Age-related differences in the functional topography of the locus coeruleus: implications for cognitive and affective functions

Reviewer #1 (Public Review):

The authors used a meta-mask based on previous LC structural studies to delineate the LC on functional scans within two large public datasets (3T CamCAN and 7T HCP).

The rostral part of the LC was characterized by connections to the posterior and anterior cingulate cortices, medial temporal lobe, hippocampus, amygdala and striatum, while the caudal part projected to the parietal cortex, occipital cortex, precentral and postcentral regions, and thalamus. Older ages were associated with less rostral-like connectivity and increased asymmetry. The gradient explained variance above the effects of age, sex and education on some emotional and cognitive measures. In particular, the old-like functional gradient (loss of rostral-like connectivity and more clustered functional organization) was associated with worse performance on emotional memory and emotion regulation tasks but not to executive functioning or self-rated sleep quality.

Participants with higher anxiety and depression also showed less rostral-like connectivity and more asymmetry. Both the aging and the anxiety/depression asymmetry manifested as less rostral-like connectivity in the left LC than the right LC.

A strength of this study is that it is the first to attempt a voxel-based approach to quantifying functional connectivity in the LC. The results finding differences between rostral and caudal LC connectivity patterns are broadly consistent with prior work indicating differences between rostral/caudal LC and should help advance understanding of the LC's connectivity patterns with cortical regions.

A limitation of the study is the challenge of assessing activity not only from the small LC brainstem nucleus but also within it. Given the current spatial limitations of whole-brain functional imaging, the current findings are bolstered by including the 7T 1.6mm isotropic data. Spatial smoothing was applied with a 3mm FWHM isotropic kernel which may have reduced precision.

Another limitation was that the authors made conclusions about clustered functional organization but it was not clear how clustering was quantified.

Reviewer #2 (Public Review):

One of the major strengths in the current study is the implementation of the fully data-driven, gradient-based method for mapping connectopies of the LC. This approach is especially suited for brain structures that are difficult to localise because the resulted connectopic mapping is relatively robust to ROI definition (Fig. 7 in Haak et al., 2018). However, as a very inclusive definition of the LC (the "meta atlas") was adopted in the study, to what extent the gradient approach can tolerate changes of accuracy and specificity for LC ROI definition is unknown. Some comparative analyses would be helpful to provide assessments on the specificity and stability of the reported gradient pattern.

Haak et al. showed distinct reproducibility within and between subjects when comparing connectopic mappings between M1 and V1. M1 connectopic mapping showed very high consistency across subjects (ICCs > 0.9) compared with V1. This is very reasonable because the functional organisation within M1 is relatively homogeneous. Regarding the reliability of the LC rostro-caudal gradient, the authors only stated that "individual gradient estimation is often not consistent", but direct measurement on the consistency across subjects for the LC gradient was missing. This is important for future LC fMRI studies as more consistent pattern might warrant the application of an atlas-based method otherwise a more individualised pipeline is needed for investigating functional dissociation in LC subregions.

It puzzles me that why a dichotomous rostral vs caudal comparison was used to demonstrate the difference in connectivity patterns along the rostro-caudal gradient which might be an oversimplistic approach as described by the authors themselves? In fact, it might be more interesting to include the central "core" LC which is structurally organized in high density (Fernandes et al., 2012) and functionally distinguishable to the peri-LC "shell" region (Totah et al., 2018; Poe et al., 2022).

The composition of rostral vs caudal connectivity pattern changes over ageing, where the loss of rostral-like connectivity was consistent in bilateral LC whereas the gain of caudal-like connectivity in older subjects was only evident in the left LC. Do authors have any explanations on this left-lateralised ageing effect which is interestingly coincided with a lot of observations such as increased left LC contrast ratios was found during ageing (Betts et al., 2017) and in PD patients (Ye et al., 2022), reduced left LC-parahippocampal gyrus connectivity was reported in aMCI patients (Jacobs et al., 2015).