Assessing hypo-arousal during reward anticipation with pupillometry in patients with major depressive disorder: replication and correlations with anhedonia

Abstract

Major depressive disorder (MDD) is a devastating and heterogenous disorder for which there are no approved biomarkers in clinical practice. We recently identified anticipatory hypo-arousal indexed by pupil responses as a candidate mechanism subserving depression symptomatology. Here, we conducted a replication and extension study of these findings. We analyzed a replication sample of 40 unmedicated patients with a diagnosis of depression and 30 healthy control participants, who performed a reward anticipation task while pupil responses were measured. Using a Bayesian modelling approach taking measurement uncertainty into account, we could show that the negative correlation between pupil dilation and symptom load during reward anticipation is replicable within MDD patients, albeit with a lower effect size. Furthermore, with the combined sample of 136 participants (81 unmedicated depressed and 55 healthy control participants), we further showed that reduced pupil dilation in anticipation of reward is inversely associated with anhedonia items of the Beck Depression Inventory in particular. Moreover, using simultaneous fMRI, particularly the right anterior insula as part of the salience network was negatively correlated with depressive symptom load in general and anhedonia items specifically. The present study supports the utility of pupillometry in assessing noradrenergically mediated hypo-arousal during reward anticipation in MDD, a physiological process that appears to subserve anhedonia.

Figure 1

Correlation pupil dilation and number of symptoms for depressed patients. (a) original study; (b) replication study; and (c) combined samples. Estimation of the correlation between pupil dilation during reward anticipation and the number of depressive symptoms taking measurement uncertainty of each measurement into account (represented by vertical and horizontal error bars) including regression lines (a, b, and c, left panels). The Bayesian model estimated the true correlation while accounting for measurement uncertainty by sampling from a multivariate Gaussian distribution, leading to a posterior distribution that indicates the likelihood of the modelled correlation (a, b, and c, right panels).

Figure 2

Correlation pupil dilation and number of depressive symptoms for HC and depressed patients. Estimation of the correlation between pupil dilation during reward anticipation and the number of depressive symptoms taking measurement uncertainty of each measurement into account (represented by vertical and horizontal error bars) including regression lines (left panel). The Bayesian model estimated the true correlation while accounting for measurement uncertainty by sampling from a multivariate Gaussian distribution, leading to a posterior distribution that indicates the likelihood of the modelled correlation (right panel).

Figure 3

Schematic depiction of the paradigm. Each trial started with a fixation cross, which was followed by one of three Gabor patches corresponding each to a specific condition. Both (a) reward condition and (b) neutral condition were followed by a light flash to which participants had to respond as fast as possible. Based on the speed of the response a specific feedback was provided indicating if the participant was fast enough or too slow. Given the chance to win 1 euro per trial in the reward condition, the cumulative total win was presented only in this condition. (c) was the control condition in which only the Gabor patch was presented and no response was required.