Itch Matrixes

Univ Brest, LIEN, Brest, France, 2 Paris-Saclay Institute of Neuroscience, Centre National de la Recherche Scientifique/ Université Paris-Saclay, Gif-sur-Yvette, France, University Hospital of Brest, Department of Dermatology, Brest, France, Univ Brest, LATIM, INSERM UMR, Brest, France, University Hospital of Brest, Department of Radiology, Brest, France, 6 L@bISEN Yncréa Ouest, ISEN, Brest, France


INTRODUCTION
In a recent meta-analysis from our group based on a systematic review we have identified brain regions reported to be responsible for central mechanisms of itch processing (1). We also have discussed the central mechanisms of itch proceeding in the brain more in depth in a review paper (2). The research papers that have studied central mechanism of itch are presented in Table 1 while their results are presented in Table 2. Here in this paper, we are exploring a new idea in which we categorize the itch matrix in the brain into three matrixes that each of them is contributing to a specific aspect of itch perception. This conceptualizes the processing of itch signals into different itch matrices could be useful in order to model different aspects of itch. For example, it is possible, that an overactivity in second matrix cause a higher susceptivity to contagious itch. Unlike the visual system pain and itch can evoke multitude of regions in the brain, which we call pain matrix and itch matrix respectively. Recent studies have proposed that the pain matrix can be categorized into three different pain matrixes (40,41): one contributing to perception and the location of pain; another matrix responsible for the affective aspect of the pain; and a third involving decoding the cognitive aspect of pain. In the same manner, we guardedly propose that the itch processing network can be broken down into three main matrixes although many data are still lacking. These three matrixes have been presented in Figure 1.

FIRST ITCH MATRIX
The first itch matrix includes but is not restricted to the primary sensorimotor cortex, the parietal/central operculum, and the posterior insular cortex (Figure 2A).
Among these three regions the primary sensorimotor cortex is involved in the encoding of the recognition, localization, and intensity of painful stimuli (42). In pain studies, activation in this region bears a linear relationship with pain intensity (43)(44)(45)(46)(47). In a positron emission tomography (PET) study by Drzezga et al. (5) the authors reported that SI activity, is positively correlated with itch intensity. Six years after Drzezga, in 2007, Mochizuki et al. added the secondary somatosensory cortex (SII) demonstrating an increase of activity in this region after itch induction with histamine (10). The increase was statistically not different than the proven one observed in the painful condition (pain vs. itch) but did not reach a statistically corrected threshold when comparing itch against no itch.
In another study which includes both AD patients and healthy controls, itch was found to activate the post-central gyrus in the right hemisphere (12). This study together with Drzezga study in 2001 are reported in the meta-analysis on Itch from Lee et al. (48). Out of 56 regions listed in the parietal cortex (31 Left and 25 Right) from 18 studies ( Table 1). Brain activity upon itch stimulation, in (48), left SI appears to be activated eight times against two only in the right hemisphere. On the contrary, right SII is reported five times against two only in the left hemisphere. The other regions mentioned (n = 39) are in both left and right parietal cortices sometime very near to the SI/SII regions (i.e., SMG, SPL, IPL, anterior parietal cortex).
In the meta-analysis from Roberts et al. (49), the authors suggest the possibility of a specificity of these regions for the itching process as they appear to be better activated by itching than by pain. Interestingly, they also group these regions with the central operculum. In a recent meta-analysis of our group (1), SI/SII region was not clearly identified but we discussed this point regarding the diversity of studies we included. Our results on correlations with itch intensity also showed two important clusters in bilateral insular cortices (5068 voxels right 4589 voxels left) that spread to a great extent on the postcentral gyri.
The co-activation of the central operculum together with SI/SII cortex is widely reported in itch literature both in healthy subjects and patients. Indeed, central operculum corresponding to the junction of pre-and post-central gyri accompanied with the region located laterally to the posterior convolution of the insula is often confounded with insula itself or even SI. In the regions abbreviated OPC, also named rolandic operculum elsewhere, itch intensity was also correlated with PET signal both in healthy subjects and AD patients (4,33).
Finally, we propose that the insular cortex, and especially its posterior portion, takes part into this first matrix. As a common point between these regions, their gradual response with itch intensity seems important to highlight. In Leknes et al. bilateral insular and left posterior insular activity (BOLD) is correlated with histamine-induced itch intensity (9). Following Craig (50,51), Mochizuki et al. postulate that the posterior part of insula plays a different role than its anterior part (52,53). A distinction that can also find its basis on cytoarchitectural composition of these structures and their connectives with other brain areas (50,54).
Despite weak evidences in itch literature, other evidences can help to understand the insula role in processing the sensations which are common to itch and pain. Mazzola et al. explain that the two thirds of posterior insula submitted to low electrical stimulation (SEEG) directly translate these stimulations as pain sensations (55 cortex activity still correlates with noxious thermal stimulation intensity (47).
In summary, all these regions encode the feeling of itchy sensation and are somewhat translating its intensity level as well as their location following a somatotopic representation. When compared to Xiang et al. study (41), this first matrix includes all already reported regions for pain. However, studies reporting activities in those regions only for itch are rare and some studies need to be carefully interpreted given approximations inherent to main peak reporting. Effectively, secondary peaks of wide clusters or percentage of anatomical regions covered by these clusters are most often not indicated. As an example, the absence of parietal operculum in Roberts et al. study (49) needs to be put in perspective. Indeed, the point that the contrast pain-itch shows an increased activity in the parietal operculum does not mean that this region is silent in itch. Moreover, in the same study, the opposite contrast itch-pain, which reveals an implication of both right supramarginal gyrus and central operculum, could have led us to add more parietal areas to this first matrix.
So far, we have dealt with the membership of each of these brain regions in the matrix separately. However, interesting arguments reside in the fact that new pathological conditions can appear when these regions grouped and malfunction together. Hence, some studies reported that SI/SII together with the insular cortex participate in creating the allodynia phenomenon (56-59). Consecutively, these regions once activated lead to an ignition of the pain network inducing activity in the PAG, the prefrontal cortex, the thalamus, the amygdala, the ACC and many other regions within the pain network. Allodynia has repercussions on the way normal brain areas react to tactile stimuli and authors do not only consider the condition through the scope of pain matrix. Many brain areas are those involved in tactile or thermal sensitivity and this allows more faithful comparison with itch perception. The difficulty with allodynia is that even when it is spontaneous, painful sensation is quickly reached and its intensity then depends on other brain region listed above.
To illustrate this phenomenon, we adduce together both Ducreux et al. study (60) and an article from Geuter et al. (61) about predictive coding. In Ducreux et al. authors demonstrated with noxious and non-noxious cold stimulation (4 • and 22 • C) that while non-noxious cold in control subjects activates SII and the insular cortex (mostly its anterior part), the same nonnoxious stimulation did activates SII and mid-posterior insula in allodynic patients together with other regions of the pain network (60). In Geuter et al. work, the authors used the predictive coding theory of brain functioning to demonstrate a difference within the anterior and the posterior part of the insula. While the anterior part would be dedicated to pain feelings as a prediction error on perceived sensations, the posterior part only responds to pain intensity with no comparisons to any predicted sensation (61). We propose that in Ducreux et al. even if the feeling is non-noxious in control subjects, it remains unpredictable and then activates the anterior part of the insula. However, allodynic patients are prepared to feel painful

SECOND ITCH MATRIX
The second itch matrix could consist of the ACC, aMCC, aIC, amygdala, striatum and hippocampus ( Figure 2B). This network could encode the affective and motivational aspects of itch. Significant activation in the ACC, especially dorsal, extending to the anterior part of the middle cingulate cortex (aMCC), has been linked to the reward network and the positive or negative emotional response (40). Noteworthy, Vogt has reported that the aMCC reflects emotional awareness and fear leading to the questioning of the enrolment of the aMCC to the ACC gross function (62,63). Considering the anterior insula, it is reported to be involved in the awareness of emotions and subjective feelings (50) as well as errors of predictions like mentioned above. Another literature about lesions in the aIC would cause deficits in emotional awareness (e.g., alexithymia) (64). Several studies have reported that activity in the aIC is significantly correlated with the unpleasantness of itch (8-10, 18, 21). For the hippocampus, it has been also shown that this structure is fully integrated in the itch network (13,21,22). For example, only active scratching can relief activity in ipsi-hippocampal structure (53). The role of hippocampus together with amygdala, dACC and insular cortex are well-documented in Sanders and Akiyama (65). The authors noticed and argued that "amygdala and hippocampus activation appears to go hand-in-hand in most studies of itch, suggesting that the memory of previous itch experiences may be a significant factor in itch-related anxiety." Stratum possibly       1 | The itch matrix categorized into three itch matrixes. First itch matrix consisted of primary and secondary sensorimotor cortex (SI and SII, respectively), the parietal/central operculum, and the posterior insular cortex (pIC) (here presented in brown, this matrix is also presented in Figure 2A). The second itch matrix consisting of anterior singular cortex (ACC), anterior part of the middle cingulate cortex (aMCC), anterior part of the insular cortex (aIC), amygdala, striatum and hippocampus (here presented in blue, this matrix is also presented in Figure 2B). The third matrix contains prefrontal cortex, posterior part of the middle cingulate cortex (pMCC), and posterior cingulate cortex (PCC) (here presented in red, this matrix is also presented in Figure 2C).
involved with motivation aspects of itch and/or the carving for scratching.
According to original paradigms, two other studies have reported diminished activation of these regions in tasks that change the nature of pain perception with context variations (66) or with analgesia induced by meditation (67). While the first of these shows a diminished activation in dorsal ACC and insula as the subjects switch their perception from unpleasant to pleasant (or less unpleasant) revealing the link between emotional and motivational function. The second demonstrate that experienced Zen meditators can reduce activity of their prefrontal medial cortex, amygdala and hippocampus regions at the expense of an increased activity in dorsal ACC or insula which still belong to this second matrix but are more related to mindfulness. These articles suggest that making things more conscious by bringing activities closer to the awareness matrix (with insula as a common region) putatively lead to less harmful psychological consequences. This second matrix is more robust than the first one. Many arguments in the itch literature exist and converge about its functional role.

THIRD ITCH MATRIX
The third itch matrix would include parts of the prefrontal cortex, pMCC, and PCC ( Figure 2C). This network should be involved in the subjective perception of itch. The cognitive state of the mind can affect the itch sensation e.g., emotions, obsessions, religious beliefs, disgusts, expectations, and past experiences. This pattern of activation is also present in the distraction from itch caused by the Stroop task (e.g., in the DLPFC) (14,30,34). The third matrix receives and integrates information from the foregoing two and triggers behavioral response.

CONCLUSION
Knowledge of itch processing in the brain is growing thanks to brain imaging (2,68). A better understanding of interactions between itch matrixes would allow a better understanding

AUTHOR CONTRIBUTIONS
LM, J-LC, DB, and OD contributed to conception and design of the study. PN organized the database and wrote the first draft of the manuscript. OD wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

ACKNOWLEDGMENTS
We thank the French Society of Dermatology for providing a grant.