Prefrontal Cortical Connectivity Mediates Locus Coeruleus Noradrenergic Regulation of Inhibitory Control in Older Adults

Abstract

Response inhibition is a core executive function enabling adaptive behavior in dynamic environments. Human and animal models indicate that inhibitory control and control networks are modulated by noradrenaline, arising from the locus coeruleus. The integrity (i.e., cellular density) of the locus coeruleus noradrenergic system can be estimated from magnetization transfer (MT)-sensitive magnetic resonance imaging (MRI), in view of neuromelanin present in noradrenergic neurons of older adults. Noradrenergic psychopharmacological studies indicate noradrenergic modulation of prefrontal and frontostriatal stopping-circuits in association with behavioral change. Here, we test the noradrenergic hypothesis of inhibitory control, in healthy adults. We predicted that locus coeruleus integrity is associated with age-adjusted variance in response inhibition, mediated by changes in connectivity between frontal inhibitory control regions. In a preregistered analysis, we used MT MRI images from N = 63 healthy humans aged above 50 years (of either sex) who performed a Stop-Signal Task (SST), with atlas-based measurement of locus coeruleus contrast. We confirm that better response inhibition is correlated with locus coeruleus integrity and stronger connectivity between presupplementary motor area (preSMA) and right inferior frontal gyrus (rIFG), but not volumes of the prefrontal cortical regions. We confirmed a significant role of prefrontal connectivity in mediating the effect of individual differences in the locus coeruleus on behavior, where this effect was moderated by age, over and above adjustment for the mean effects of age. Our results support the hypothesis that in normal populations, as in clinical settings, the locus coeruleus noradrenergic system regulates inhibitory control.SIGNIFICANCE STATEMENT We show that the integrity of the locus coeruleus, the principal source of cortical noradrenaline, is related to the efficiency of response inhibition in healthy older adults. This effect is in part mediated by its effect on functional connectivity in a prefrontal cortical stopping-network. The behavioral effect, and its mediation by connectivity, are moderated by age. This supports the psychopharmacological and genetic evidence for the noradrenergic regulation of behavioral control, in a population-based normative cohort. Noradrenergic treatment strategies may be effective to improve behavioral control in impulsive clinical populations, but age, and locus coeruleus integrity, are likely to be important stratification factors.

Keywords: functional connectivity; healthy ageing; locus coeruleus; neuromelanin; response inhibition; stop-signal task.

Figure 1.

Demographics table (A) and distribution of participants' age and cognitive scores (B). ACE-R, Addenbrooke's Cognitive Examination Revised. The boxplots display the minimum, the maximum, the sample median, and the first and third quartiles. Dots indicate individual data.

Figure 2.

SST (A) and SSRTs estimated by the ex-Gaussian race model of response inhibition (B). A, In the SST, participants respond to the direction of a black arrow by pressing the corresponding key as accurately and as quickly as possible. Occasionally, a red arrow and a tone (stop signal) require the participants to inhibit their response. The stop signal could either appear immediately after the fixation point (no-go trials) or after a short delay (SSD) that varies across trials. B, Distributions of mean SSRT and go RT. The ex-Gaussian race model depicts task performance as a race between a stop process and a go process. Successful inhibition in stop and no-go trials occurs when the stop process finishes its race before the go process. The black circles indicate the medians, the thick black segments depict the 66% quantile intervals, and the thin black segments depict the 95% quantile interval. C, Posterior predictive checks: comparing empirical data to simulated results from the fitted model. The first panel shows histograms of the observed Go RT distributions. The black lines show 1000 Go RT distributions predicted by the model. Within each of the other panels, the group-level median values are plotted separately for each response (left, right and NR, no response, when applicable) and trial type (go, stop). Please note that for reaction times and accuracy, responses in stop trials correspond to commission errors, whereas for Proportion Responses, no response in go trials are omission errors. Model predictions are represented by the median (filled circles) and 95% quantile intervals (error bars) of 100 simulated participants, randomly drawn from the joint posterior distribution.

Figure 3.

A, Sagittal, coronal, transversal view (clockwise) of group-averaged MT images. On each image the locus coeruleus (LC) is highlighted by a red square. B, Study-specific atlas of the LC (light blue, threshold 5%) and reference region in the central pons (green). C, Probabilistic LC maps thresholded at 5% presented in a 3D-reconstructed brainstem and segregated in three subregions (equal number of z slices).

Figure 4.

A, B, SSRT estimates as a function of age adjusted locus coeruleus CNR and age group. The interaction with age is estimated as a continuous variable (see text) but binarized for visualization purposes only. C, Bayesian evidence for an association between the integrity of rostral, and caudal subregions of the locus coeruleus with response inhibition.

Figure 5.

A, Age moderates the influence of connectivity between the preSMA and rIFG on response inhibition (SSRT). B, SSRT estimates as a function of connectivity between preSMA and rIFG. The interaction with age is estimated as a continuous variable (see text) but binarized for visualization purposes only.

Figure 6.

A, Moderated mediation model, with locus coeruleus integrity as a predictor (LC-CNR), functional connectivity within the prefrontal stopping-network as a mediator (connectivity), and SSRTs as an outcome. Age was included as a second-level moderator. B, Direct and indirect (i.e., mediated by functional connectivity) effects of locus coeruleus integrity on response inhibition, conditional on age. C, Results from the moderated mediation. Labels match with those identifying the model's paths on panel A.