| 其他摘要 | Pain imposes enormous health and economic burdens to both individuals and the whole society. To better combat the problem of pain, it is imperative to understand the neural mechanisms of pain and develop reliable objective indicators of pain. However, the majority of previous studies only focus on absolute pain sensitivity, attempting to uncover the neural indicators of the ability to perceive the same painful stimulus as more or less painful. Differential pain sensitivity (or pain discriminability), the ability to distinguish different painful stimuli, is largely overlooked. Notably, objective neural indicators of pain discriminability have important theoretical and clinical implications, such as understanding how the differences in multiple painful stimuli are encoded in the brain and assessing pain discriminability in certain patients.
To fill this gap, the present study aimed to reveal neural indicators of pain discriminability using electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) techniques. To quantify sensory discriminability, we applied signal detection theory (SDT) to six large datasets (three EEG [Datasets 1}4, total n=461}and two fMRI [Datasets 5}6, total n=399] datasets), in which Datasets 1 and 5 were used for exploration of potential neural indicators of pain discriminability, and others for assessment of the replicability and pain-selectivity of the discovered neural indicators. In each dataset except for Dataset 4, participants received transient stimuli of four sensory modalities (pain [i.e., nociceptive laser stimuli], touch [i.e., non-nociceptive electrical pulses], audition [pure tones], and vision [brief flashes of a grey round disk]) and two intensities (high and low), and reported their perceptual ratings using a 0一10 numeric rating scale, where 0 stood for "no perception" and 10 "the strongest sensation imaginable (in each stimulus modality)". In Dataset 4, participants received laser stimuli of four intensities and then rated their pain intensity ratings using the 0一10 rating scale.
EEG results showed that, in Dataset 1，high intensity painful stimuli evoked larger N1，N2, and P2 amplitudes than low intensity painful stimuli, and that the amplitude differences of these three waves in the high and low intensity conditions (high-low) correlated significantly with pain discriminability quantified with AUC. This finding was well replicated in independent Datasets 2 and 3.Results from Dataset 4 further showed that pain-evoked EEG responses consistently encoded pain discriminability when the differences of laser intensity were not too large. On the other hand, even though high intensity tactile, auditory, and visual stimuli also evoked larger N2 and P2 waves than low intensity stimuli, the amplitude differences (high-low) did not reliably correlate with tactile, auditory, and visual discriminability. Further analyses ruled out possible confounding factors: the quantification of sensory discriminability (AUC, d', and rating differences in the high and low intensity conditions [high-low]), perceptual rating differences between modalities, and AUC distribution differences between modalities all had no substantial effect on the main findings. Furthermore, a machine learning-based predictive model with pain-evoked EEG waves as features could predict pain discriminability, but failed to accurately predict tactile, auditory, and visual discriminability.
FMRI data revealed similar findings. In Dataset 5, a wide ranges of brain regions including the primary somatosensory cortex (S1), secondary somatosensory cortex (S2), insula, anterior cingulate cortex (ACC), and thalamus were more activated by high intensity painful stimuli than by low intensity painful stimuli, and the differential activations (high-low) in these regions were correlated with pain AUC values. In order to examine the replicability of these findings, region-of-interest (ROI) masks were defined as the significant voxels in the S1，S2, insula, ACC, and thalamus, and then applied to Dataset 6 to extract brain activations in these ROIs. Brain activities in the S 1，thalamus, insula, and ACC were significantly correlated with pain AUC values. On the other hand, brain activations differed between high intensity tactile, auditory, and visual stimuli conditions and low intensity conditions, but the differential activations in few or even no voxels could correlate with corresponding AUC values. Using discriminability measures other than AUC and matching AUC values between different modalities had little effect on the results. In addition, a predictive model based on pain-evoked brain activations could predict pain discriminability, but not tactile, auditory, and visual discriminability.
Overall, these results demonstrate that transient pain-evoked brain responses can serve as replicable and selective neural indicators of pain discriminability. These findings also provide a novel interpretation of classical pain-evoked brain response, deepens our understanding of pain processing, and shed light on the effective management of chronic pain. |
修改评论