of choosing one or two numbers, where winning probability is less than 34%, resulting in high frequent losses in the long run (Brand et al., 2008). 2.2.3. Iowa Gambling Task (IGT) The IGT measures decision making under ambiguous conditions. Probabilities of winning and losing have to be discovered during performance of the task by using feedback
Table 1
from previous trials. A computerized IGT was used (Bechara et al., 2000). Participants choose between four decks of cards. Unknown to the subject, two decks give high rewards, but also result in high losses, and are disadvantageous in the long run. The other decks give lower rewards, but also lower losses, and result in a net gain in the long run. Participants have to learn to avoid choosing the disadvantageous decks and to choose cards from the advantageous decks. The task ends after 100 trials. Total number of cards selected from the advantageous decks is taken as the dependent measure.
2.3. Statistical analyses T-tests were performed to investigate differences in demographic and clinical variables. Separate MANOVAs were performed for each task. Group (testosterone vs. estradiol) was entered as between subjects factor, and session (baseline vs. session 2, after 1-week hormone treatment) was included as within subject factor. When multiple dependent measures were included, post hoc comparisons were made for each dependent measure separately. In addition, correlations of the dependent measures with testosterone or estradiol were examined.
3. Results 3.1. Demographics and hormone levels The two groups did not differ significantly with respect to age (t = 1.11, p = 0.28) or BMI (t = 0.94, p = 0.36). The group by time interactions were significant for hormone levels in both groups, showing an increase in testosterone levels from baseline to session 2 in the testosterone group, and a decrease in testosterone levels in the estradiol group, F(1,18) = 132.3, p < 0.0001; all testosterone levels were (slightly above) high-normal in the testosterone group, and (slightly below) low-normal in the estradiol group. Reversely, estradiol levels increased in the estradiol group but decreased in the testosterone group, F(1,18) = 26.15, p < 0.0001, resulting in (slightly below) low-normal levels in the testosterone group, and high-normal levels in the estradiol group (see Table 1).
Demographics and hormone levels. Estradiol group (n = 9) Baseline session
Age (years) Body mass index Total testosterone (ng/dL) Free testosterone (ng/dL) Total estradiol (pg/mL) Free estradiol (pg/mL)
Mean
SD
30.6 23.4 425.36 9.51 17.09 0.33
4.7 2.24 136.87 2.25 2.64 0.05
Testosterone group (n = 10) Test session 2 (aromatase 2.5 mg + estradiol 50 mg) Mean SD
Baseline session
Test session 2 (aromatase 2.5 mg)
Mean
SD
Mean
SD
245.72 5.16 25.26 0.46
33.5 24.4 495 8.8 20.5 0.37
6.7 2.40 137 2.0 3.54 0.07
988 21.5 8.9 0.18
137 4.9 0.93 0.02
126.83 2.64 13.61 0.24
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
3.2. Risk-taking tasks Dependent measures for groups and test sessions are presented in Table 2. 3.2.1.1. BART A significant time effect was present, F(3,15) = 4.26, p = 0.023, indicating a change in performance from baseline to the second session. A trend for a group effect was present, F(3,15) = 3.05, p = 0.061. A significant group time effect, F(3,15) = 4.06, p = 0.027, indicates that the testosterone group exhibited a higher increase in risk-taking behaviour during the second test session compared to the estradiol group. Post hoc comparisons revealed a non-significant higher increase in adjusted number of pumps, from baseline to the second test session, F(1,17) = 1.97, p = 0.18, the amount of money earned increased more in the testosterone group compared to the estradiol group, F(1,17) = 4.82, p = 0.04, and the number of explosions was non-significantly higher from baseline to the second session in the estradiol group compared to the testosterone group, F(1,17) = 1.63, p = 0.22. 3.2.1.2. IGT A trend for a time effect was present, indicating better performance in the second test session, compared to baseline, F(1,17) = 3.91, p = 0.064. Neither the group effect, F(1,17) = 0.19, p = 0.67, nor the group time interaction, F(1,17) = 0.05, p = 0.82, was significant 3.2.1.3. GDT A time effect was present, F(1,17) = 5.62, p = 0.03, indicating more advantageous decisions in the second session compared to baseline. Again, neither the group effect, F(1,17) = 0.06, p = 0.81, nor the group time interaction, F(1,17) = 0.66, p = 0.43, reached significance. 3.2.2. Correlations risk-taking and hormone levels A significant correlation between total testosterone and amount of money earned in the BART (r = 0.50, p < 0.05) was present in the second session, but not in the baseline Table 2
1419
session. IGT performance tended to correlate with levels of free estradiol (r = 0.41, p < 0.10) in the second session, but not in the baseline session. As can be seen in Table 3, most of the BART measures correlated significantly during session 1, but less correlation between the BART measures was present during session 2. In addition, IGT performance correlated significantly with BART number of pumps during session 1, but not during session 2, whereas the Game of Dice Task did not correlate significantly with BART or IGT measures during either of the sessions.
4. Discussion In this study, we tested the hypothesis that a group with highnormal levels of testosterone/low-normal estradiol (after a 7-day administration of an aromatase inhibitor) in contrast to a group with low-normal testosterone/high-normal estradiol levels (after a 7-day administration of an aromatase inhibitor and estradiol), would lead to higher risk-taking behaviour in neuropsychological decision-making tasks, especially in tasks with unknown probabilities. In accordance with this hypothesis, we found higher risk-taking behaviour in the BART (unknown probabilities) in the high-normal testosterone/ low-normal estradiol group, compared to the high-normal estradiol/low-normal testosterone group. These effects were not found for both IGT (in which knowledge about probabilities typically increases over task duration) and GDT (known probabilities from the beginning of the task). The specific link between high-normal testosterone/lownormal estradiol and increased risk-taking in the BART, but not in the IGT or GDT, implies that risk-taking is affected by fluctuations of testosterone within or close to normal levels, particularly when no information about the actual risk is provided. Other studies have shown that elevation of testosterone levels to supraphysiological levels shows more consistent effects compared to studies employing lower levels of testosterone. Thus, the positive effects of testosterone on risk-taking as seen in this study, are consistent with the positive effects of natural testosterone in the day traders study (Coates and Herbert, 2008). Our results are inconsistent, however, with the two previous studies which have
Balloon Analogue Risk Task, Iowa Gambling Task, and Game of Dice Task measures in the first and second test session. Estradiol Group (n = 9)
Testosterone group (n = 10)
Baseline session
Test session 2 (aromatase 2.5 mg + estradiol 50 mg)
Baseline session
Test session 2 (aromatase 2.5 mg)
Mean
SD
Mean
SD
Mean
SD
Mean
SD
32.33 3268.9 8.0
15.00 1174.9 4.58
36.72 3551.5 9.5
12.35 845.9 4.38
28.09 3390.0 8.5
10.55 558.5 4.77
41.32 4262.5 8.5
14.83 156.1 4.77
IGT N advantageous choices
45.89
11.84
55.11
18.12
44.60
11.77
51.90
15.01
GDT N advantageous throws
13.78
4.84
15.78
2.68
12.30
6.92
16.40
2.66
BART Adjusted average pumps Total money earned Number of explosions
1420 Table 3
A.E. Goudriaan et al. Pearson’s correlations between risk-taking dependent measures. Session 1 Adjusted average pumps
Total money earned
Number of explosions
N advantageous choices
BART Adjusted average pumps Total money earned Number of explosions
1.000 0.83 *** 0.63 **
1.000 0.63 **
1.000
IGT N advantageous choices
0.67 **
0.411
0.406
1.000
GDT N advantageous throws
0.07
0.01
0.17
0.11
N advantageous throws
1.000
Session 2 Adjusted average pumps
Total money earned
Number of explosions
N advantageous choices
BART Adjusted average pumps Total money earned Number of explosions
1.000 0.36 0.89 ***
1.000 0.30
1.000
IGT N advantageous choices
0.02
0.14
0.07
1.000
GDT N advantageous throws
0.34
0.08
0.43
0.08
N advantageous throws
1.000
*p < 0.05, **p < 0.01, ***p < 0.001.
indicated a negative effect of testosterone on IGT performance (Reavis and Overman, 2001; van Honk et al., 2004). One explanation for the divergent findings may be that the less extreme elevations of testosterone play a smaller role in IGT performance, especially in older men, as suggested by the study of Reavis and Overman (2001). In addition, the nonlinear relationship between testosterone and risk-taking, may explain the differential findings of the reported studies (Takahashi et al., 2006; Sapienza et al., 2009). However, it may also be that we failed to show such effects because of the relatively small sample size and a high variance in IGT performance, which are clear limitations of our study. In addition, our findings may be related to the simultaneous variation in estradiol. An animal study showed that estradiol receptors are related to the influence of femaleassociated cues on male risk taking (Kavaliers et al., 2008), but human studies are scarce. Given our preliminary findings, the mutual influence of testosterone and estradiol on risk-taking in humans has to be examined in more detail. Finally, this study was limited to adult men, and a recent null-finding in post-menopausal women shows that testosterone administration in this population did not change risk-attitudes (Zethraeus et al., 2009). Future studies therefore have to replicate these findings in larger samples, including both sexes, and both young adults and older adults. The role of stress hormones such as cortisol in decision making should also be investigated in future studies, because stress may also influence risk-taking (Starcke et al., 2008). Different levels of testosterone (supraphysiological, normal to high-normal range) seem to affect
risk-taking and impulsivity differentially. Therefore, administering several levels of testosterone could increase our knowledge on the link between testosterone levels and risk-taking and impulsivity. In conclusion, results from this study imply that an increase of testosterone in the normal to high-normal range can increase risk-taking behaviour and improve task performance on a specific risk-taking task with unknown risks.
Role of funding source This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector. Anna E. Goudriaan is supported by a new researcher grant as provided by the Dutch Scientific Organisation (NWO ZonMw, #91676084, 2007-2011).
Conflict of interest All authors have no conflicts of interest to disclose.
Acknowledgements We are indebted to Kaatje Toye and Bea Vervinckt for the meticulous realization of the study protocol and to Inge Bocquart and Kathelijne Mertens for performing the immunoassays. We thank all volunteers who participated as study subjects.
Neurocognitive risk-taking in healthy men: modulation by testosterone and estradiol levels
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.psyneuen. 2010.04.005.
References Bechara, A., Tranel, D., Damasio, H., 2000. Characterization of the decision-making deficit of patients with ventromedial prefrontal cortex lesions. Brain 123, 2189—2202. Brand, M., Heinze, K., Labudda, K., Markowitsch, H.J., 2008. The role of strategies in deciding advantageously in ambiguous and risky situations. Cogn. Process. 9, 159—173. Coates, J.M., Herbert, J., 2008. Endogenous steroids and financial risk taking on a London trading floor. Proc. Natl. Acad. Sci. U.S.A. 105, 6167—6172. Haug, M., Brain, P.F., Kamis, A.B., 1986. A brief review comparing the effects of sex steroids on two forms of aggression in laboratory mice. Neurosci. Biobehav. Rev. 10, 463—468. Hunt, M.K., Hopko, D.R., Bare, R., Lejuez, C.W., Robinson, E.V., 2005. Construct validity of the Balloon Analog Risk Task (BART): associations with psychopathy and impulsivity. Assessment 12, 416—428. Kavaliers, M., Devidze, N., Choleris, E., Fudge, M., Gustafsson, J.A., Korach, K.S., Pfaff, D.W., Ogawa, S., 2008. Estrogen receptors alpha and beta mediate different aspects of the facilitatory effects of female cues on male risk taking. Psychoneuroendocrinology 33, 634—642. Kouri, E.M., Lukas, S.E., Pope Jr., H.G., Oliva, P.S., 1995. Increased aggressive responding in male volunteers following the administration of gradually increasing doses of testosterone cypionate. Drug Alcohol Depend. 40, 73—79. Lejuez, C.W., Read, J.P., Kahler, C.W., Richards, J.B., Ramsey, S.E., Stuart, G.L., Strong, D.R., Brown, R.A., 2002. Evaluation of a behavioral measure of risk taking: the Balloon Analogue Risk Task (BART). J. Exp. Psychol. Appl. 8, 75—84. Perry, P.J., Kutscher, E.C., Lund, B.C., Yates, W.R., Holman, T.L., Demers, L., 2003. Measures of aggression and mood changes in
1421
male weightlifters with and without androgenic anabolic steroid use. J. Forensic Sci. 48, 646—651. Raven, G., de Jong, F.H., Kaufman, J.M., de, R.W., 2006. In men, peripheral estradiol levels directly reflect the action of estrogens at the hypothalamo-pituitary level to inhibit gonadotropin secretion. J. Clin. Endocrinol. Metab. 91, 3324—3328. Reavis, R., Overman, W.H., 2001. Adult sex differences on a decisionmaking task previously shown to depend on the orbital prefrontal cortex. Behav. Neurosci. 115, 196—206. Sapienza, P., Zingales, L., Maestripieri, D., 2009. Gender differences in financial risk aversion and career choices are affected by testosterone. Proc. Natl. Acad. Sci. U.S.A. 106, 15268—15273. Spencer, J.L., Waters, E.M., Romeo, R.D., Wood, G.E., Milner, T.A., McEwen, B.S., 2008. Uncovering the mechanisms of estrogen effects on hippocampal function. Front. Neuroendocrinol. 29, 219—237. Starcke, K., Wolf, O.T., Markowitsch, H.J., Brand, M., 2008. Anticipatory stress influences decision making under explicit risk conditions. Behav. Neurosci. 122, 1352—1360. Takahashi, T., Sakaguchi, K., Oki, M., Homma, S., Hasegawa, T., 2006. Testosterone levels and discounting delayed monetary gains and losses in male humans. Neuro. Endocrinol. Lett. 27, 439—444. Trainor, B.C., Kyomen, H.H., Marler, C.A., 2006. Estrogenic encounters: how interactions between aromatase and the environment modulate aggression. Front. Neuroendocrinol. 27, 170—179. van Honk, J., Schutter, D.J., Hermans, E.J., Putman, P., Tuiten, A., Koppeschaar, H., 2004. Testosterone shifts the balance between sensitivity for punishment and reward in healthy young women. Psychoneuroendocrinology 29, 937—943. Vermeersch, H., T’Sjoen, G., Kaufman, J.M., Vincke, J., 2008. The role of testosterone in aggressive and non-aggressive risk-taking in adolescent boys. Horm. Behav. 53, 463—471. Zethraeus, N., Kocoska-Maras, L., Ellingsen, T., von, S.B., Hirschberg, A.L., Johannesson, M., 2009. A randomized trial of the effect of estrogen and testosterone on economic behavior. Proc. Natl. Acad. Sci. U.S.A. 106, 6535—6538. Zheng, P., 2009. Neuroactive steroid regulation of neurotransmitter release in the CNS: action, mechanism and possible significance. Prog. Neurobiol. 89, 134—152.