Impaired reinforcement learning following chronic escitalopram as revealed by computational modelling and neuroimaging correlates: A double-blind, placebo-controlled semi-randomised study

Abstract:

Introduction: Reinforcement learning is vital for adaptive behaviour, as it in- volves changing our behaviour based on rewarding and punishing feedback in the environment. Several studies have probed acute or sub-chronic selective serotonin reuptake inhibitor (SSRIs) effects on reinforcement learning and have shown altered processing [1-2]. However, the chronic effects, which hold more clinical relevance, remain unclear. Understanding both acute and chronic im- pacts is pivotal, given the chronic use of SSRIs in neuropsychiatric conditions such as Major Depressive Disorder. Consequently, our study aims to explore how chronic administration of the selective serotonin reuptake inhibitor (SSRI), escitalopram, affects reinforcement learning using functional magnetic reso- nance imaging. Methods: This preregistered study (NCT04239339) used a double-blind, pla- cebo-controlled design with 66 healthy volunteers, semi-randomised to receive 20mg of escitalopram (n1⁄432) or placebo (n1⁄434) for at least 21 days. Groups were balanced for age, sex and intelligence quotient (IQ). The behavioural probabilistic reversal learning task (PRL) involved choosing between a red and a green stimulus. One stimulus would be correct 80% of the time and incorrect 20% of the time. After 40 trials there would be a reversal and the previously correct stimulus would then become incorrect. During a functional magnetic resonance imaging scan, participants completed a probabilistic learning para- digm that required making choices to maximize wins and minimize losses. In each trial one of three possible probabilistic pairs of abstract pictures was pre- sented: a rewarding (70% chance of win, 30% chance of no gain), punishing (70% of loss, 30% chance of no win) or neutral pair (100% no-change). We analysed group differences in reinforcement learning using both traditional analysis of covariance and general linear models as well as innovative hierar- chical Bayesian modelling techniques. The neuroimaging analysis was conducted using the general linear model framework and analysed according to an event- related design. Results: The escitalopram group exhibited lower reinforcement sensitivity on the PRL task compared to placebo (MD1⁄4 -2.676 [90% HDI, -5.285 to -.396]). Additionally, they displayed diminished learning from punishment during pun- ishment trials on the probabilistic learning paradigm (MD1⁄4.15 [90% HDI -.31 to -.01]) compared to controls. Another key and novel finding was that the placebo group showed heightened activation in the intraparietal sulcus during reward trials. Given its role in learning from uncertainty and encoding of outcome value [3,4] it may be that reinforcement learning is altered by serotonin through its effects on encoding of value outcomes. This alteration may reduce sensitivity to reinforcers, impacting adaptive feedback learning. Conclusions: The results have clinical implications as they may reflect the ‘blunting’ effect often reported by MDD patients receiving chronic SSRI treat- ments [5]. Understanding the impact of chronic SSRIs is essential for optimizing treatment and enhancing quality of life for patients with neuropsychiatric con- ditions, given the significance of learning from reward and punishment feedback in daily functioning and the widespread use of SSRI treatment. Our research has important clinical implications and enhances the understanding of the neural mechanisms by which serotonin influences affect, cognition, and behaviour. References [1]Kanen, J. W., Apergis-Schoute, A. M., Yellowlees, R., Arntz, F. E., van der Flier, F. E., Price, A., ... & Robbins, T. W. (2021). Serotonin depletion impairs both Pavlovian and instrumental reversal learning in healthy humans. Molecular psychiatry, 26(12), 7200-7210. [2]Skandali, N., Rowe, J. B., Voon, V., Deakin, J. B., Cardinal, R. N., Cormack, F., ... & Sahakian, B. J. (2018). Dissociable effects of acute SSRI (escitalopram) on executive, learning and emotional functions in healthy humans. Neuropsychopharmacology, 43(13), 2645-2651. [3]Tom SM, Fox CR, Trepel C, Poldrack RA. The neural basis of loss aversion in decision- making under risk. Science. 2007;315(5811):515-8. [4]Peck CJ, Jangraw DC, Suzuki M, Efem R, Gottlieb J. Reward modulates attention independently of action value in posterior parietal cortex. Journal of Neuroscience. 2009;29(36):11182-91. [5]Marazziti, D., Mucci, F., Tripodi, B., Carbone, M. G., Muscarella, A., Falaschi, V., & Baroni, S. (2019). Emotional blunting, cognitive impairment, bone fractures, and bleeding as possible side effects of long-term use of SSRIs. Clinical neuropsychiatry, 16(2), 75. Conflict of interest This study was funded by a Lundbeck Foundation Grant to Professor Barbara J Sahakian of the University of Cambridge in collaboration with Professor Gitte Moos Knudsen of the Copenhagen University Hospital Rigshospitalet. CL, GKM, SA, FK, AJ, JF, PSJ, DSS, PF have no conflicts of interest. GMK served as a speaker for Angelini, Abbvie, Cybin, and H. Lundbeck, as an advisor for Sanos, Onsero, Pangea Botanica, Gilgamesh, Pure Technologies, and her lab is a research site for Reunion and Delix Therapeutics. BJS consults for Cambridge Cognition.