Evoked pain intensity representation is distributed across brain systems: A multistudy mega-analysis

Abstract

<jats:title>ABSTRACT</jats:title><jats:p>Information is coded in the brain at different scales for different phenomena: locally, distributed across regions and networks, and globally. For pain, the scale of representation is controversial. Although generally believed to be an integrated cognitive and sensory phenomenon implicating diverse brain systems, quantitative characterizations of which regions and networks are sufficient to represent pain are lacking. In this meta-analysis (or mega-analysis) using data from 289 participants across 10 studies, we use model comparison combined with multivariate predictive models to investigate the spatial scale and location of acute pain representation. We compare models based on (a) a single most pain-predictive module, either previously identified elementary regions or a single best large-scale cortical resting-state network module; (b) selected cortical-subcortical systems related to evoked pain in prior literature (‘multi-system models’); and (c) a model spanning the full brain. We estimate the accuracy of pain intensity predictions using cross validation (7 studies) and subsequently validate in three independent holdout studies. All spatial scales convey information about pain intensity, but distributed, multi-system models better characterize pain representations than any individual region or network (e.g. multisystem models explain &gt;20% more of individual subject pain ratings than the best elementary region). Full brain models showed no predictive advantage over multi-system models. These findings quantify the extent that representation of evoked pain experience is distributed across multiple cortical and subcortical systems, show that pain representation is not circumscribed by any elementary region or conical network, and provide a blueprint for identifying the spatial scale of information in other domains.</jats:p><jats:sec><jats:title>Significance Statement</jats:title><jats:p>We define modular, multisystem and global views of brain function, use multivariate fMRI decoding to characterize pain representations at each level, and provide evidence for a multisystem representation of evoked pain. We further show that local views necessarily exclude important components of pain representation, while a global full brain representation is superfluous, even though both are viable frameworks for representing pain. These findings quantitatively juxtapose and reconcile divergent conclusions from evoked pain studies within a generalized neuroscientific framework, and provide a blueprint for investigating representational architecture for diverse brain processes.</jats:p></jats:sec>

Effect sizes and test-retest reliability of the fMRI-based Neurologic Pain Signature

“…The complete NPS performance was better than constituent local brain regions for both effect size and test-retest reliability. This finding is consistent with the argument that pain is encoded in distributed brain networks instead of a specific and isolated brain region (Petre et al, 2020;. Interestingly, the six regions (i.e., bilateral insula, right dorsal posterior insula, dACC, right S2, and right thalamus) with relatively larger effect sizes and reliabilities were the targets of ascending nociceptive afferents and activated in response to pain stimuli.…”