A Novel Investigation of an In-Scanner Alternative to the Cold Pressor Test in Healthy Individuals

Abstract

The cold pressor task (CPT) is widely used to study tonic pain during acute and chronic conditions and is often used as a conditioning stimulus to activate descending pain control systems. However, logistical challenges in magnetic resonance imaging (MRI) limit its application, hindering the understanding of CPT's neural dynamics. To address this, we acquired resting-state functional MRI (fMRI) data from 30 healthy participants before, during and after immersion in gelled-cold water, the closest in-scanner alternative to date to CPT for prolonged stimulation. Participants provided subjective pain intensity ratings after each scan, as well as average pain perceived during noxious stimulation, using a numeric rating scale (NRS). Following fMRI, participants rated their pain continuously during identical tonic noxious stimulation of the contralateral hand using a visual analogue scale (VAS). We employed three complementary methods to examine changes in brain function across fMRI conditions: a data-driven approach via independent component analysis (ICA), seed-to-whole-brain connectivity analysis with the periaqueductal grey (PAG) as seed and spectral dynamic causal modelling (spDCM) to explore effective connectivity changes across the dorsal anterior cingulate cortex (dACC), anterior insulae (AI), thalamus and PAG. NRS scores were significantly higher following tonic cold compared to baseline and recovery conditions. Continuous VAS reflected sustained mild-to-moderate pain over 6 min, with average VAS scores not significantly differing from NRS ratings recorded in the scanner. ICA identified engagement of descending pain control and sensorimotor networks during pain, with the latter persisting during recovery. Seed-based analysis revealed a disengagement between the PAG and cortical/subcortical regions involved in pain processing, such as the dACC, midcingulate cortex, AI, intraparietal sulcus and precuneus. Finally, spDCM revealed tonic pain neural signature was most likely characterised by top-down inhibitory and bottom-up excitatory connections. This study establishes the cold gelled-water paradigm as a potential in-scanner alternative to CPT. By uncovering key neural dynamics of CPT, we provide new insights into the brain and brainstem mechanisms of tonic cold pain paradigms routinely used in psychophysical pain studies.

Keywords: ICA; cold pressor task; fMRI; pain; seed‐based; spDCM.

FIGURE 1

Experimental design. Participants provided their current general perceived levels of pain in a numerical rating scale (NRS) from 0 (‘no pain’) to 100 (‘worst pain imaginable’) following each functional scan. Following the tonic pain scan, participants also provided the average perceived pain during tonic cold stimulation in an NRS. During the tonic pain condition, all participants immersed their right hand in a gelled‐cold water tub. Immediately following fMRI scanning, participants immersed their left hand in a gelled‐cold water tub and provided continuous pain intensity ratings in a visual analogue scale (VAS), consisting of a horizontal line anchored from 0 (‘no pain’) to 100 (‘worst pain imaginable’).

FIGURE 2

Summary of behavioural results. (A) ‘Current pain’ NRS scores were significantly higher following the Tonic Pain condition than on the other two. Average NRS scores about perceived pain intensity estimated by participants during the Tonic Pain scan did not differ from average VAS provided outside of the scanner for tonic cold stimulation on the opposite hand. (B) Average continuous VAS ratings for 30‐s bins across 6 min. Error bars represent standard error of the mean (SEM).

FIGURE 3

ICA results from components 1–4. Functional networks identified across each condition at group level (green = baseline, red = tonic pain, blue = recovery). First four components were identified by researchers as the ones capturing variability of interest and not artefactual. L = left; R = right.

FIGURE 4

Dual regression results. All maps reflect significant extension of group ICA components at alpha 0.05. Sagittal slices displayed at MNI coordinates (from right to left): 62, 50, 36, 22, 8, −6, −20, −34, −48, −62. TFCE = threshold‐free cluster enhancement. MNI = Montreal Neurological Institute. L = left; R = right.

FIGURE 5

Seed‐based connectivity results. Significant map depicts decreases in functional connectivity (FC) with the seed (i.e., PAG) during the tonic pain condition compared to baseline. Results are significant following threshold‐free cluster enhancement (TFCE) correction at alpha 0.05. ACC = anterior cingulate cortex; IPS = intraparietal sulcus; L = left; MCC = midcingulate cortex; MNI = Montreal Neurological Institute; R = right.

FIGURE 6

DCM results summary. (A) From left to right: Locations of seven nodes comprising the descending pain control pathway used for spDCM analysis. Due to the exploratory nature of the analysis, a fully connected model was initially described. (B) Results from greedy search through nested models; depiction of parametric empirical Bayes (PEB) design matrix, comprising scans from all participants across the three experimental conditions (‘Baseline’, ‘Tonic Pain’ and ‘Recovery’). Matrix columns reflect a group mean, and pairwise comparisons were specified (white colours indicate contrast loads = 1, grey colours indicate contrast loads = 0, black indicates contrast loads = −1). (B) Top row shows a depiction of connections with posterior probabilities greater than 0.75 across each contrast of interest. Off‐diagonal positive numbers (represented by yellow arrows on the corresponding bottom diagram) represent excitatory connections (from locations indicated by rows and to locations indicated by columns), and off‐diagonal negative numbers (represented by blue arrows on the corresponding bottom diagram) represent inhibitory connections. (C) From left to right: Leave‐one‐out cross‐validation (LOOCV) results. Parameters from connections with posterior probabilities greater than 0.9 for pairwise comparison results were used to predict the left‐out condition. The correlation between the predicted experimental condition and the actual condition, r is the correlation coefficient; connectivity strength across conditions from connections included in LOOCV. AI_L = left anterior insula; AI_R = right anterior insula; dACC_L = left dorsal anterior cingulate cortex; dACC_R = right dorsal anterior cingulate cortex; tha_L = left thalamus; tha_R = right thalamus; PAG = periaqueductal grey.