Research Paper

Primary somatosensory/motor cortical thickness distinguishes paresthesia-dominant from pain-dominant carpal tunnel syndrome Yumi Maedaa,b, Norman Kettnerb, Jieun Kima,c, Hyungjun Kima,c, Stephen Cinaa, Cristina Malatestad, Jessica Gerbera, Claire McManusd, Alexandra Libbya, Pia Mezzacappaa, Ishtiaq Mawlaa, Leslie R. Morsee, Joseph Audettef, Vitaly Napadowa,b,*

Abstract Paresthesia-dominant and pain-dominant subgroups have been noted in carpal tunnel syndrome (CTS), a peripheral neuropathic disorder characterized by altered primary somatosensory/motor (S1/M1) physiology. We aimed to investigate whether morphometry dissociates these subgroups. Subjects with CTS were evaluated with nerve conduction studies, whereas symptom severity ratings were used to allocate subjects into paresthesia-dominant (CTS-paresthesia), pain-dominant (CTS-pain), and pain/ paresthesia nondominant (not included in further analysis) subgroups. Structural brain magnetic resonance imaging data were acquired at 3T using a multiecho MPRAGE T1-weighted pulse sequence, and gray matter cortical thickness was calculated across the entire brain using validated, automated methods. CTS-paresthesia subjects demonstrated reduced median sensory nerve conduction velocity (P 5 0.05) compared with CTS-pain subjects. In addition, cortical thickness in precentral and postcentral gyri (S1/M1 hand area) contralateral to the more affected hand was significantly reduced in CTS-paresthesia subgroup compared with CTS-pain subgroup. Moreover, in CTS-paresthesia subjects, precentral cortical thickness was negatively correlated with paresthesia severity (r(34) 520.40, P 5 0.016) and positively correlated with median nerve sensory velocity (r(36) 5 0.51, P 5 0.001), but not with pain severity. Conversely, in CTS-pain subjects, contralesional S1 (r(9) 5 0.62, P 5 0.042) and M1 (r(9) 5 0.61, P 5 0.046) cortical thickness were correlated with pain severity, but not median nerve velocity or paresthesia severity. This double dissociation in somatotopically specific S1/M1 areas suggests a neuroanatomical substrate for symptom-based CTS subgroups. Such fine-grained subgrouping of CTS may lead to improved personalized therapeutic approaches, based on superior characterization of the linkage between peripheral and central neuroplasticity. Keywords: Carpal tunnel syndrome, Neuropathic pain, Paresthesia, Entrapment neuropathy, Primary somatosensory/

1. Introduction common entrapment neuropathy, with a US prevalence of 42 Many chronic neuropathic pain disorders are characterized by 3.72%. Carpal tunnel syndrome pathophysiology includes compression of the distal median nerve by an elevated pressure not just pain, but a diverse set of symptoms including numbness, 51 tingling, etc. The impact of different symptom presentations is not in the carpal tunnel, leading to altered median nerve conduction fully understood. Carpal tunnel syndrome (CTS) is the most and a range of sensory symptoms primarily in the first through fourth digits of the hand. However, multiple studies have reported lacking correlation between general CTS symptomatology and 21,35 Sponsorships or competing interests that may be relevant to content are disclosed median nerve conduction, assessed by electroneurography. at the end of this article. This discrepancy may be related to distinct, symptom-dominant a Department of Radiology, Athinoula A. Martinos Center for Biomedical Imaging, CTS subgroups. Electroneurography mainly measures A-beta b Massachusetts General Hospital, Charlestown, MA, USA, Department of fiber transmission, which has been linked to paresthesia,54,55 Radiology, Logan University, Chesterfield, MO, USA, c Korean Institute of Oriental Medicine, Daejeon, Korea, d Department of Physical Medicine and Rehabilitation, whereas pain in CTS has instead been linked to A-delta and 52 Spaulding Rehabilitation Hospital, Medford, MA, USA, e Department of Physical C-fiber transmission. Paresthesia represents the most com- Medicine and Rehabilitation, Harvard Medical School, Spaulding Rehabilitation mon CTS symptom38 and may result from ectopic impulse f Hospital, Boston, MA, USA, Department of Pain Medicine, Harvard Vanguard activity generated by ischemic nerve damage.34,39 In addition, Medical Associates, Atrium Health, Boston, MA, USA previous studies have used factor analysis to separate *Corresponding author. Address: Martinos Center for Biomedical Imaging, #2301 149 13th St, Charlestown, MA 02129, USA. Tel.: 617-724-3402; fax: 617-726- paresthesia-dominant symptom patterns from pain-dominant 7422. E-mail address: [email protected] (V. Napadow). symptom patterns, with the former more closely correlated to 56 Supplemental digital content is available for this article. Direct URL citations appear nerve conduction. in the printed text and are provided in the HTML and PDF versions of this article on Another reason for a lack of correlation between electro- the journal’s Web site (www.painjournalonline.com). neurography and symptom report may be due to variable central PAIN 157 (2016) 1085–1093 nervous system plasticity. Recent studies have demonstrated © 2016 International Association for the Study of Pain altered functional10,29,37 and structural30 properties for CTS in http://dx.doi.org/10.1097/j.pain.0000000000000486 primary somatosensory (S1) cortices of the brain. We found that

May 2016· Volume 157· Number 5 www.painjournalonline.com 1085

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. 1086 Y. Maeda et al.·157 (2016) 1085–1093 PAIN®

reduced gray matter volume within the S1 hand area contralateral pain and paresthesia (numbness/tingling). These ratings were to the more affected hand was associated with median nerve used for subgrouping subjects with CTS: paresthesia dominance sensory velocity.30 Other studies have also demonstrated (CTS-paresthesia;n5 38, greater paresthesia than pain), pain reduced gray matter volume in patients with chronic pain1 and dominance (CTS-pain;n5 11, greater pain than paresthesia), reduced cortical thickness in cases of more severe peripheral and relative pain/paresthesia equivalence (n 5 10, identical pain deafferentation, such as peripheral nerve transection.53 How- and paresthesia ratings; data not included in further analysis). To ever, the association between such structural brain changes and inform whether this specific pain and paresthesia rating scale specific symptomatology or peripheral nerve function in patients may be a stable trait of subjects’ clinical presentation, we also with chronic neuropathic pain is not well understood. assessed other, more nuanced, aspects of symptom severity Different symptom presentations may differentially impact and functional status referring to a “typical 24-hour period during brain physiology, which may improve our understanding of the past 2 weeks,” using the Boston Carpal Tunnel Questionnaire specific pathophysiology and help determine appropriate thera- (BCTQ),27 which uses a scale of 1 through 5 (1: none; 5: severe). peutic options for symptom-based patient subgroups. We Pain-related questions were combined into a subscore, with evaluated brain cortical thickness in CTS with dominant pain, averaged ratings on questions 1 through 5, whereas a paresthe- compared with dominant paresthesia. We hypothesized that sia subscore averaged ratings on questions 6 through 10. compared to CTS with dominant pain, CTS with dominant Subscales scores were used to calculate a pain/paresthesia paresthesia will demonstrate decreased cortical thickness in the ratio, which assessed the relative contribution of pain vs S1 cortical representation for the hand/wrist and that thickness paresthesia to subjects’ symptom presentation. We also will be associated with distinct (pain vs paresthesia) symptom assessed the separate BCTQ function scale (average score from severity for these specific CTS subgroups. 8 questions). Clinical data were tested for normality (Shapiro–Wilk test) and equal variance (Levene test), significant at P # 0.05 (SPSS 2. Methods version 20; Chicago, IL). Nerve conduction study data and BCTQ scores were compared between CTS-paresthesia and CTS-pain 2.1. Subject enrollment and demographics subgroups using either the Student t test or Mann–Whitney U test We enrolled 59 women with CTS (48.9 6 9.6 years, mean 6 according to statistical distribution, significant at P # 0.05. SD) with a history of pain and/or paresthesia symptoms for greater than 3 months in median nerve–innervated regions of 2.3. Magnetic resonance imaging data acquisition the hand/fingers. In addition, 24 female age-matched healthy control subjects (49.5 6 10.0 years) were enrolled. All subjects Structural brain MRI data were acquired with a multiecho were examined for eligibility by aphysiatrist(J.A.,L.R.M.)at MPRAGE T1-weighted pulse sequence (TR 5 2530 millisecond, Spaulding Rehabilitation Hospital, which included testing TE1/TE2 5 1.64/30.0 millisecond, TI 5 1200 millisecond, flip of median (digits 1-3) and ulnar (digit 5) sensory nerve angle 5 7˚, field of view 5 256 3 256, slices 5 176, sagittal conduction velocities (NCVs).Nerveconductionstudies acquisition, spatial resolution 5 1 3 1 3 1 mm3) on a 3T Siemens (Sierra EMG, Cadwell Industries, Inc., Kennewick, WA) Trio scanner (Siemens Medical, Erlangen, Germany) equipped used previously described methods.28 For subjects with with a 32-channel head coil. CTS, nerve conduction studies (NCS) inclusion criteria consisted of .3.7-millisecond sensory latency for median 2.4. Magnetic resonance imaging data analysis nerve or .0.5-millisecond sensory latency compared with ulnar sensory latency. Exclusion criteria for both CTS and For cortical thickness analyses, brain MRI structural data were healthy control subjects consisted of contraindications for preprocessed with FreeSurfer (recon-all, FreeSurfer v.5.3). magnetic resonance imaging (MRI), history of diabetes Preprocessing included motion correction, removal of nonbrain mellitus, cardiovascular, respiratory, or neurological illnesses, tissue,46 automated Talairach transformation,14,15 intensity rheumatoid arthritis, wrist fracture, current usage of pre- normalization,49 tessellation of the gray matter/ scriptive opioid medication, thenar atrophy, nerve entrapment boundary, automated topology correction,13,47 and surface other than median nerve, cervical radiculopathy or myelopathy, deformation after intensity gradients to optimally place the gray/ generalized peripheral neuropathy, blood dyscrasia or coagul- white and gray/cerebrospinal fluid borders at the location where opathy, or current use of anticoagulation therapy. All study the greatest shift in intensity defines the transition to the other protocols were approved by Massachusetts General Hospital tissue class.5,6,9,12,16–18 This method produces representations and Partners Human Research Committee. Written informed of cortical thickness calculated as the closest distance from the consent was obtained from all subjects before enrollment. gray/white boundary to the gray/cerebrospinal fluid boundary at each vertex on the tessellated surface12 and has been validated against histological analysis.26,44,45 Surface cortical thickness 2.2. Clinical data data were then smoothed using 2-dimensional surface smooth- Nerve conduction studies were used to calculate relative median ing (10 mm full width at half maximum, mris_preproc, Free- NCV, by subtracting ulnar (digit 5) from median (digit 2) sensory Surfer v.5.3). velocity. Although studies have used both raw and ulnar- For subjects with CTS whose more affected hand was the left normalized median sensory velocities as outcomes, the relative hand, cortical thickness data were flipped across the midsagittal measures may demonstrate higher diagnostic accuracy,2 as they plane before these data were passed up to group analyses with control for nonspecific intersubject differences, ie, they are less right hand–affected subjects. This allowed for data interpretation influenced by factors such as age, height, weight, and hand relative to the hand demonstrating median nerve pathology and is temperature. specifically important for lateralized structures (ie, S1/M1). A Subjects’ symptom severity was assessed by separate 11- nonflipped analysis explored if the main results were significantly point (0: none; 10: very strong) Numerical Rating Scales (NRS) for influenced by data flipping. In the entire sample, 16 patients were

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. May 2016· Volume 157· Number 5 www.painjournalonline.com 1087

more affected on the left hand. Of these 16 patients, only 1 had P 5 0.74) between CTS-paresthesia and CTS-pain subgroups unilateral CTS, whereas 15 had bilateral CTS, based on our (Table 1). Nerve conduction testing found reduced relative clinical/electrodiagnostic criteria. Conversely, 33 patients were (median–ulnar) sensory median NCVs for CTS-paresthesia sub- more affected on the right hand, with 11 demonstrating unilateral group compared with CTS-pain subgroup (t(46) 5 1.99; P 5 CTS, whereas 22 had bilateral CTS. 0.05). Boston Carpal Tunnel Questionnaire scores reflected CTS Preprocessed cortical thickness data were contrasted be- subgroup allocation, suggesting that the ratings on our NRS tween CTS-paresthesia and CTS-pain subgroups with age as reflected a stable trait of subjects’ clinical presentation. Specifically, a regressor of no interest using a general linear model (mri_glmfit, for CTS-paresthesia subjects, NRS pain was rated as 3.8 6 2.6 FreeSurfer v.5.3) and corrected for multiple comparisons at (mean 6 SD), whereas NRS paresthesia was rated as 6.4 6 2.3. cluster-wise P 5 0.05 (mri_glmfit-sim, FreeSurfer). Intracranial Conversely, for CTS-pain subjects, NRS pain was rated as 5.4 6 volume did not differ between groups (P 5 0.42). For region of 2.1, whereas NRS paresthesia was rated as 4.1 6 2.0. The interest follow-up analyses involving significant clusters within pain–paresthesia metric, which was used for subgroup classifica- precentral and postcentral gyri, a software-generated annotation tion, was significantly different between the 2 groups (CTS-pain: file (aparc.annot, FreeSurfer) was used to determine precentral 1.35 6 0.58, CTS-paresthesia: 22.59 6 2.28, t(46) 529.55, P , and postcentral gyral localizations for both hemispheres. 0.001; unequal variances, Levene test). The BCTQ-derived pain/ Importantly, the number of subjects required for cortical thickness paresthesia ratio was significantly greater for CTS-pain subgroup analyses is dependent on cortical location,43 with fewer subjects than for CTS-paresthesia subgroup (t(47) 5 3.16; P 5 0.003). required for several regions including those adjacent to the central There were no differences in overall BCTQ symptom severity . This is likely due to such factors as reduced intersubject (t(47) 520.44, P 5 0.66) or function scale (t(46) 520.71, P 5 variability in cortical thickness and ease of surface topology 0.48) scores between CTS-pain and CTS-paresthesia groups. In modeling for regions adjacent to the . Thus, fewer general, the BCTQ symptom scale score across all subjects in our subjects are required because of a lower variance, which stems study was 2.72 6 0.66, which is highly consistent with the BCTQ from greater consistency in folding across individuals and the fact symptom scale score reported in previous large sample (N 5 403) that intersubject registration is based on aligning cortical folding studies,40 suggesting that our cohort was at least grossly reflective patterns in FreeSurfer.43 Correlation analyses associated cortical of the general CTS population. thickness and clinical variables (NCV and BCTQ), significant at P # 0.05. In addition, as our previous study directly contrasted structural brain data from subjects with CTS (a subset, N 5 20, of 3.2. Magnetic resonance imaging data analysis: the larger sample, N 5 59, reported herein) vs healthy controls cortical thickness and reported significant gray matter volume differences between A whole-brain cortical thickness analysis demonstrated that CTS- patients with CTS and healthy adults,30 this study did not directly pain subjects had significantly greater cortical thickness in contrast CTS data with healthy controls. In this study, healthy a cluster covering left (contralesional to the more affected hand) control subject data from significant clusters identified above precentral (putative M1), (putative S1), were used to descriptively aid interpretation of how differences another cluster covering left , and a third between CTS-pain and CTS-paresthesia cortical thickness relate cluster covering right (ipsilesional to the more affected hand) to cortical thickness measures in age-matched and sex-matched precentral gyrus (M1) and superior/ (consis- healthy adults. tent with ) (Table 2, Figure 1, similar results were also seen for a nonflipped analysis, Supplementary Figure 1, 3. Results available online as supplemental digital content at http://links. lww.com/PAIN/A219). The contralesional S1/M1 cluster was 3.1. Demographic and clinical assessments localized to the hand area of the somatotopic map, as supported There were no differences in age (t(12) 520.93; P 5 0.37; unequal by overlaid outlines of finger stimulation clusters from our previous variances, Levene test) or symptom duration (t(47) 520.33; functional MRI results29 and more clearly presented with idealized

Table 1 Demographics and clinical assessments. Healthy control CTS-pain CTS-paresthesia CTS-pain vs (N 5 24, 24 F) (N 5 11, 11 F) (N 5 38, 38 F) CTS-paresthesia P Age (yr) 49.5 6 10.0 47.4 6 11.5 50.7 6 7.0 0.37 Symptom duration (yr) n/a 8.4 6 9.0 9.3 6 8.5 0.74 Nerve conduction study Median–ulnar sensory velocity (m/s) 20.8 6 4.2 214.4 6 5.1 219.3 6 7.4 0.05 Symptom severity Numerical Rating Scale (0-10) Pain rating n/a 5.4 6 2.1 3.8 6 2.6 0.07 Paresthesia rating n/a 4.1 6 2.0 6.4 6 2.3 0.005 Pain–paresthesia (used for classification) n/a 1.35 6 0.58 22.59 6 2.28 <0.001 Boston Carpal Tunnel Syndrome questionnaire Symptom severity score (1-5) n/a 2.6 6 0.5 2.7 6 0.7 0.66 Function status score (1-5) n/a 1.9 6 0.8 2.2 6 0.8 0.48 Pain/paresthesia ratio n/a 1.17 6 0.24 0.87 6 0.28 0.003 Data are shown as mean 6 SD. n/a, not applicable. Bold values refer to significant P at level 0.05. CTS, carpal tunnel syndrome.

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. 1088 Y. Maeda et al.·157 (2016) 1085–1093 PAIN®

Table 2 Brain cortical thickness differences between carpal tunnel syndrome CTS-pain and CTS-paresthesia subgroups. Region X (mm) Y (mm) Z (mm) CTS-pain - CTS-paresthesia z-stat Total cluster size (mm2) Left (contralesional to more affected hand) hemisphere Precentral gyrus (M1) 241.2 211.9 59.3 3.86 829 Postcentral gyrus (S1) 239.7 232.5 47.4 2.56 829 Middle temporal gyrus 262.9 235.4 213.3 3.33 947 Right (ipsilesional to more affected hand) hemisphere 19.2 23.2 64.4 4.79 2621 Precentral gyrus (M1) 37.5 210.8 61.8 2.67 2621

Brodmann area borders for M1 and S1, following Moore et al.36 There were no brain regions that exhibited greater cortical Cortical thickness in the left precentral and postcentral gyri (but thickness for CTS-paresthesia subgroup than for CTS-pain not middle temporal gyrus) for healthy control subjects (plotted subgroup. for descriptive interpretation) demonstrated an intermediate For CTS-paresthesia subjects, contralesional precentral (M1) mean value, between the CTS-pain and CTS-paresthesia cortical thickness was positively correlated with median nerve subgroups (Figure 2). As cortical thickness differences may sensory velocity (r(36) 5 0.51, P 5 0.001) and negatively have been due to differences in median nerve conduction correlated with BCTQ paresthesia score (r(34) 520.40, P 5 between groups, we also ran similar analyses controlling for 0.016, Figure 3, Table 3). Thus, for the CTS-paresthesia relative median NCV, which yielded similar results with a subset subgroup, worse clinical presentation, as evidenced by greater of the brain regions noted above (right/ipsilesional M1, superior paresthesia severity, was associated with thinner gray matter frontal and temporal gyrus; contralesional superior temporal in contralesional M1. Conversely, for subjects with CTS-pain, gyrus, Supplementary Figure 2, available online as supple- apositivecorrelationwasfound between cortical thickness mental digital content at http://links.lww.com/PAIN/A219). in both postcentral (S1; r(9) 5 0.62, P 5 0.042) and precentral

Figure 1. Carpal tunnel syndrome (CTS)-pain and CTS-paresthesia cortical thickness difference map. A whole-brain cortical thickness analysis demonstrated that CTS-pain subjects had significantly greater gray matter thickness in bilateral precentral gyrus (putative M1), left (contralesional to the more affected hand) postcentral gyrus (putative S1), left middle temporal gyrus, and right (ipsilesional to the more affected hand) superior/middle frontal gyrus (consistent with premotor cortex). The S1/M1 cluster was localized to the hand area of the somatotopic map, as supported by the overlaid outline (yellow) of activation clusters for median nerve-innervated digits of the affected hand (from our previous functional magnetic resonance imaging publication29). The cluster showed separate local maxima in both purported M1 (BA4) and S1 (BA1/BA2), more clearly presented with idealized borders for M1 and S1, following Moore et al.36

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. May 2016· Volume 157· Number 5 www.painjournalonline.com 1089

Figure 2. Cortical thickness in carpal tunnel syndrome (CTS)-pain and CTS-paresthesia subjects compared with healthy controls (HCs). Matched HC subjects exhibited an intermediate mean cortical thickness between CTS-pain and CTS-paresthesia subgroups in S1/M1 but not in the middle temporal gyrus. Error bars represent SEM.

(M1; r(9) 5 0.61, P 5 0.046) gyri and BCTQ pain score Although paresthesia and pain are common symptoms (Figure 3, Table 3). Thus, for the CTS-pain subgroup, worse associated with CTS, evidence supports distinct pathophysiology clinical presentation, as evidenced by greater pain severity, was underlying these different symptoms. First, a factor analysis on associated with thicker gray matter in contralesional S1 BCTQ found that paresthesia and pain load onto different factors and M1. and that the primary paresthesia factor was more closely For CTS-paresthesia subgroup, cortical thickness in contrale- correlated with nerve conduction measures.56 This result was sional M1 (r(36) 520.21, P 5 0.21) and S1 (r(36) 5 0.07, P 5 0.69) recently corroborated in a large sample cohort.40 Both pares- was not correlated with BCTQ pain score; whereas for CTS-pain thesia and pain result from median nerve compression due to subgroup, cortical thickness in contralesional M1 (r(9) 520.23, P 5 increased carpal tunnel pressure, leading to ischemia and 0.50) and S1 (r(9) 5 0.34, P 5 0.30) was not correlated with BCTQ pathohistochemical sequelae. However, experimentally induced paresthesia score. Furthermore, for CTS-pain subgroup, cortical ischemia produces paresthesia and numbness, not necessarily thickness in contralesional M1 (r(9) 5 0.12, P 5 0.73) and S1 (r(9) 5 pain, in patients with CTS.22 Paresthesia severity has been 0.35, P 5 0.33) was not correlated with median NCV (Table 3). In closely linked with median nerve conduction studies.55 As nerve addition, specificity for pain and paresthesia was highlighted by the conduction studies are believed to preferentially assay large fact that gross BCTQ symptom (SSS) and function (FSS) scale diameter, myelinated A-beta fibers,4 CTS-induced paresthesia scores were not correlated with cortical thickness in contralesional has been strongly associated with damage and demyelination of S1 (BCTQ-SSS: r(36) 520.09, P 5 0.57; BCTQ-FSS: r(35) 52 nonnociceptive A-beta fibers,55 similar to neuropathic pain 0.09, P 5 0.59) for CTS-paresthesia subjects or in contralesional syndromes.23 In fact, studies of nerve compression in awake M1 (r(9) 5 0.39, P 5 0.23 and r(9) 520.69, P 5 0.83, respectively) human subjects have demonstrated that ischemia is related to or S1 (r(9) 5 0.48, P 5 0.13 and r(9) 520.08, P 5 0.82, paresthesia, resulting from an abnormal spatiotemporal pattern respectively) for CTS-pain subjects. of ectopic impulses from different sensory units, accompanied by prolonged bursts of high-frequency afference in the most intense stage of paresthesia.34,39 Conversely, pain in CTS has been 4. Discussion linked with small diameter, myelinated A-delta fiber damage.55 Our study investigated the differentiation in structural brain Because pain severity does not correlate with nerve conduction morphometry by CTS symptom patterns and associated median measures, pain in CTS may arise from weaker, but still chronic, nerve function. We found a double dissociation for pain-dominant median nerve compression, which preferentially damages vs paresthesia-dominant CTS subgroups. First, compared with smaller A-delta fibers.55 Our results support this hypothesis, as CTS-pain subjects, CTS-paresthesia subjects demonstrated we also found median nerve conduction to be more reduced in worse median nerve function and reduced cortical thickness for CTS-paresthesia subgroup, compared with CTS-pain subgroup, several brain regions, including S1 (hand area, contralateral to with no differences in disease duration or BCTQ symptom more affected hand) and M1 (bilateral). In addition, contralesional severity scores between the 2 groups. Thus, pain or paresthesia M1 cortical thickness in CTS-paresthesia subjects was correlated dominance in CTS is not simply a consequence of disease with median nerve velocity and paresthesia severity, but not pain duration or severity and likely depends on distinct pathophysio- severity. Conversely, in CTS-pain subjects, contralesional S1 and logical factors.41 M1 cortical thickness was correlated with pain severity, but not Compared with subjects with CTS-pain, CTS-paresthesia median nerve velocity or paresthesia severity. These results subjects demonstrated reduced S1 and M1 cortical thickness, support the existence of dichotomous subgroups of patients with which was correlated with peripheral nerve dysfunction and CTS based on symptomatology and suggest a neuroanatomical self-rated paresthesia severity. Our previous study30 used substrate for this dichotomy. whole-brain voxel-based morphometry analysis and noted

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. 1090 Y. Maeda et al.·157 (2016) 1085–1093 PAIN®

Figure 3. Cortical thickness correlates with clinical measures in carpal tunnel syndrome (CTS)-pain and CTS-paresthesia subjects. For CTS-paresthesia subjects, contralesional precentral (M1) cortical thickness was positively correlated with relative median sensory nerve conduction velocity (r(36) 5 0.51, P 5 0.001) and negatively correlated with Boston Carpal Tunnel Questionnaire (BCTQ) paresthesia score (r(34) 520.40, P 5 0.016). Thus, worse clinical presentation, as evidenced by slower median nerve sensory conduction velocity and greater paresthesia severity, was associated with thinner gray matter in contralesional M1. Conversely, for CTS-pain subjects, contralesional postcentral (S1) and precentral (M1) cortical thickness was positively correlated with BCTQ pain score (S1: r(9) 5 0.62, P 5 0.042; M1: r(9) 5 0.61, P 5 0.046, respectively) but not paresthesia severity or nerve conduction velocity (S1: P 5 0.30, P 5 0.33; M1: P 5 0.50, P 5 0.73, respectively).

reduced gray matter volume in contralesional S1 hand area for dissociation noted in this study between specific (ie, pares- subjects with CTS. In that mixed (combined paresthesia- thesia vs pain) symptoms and different symptom-dominant dominant and pain-dominant) sample, reduced gray matter CTS subgroups. volume was weakly correlated with median NCV but not with Interestingly, altered cortical thickness was noted not just in symptomatology. Reduced gray matter volume is broadly contralesional S1 but also in M1. Moreover, contralesional M1 consistent with our current observation of reduced S1 cortical thickness in CTS-paresthesia subgroup was positively correlated thickness for CTS-paresthesia subgroup, as this subgroup with NCV and negatively correlated with paresthesia severity—ie, represented most subjects with CTS previously evaluated, worse clinical presentation was associated with reduced and paresthesia is the most common CTS symptom.38 Lack contralesional M1 thickness. Reduced cortical thickness in of significant correlation with symptom severity for the previ- contralesional M1 may be due to reduced hand and arm ous CTS cohort is also consistent with the double movement in CTS-paresthesia subjects vs CTS-pain subjects.

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. May 2016· Volume 157· Number 5 www.painjournalonline.com 1091

Table 3 Correlations between cortical thickness and clinical measures. BCTQ pain BCTQ paresthesia Median–ulnar NCV CTS-paresthesia subgroup Left precentral gyrus (contralesional M1) Correlation coefficient R 20.212 20.403 0.506 P 0.202 0.012 0.02 Left postcentral gyrus (contralesional S1) Correlation coefficient R 0.066 20.063 0.053 P 0.692 0.708 0.752 CTS-pain subgroup Left precentral gyrus (contralesional M1) Correlation 0.614 20.227 0.120 P 0.045 0.502 0.725 Left postcentral gyrus (contralesional S1) Correlation 0.619 0.343 0.346 P 0.042 0.302 0.327 BCTQ, Boston Carpal Tunnel Questionnaire; CTS, carpal tunnel syndrome; NCV, nerve conduction velocity. Bold values refer to significant P at level 0.05.

In fact, patients with dystonia, whose affected hand was severity. Thus, greater pain severity was associated with greater immobilized for 4 weeks, demonstrated reduced contralesional cortical thickness in these regions. Increased gray matter volume M1 gray matter density.20 However, CTS-paresthesia subjects or cortical thickness has been reported in other chronic pain did not differ in disease duration or BCTQ functional scores disorders.50 Specifically for S1/M1 regions, patients with low- (which refer to hand and arm movement activities) compared with back pain exhibit increased S1 (low back representation) cortical CTS-pain subjects, and M1 cortical thickness was not associated thickness,24 whereas patients with trigeminal neuropathic pain8 with BCTQ function scores. This suggests that neither disrupted and migraine7 also demonstrate increased cortical thickness in behavioral patterns nor chronicity can explain the altered cortical somatosensory cortices. In addition, greater thermal and pain thickness in contralesional M1. In addition, our previous study sensitivity correlates with cortical thickness in somatosensory noted altered white matter microstructure in contralesional S1/ cortex.11 Thus, chronic ongoing pain in patients with CTS M1 U-fiber cortico-cortical tracts, which was also associated with suffering with predominant pain symptomatology in the hand/ NCV.30 Thus, reduced M1 cortical thickness in CTS-paresthesia wrist, secondary to A-delta fiber damage,55 also leads to subgroup may be a direct consequence of more profound experience-dependent cortical thickening in hand area S1 peripheral nerve dysfunction and concomitant paresthesia regions. Consequently, such A-delta fiber damage may heighten perception in these subjects, leading to disrupted hand area inflammatory response and central sensitization with upregulated S1/M1 connectivity and information transfer along fiber tracts afferent input to the somatosensory cortex,3 resulting in connecting these 2 regions. experience-dependent thickening of these and closely associ- In addition to CTS, purported cortical thinning has been ated (eg, M1) brain regions. reported for other chronic pain populations, including temporo- The direct physiological linkage between peripheral nerve mandibular disorder33 and fibromyalgia.25 Moreover, cortical pathophysiology in CTS and purported structural brain neuro- thinning has been reported for severe peripheral nerve injury53 plasticity is not well understood. Median nerve compression and diabetic neuropathy,19 where thinning was demonstrated results in ischemia, inflammation, and elevated tunnel pressure, specifically in the primary somatosensory cortex. Thus, peripheral thereby producing altered afferent input from affected digits and nerve damage, secondary to trauma and/or associated with decreased nerve conduction along median nerve sensory fibers. neuropathic pain, seems to be closely linked with S1 cortical Thus, S1 cortical thickness changes may result from partial thinning and/or gray matter volume reduction. Carpal tunnel deafferentation or desynchronization in S1 input. Cortical thinning syndrome–induced paresthesia is believed to result specifically and gray matter reduction may be mediated by neuronal or glial from A-beta fiber damage and represents a more paroxysmal, death,31 or loss of dendritic spine density.32 Ultimately, neuronal, intermittent symptom compared with pain.55 In addition, numb- synaptic, glial, and even vascular remodeling can influence ness is more closely linked with paresthesia than pain symptoms macroscopic MRI-based markers of plasticity.57 Furthermore, as in CTS.56 Paroxysmal paresthesia and numbness likely result structural plasticity after chronic pain may be reversible,48 future from partial deafferentation and reduced median nerve velocities, longitudinal studies should evaluate whether effective treatment leading to S1/M1 cortical thinning when paresthesia predom- that ameliorates symptomatology and improves median NCVs inates over pain in patients with CTS. In our data, S1/M1 thinning can also normalize gray matter volume and/or cortical thickness. may have resulted from differences in median NCVs between Limitations of our study should also be noted. For instance, the groups (and not symptom dominance per se). Thus, we repeated number of subjects in CTS-paresthesia subgroup was significantly our analyses, controlling for relative median NCV, and demon- greater than in CTS-pain subgroup. To address this potential bias, strated that several previously identified regions (eg, ipsilateral we performed additional analyses using the same number of age- S1/M1) did indeed still demonstrate significantly reduced cortical matched and sex-matched subjects in both groups (the reduced thickness (Supplementary Figure 2, available online as supple- CTS-paresthesia subgroup was extracted from the main sub- mental digital content at http://links.lww.com/PAIN/A219). group by highest BCTQ paresthesia scores, yielding otherwise In contrast, CTS-pain subjects not only demonstrated greater matched demographics, Supplementary Table 1, available online S1/M1 cortical thickness than CTS-paresthesia subjects but also as supplemental digital content at http://links.lww.com/PAIN/ showed a positive association between cortical thickness in the A219, and an otherwise identical analysis). We found very similar hand representation area and self-rated pain (but not paresthesia) cortical thickness differences compared with the whole group

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. 1092 Y. Maeda et al.·157 (2016) 1085–1093 PAIN®

analysis (Supplementary Table 2, available online as supplemental [6] Dale AM, Sereno MI. Improved Localizadon of cortical activity by digital content at http://links.lww.com/PAIN/A219), suggesting combining EEG and MEG with MRI cortical surface reconstruction: a linear approach. J Cogn Neurosci 1993;5:162–76. that sample size differences did not contribute significantly to our [7] DaSilva AF, Granziera C, Snyder J, Hadjikhani N. Thickening in the neuroimaging results. Furthermore, we should note that the somatosensory cortex of patients with migraine. Neurology 2007;69: number of subjects required for cortical thickness analyses is 1990–5. dependent on cortical location,43 with fewer subjects required for [8] DaSilva AF, Becerra L, Pendse G, Chizh B, Tully S, Borsook D. several regions including those adjacent to the central sulcus, Colocalized structural and functional changes in the cortex of patients with trigeminal neuropathic pain. PLoS One 2008;3:e339. where our key findings were localized. [9] Desikan RS, Segonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, In conclusion, we found a double dissociation for pain- Buckner RL, Dale AM, Maguire RP, Hyman BT, Albert MS, Killiany RJ. An dominant vs paresthesia-dominant CTS subgroups. CTS- automated labeling system for subdividing the human on paresthesia subjects demonstrated worse median nerve MRI scans into gyral based regions of interest. Neuroimage 2006;31: 968–80. function and reduced S1/M1 cortical thickness, with the latter [10] DhondR,RuzichE,WitzelT,MaedaY,MalatestaC,MorseL,AudetteJ, associated with median nerve velocity and paresthesia Hamalainen M, Kettner N, Napadow V. Spatiotemporal mapping severity, but not pain severity. Conversely, in CTS-pain cortical neuroplasticity in carpal tunnel syndrome. Brain 2012;135: subjects, S1/M1 cortical thickness was correlated with pain 3062–73. severity, but not median nerve velocity or paresthesia severity. [11] Erpelding N, Moayedi M, Davis KD. Cortical thickness correlates of pain and temperature sensitivity. Pain 2012;153:1602–9. Thus, dichotomous symptom-based CTS subgroups may be [12] Fischl B, Dale AM. Measuring the thickness of the human cerebral cortex supported by distinct neuroplasticity in S1/M1. More refined from magnetic resonance images. Proc Natl Acad Sci U S A 2000;97: subgrouping for CTS and other chronic neuropathic pain 11050–5. disorders based on objective measures of neuroplasticity may [13] Fischl B, Liu A, Dale AM. Automated manifold surgery: constructing geometrically accurate and topologically correct models of the human lead to improved, personalized therapy that may focus on cerebral cortex. IEEE Trans Med Imaging 2001;20:70–80. central (S1/M1) and also peripheral (median nerve) targets. [14] Fischl B, Salat DH, van der Kouwe AJ, Makris N, Segonne F, Quinn BT, Future longitudinal studies should evaluate whether changes in Dale AM. Sequence-independent segmentation of magnetic resonance cortical thickness precede or follow changes in symptom images. Neuroimage 2004;23(suppl 1):S69–84. severity, specific to paresthesia and pain, to help determine [15] Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, van der Kouwe A, Killiany R, Kennedy D, Klaveness S, Montillo A, Makris N, Rosen such therapeutic targeting. B, Dale AM. Whole brain segmentation: automated labeling of neuroanatomical structures in the . Neuron 2002;33: 341–55. Conflict of interest statement [16] Fischl B, Sereno MI, Dale AM. Cortical surface-based analysis. II: inflation, flattening, and a surface-based coordinate system. Neuroimage 1999;9: The authors have no conflicts of interest to declare. 195–207. This work was supported by the following NIH grants: R01- [17] Fischl B, Sereno MI, Tootell RB, Dale AM. High-resolution intersubject AT004714, R01-AT004714-02S1, P01-AT002048, P41EB015896, averaging and a coordinate system for the cortical surface. Hum Brain S10-RR021110, S10RR023401, and S10-RR023043. Mapp 1999;8:272–84. [18] Fischl B, van der Kouwe A, Destrieux C, Halgren E, Segonne F, Salat DH, Busa E, Seidman LJ, Goldstein J, Kennedy D, Caviness V, Makris N, Rosen B, Dale AM. Automatically parcellating the human cerebral cortex. Appendix A. Supplemental Digital Content Cereb Cortex 2004;14:11–22. Supplemental Digital Content associated with this article can be [19] Frokjaer JB, Brock C, Softeland E, Dimcevski G, Gregersen H, Simren M, M Drewes A. Macrostructural brain changes in patients with longstanding found online at http://links.lww.com/PAIN/A219. type 1 diabetes mellitus—a cortical thickness analysis study. Exp Clin Endocrinol Diabetes 2013;121:354–60. [20] Granert O, Peller M, Gaser C, Groppa S, Hallett M, Knutzen A, Deuschl Supplemental media G, Zeuner KE, Siebner HR. Manual activity shapes structure and function in contralateral human motor hand area. Neuroimage 2011;54: A video abstract accompanying this article can be found online at 32–41. available online as supplemental digital content at http://links. [21] Green TP, Tolonen EU, Clarke MR, Pathak P, Newey ML, Kershaw CJ, lww.com/PAIN/A220. Kallio MA. The relationship of pre- and postoperative median and ulnar nerve conduction measures to a self-administered questionnaire in carpal tunnel syndrome. Neurophysiol Clin 2012;42:231–9. Article history: [22] Han SE, Boland RA, Krishnan AV, Vucic S, Lin CS, Kiernan MC. Received 17 August 2015 Ischaemic sensitivity of in carpal tunnel syndrome. J Peripher Nerv Received in revised form 12 October 2015 Syst 2009;14:190–200. Accepted 23 October 2015 [23] Jensen TS, Baron R. Translation of symptoms and signs into Available online 5 January 2016 mechanisms in neuropathic pain. Pain 2003;102:1–8. [24] Kong J, Spaeth RB, Wey HY, Cheetham A, Cook AH, Jensen K, Tan Y, Liu H, Wang D, Loggia ML, Napadow V, Smoller JW, Wasan AD, Gollub RL. S1 is associated with chronic low back pain: a functional and References structural MRI study. Mol Pain 2013;9:43. [1] Apkarian AV, Sosa Y, Sonty S, Levy RM, Harden RN, Parrish TB, Gitelman [25] Kuchinad A, Schweinhardt P, Seminowicz DA, Wood PB, Chizh BA, DR. Chronic back pain is associated with decreased prefrontal and Bushnell MC. Accelerated brain gray matter loss in fibromyalgia patients: thalamic gray matter density. J Neurosci 2004;24:10410–5. premature aging of the brain? J Neurosci 2007;27:4004–7. [2] Atroshi I, Gummesson C, Johnsson R, Ornstein E. Diagnostic properties [26] Kuperberg GR, Broome MR, McGuire PK, David AS, Eddy M, Ozawa F, of nerve conduction tests in population-based carpal tunnel syndrome. Goff D, West WC, Williams SC, van der Kouwe AJ, Salat DH, Dale AM, BMC Musculoskelet Disord 2003;4:9. Fischl B. Regionally localized thinning of the cerebral cortex in [3] Baron R. Mechanisms of disease: neuropathic pain–a clinical schizophrenia. Arch Gen Psychiatry 2003;60:878–88. perspective. Nat Clin Pract Neurol 2006;2:95–106. [27] Levine DW, Simmons BP, Koris MJ, Daltroy LH, Hohl GG, Fossel AH, Katz [4] Cruccu G, Anand P, Attal N, Garcia-Larrea L, Haanpaa M, Jorum E, Serra JN. A self-administered questionnaire for the assessment of severity of J, Jensen TS. EFNS guidelines on neuropathic pain assessment. Eur J symptoms and functional status in carpal tunnel syndrome. J Bone Joint Neurol 2004;11:153–62. Surg Am 1993;75:1585–92. [5] Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. [28] Ma D, Liveson J. Nerve Conduction Handbook. Philadelphia: FA Davis Segmentation and surface reconstruction. Neuroimage 1999;9:179–94. Co, 1983.

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited. May 2016· Volume 157· Number 5 www.painjournalonline.com 1093

[29] Maeda Y, Kettner N, Holden J, Lee J, Kim J, Cina S, Malatesta C, Gerber [44] Rosas HD, Liu AK, Hersch S, Glessner M, Ferrante RJ, Salat DH, van der J, McManus C, Im J, Libby A, Mezzacappa P, Morse LR, Park K, Audette Kouwe A, Jenkins BG, Dale AM, Fischl B. Regional and progressive J, Tommerdahl M, Napadow V. Functional deficits in carpal tunnel thinning of the cortical ribbon in Huntington’s disease. Neurology 2002; syndrome reflect reorganization of primary somatosensory cortex. Brain 58:695–701. 2014;137:1741–52. [45] Salat DH, Buckner RL, Snyder AZ, Greve DN, Desikan RS, Busa E, Morris [30] Maeda Y, Kettner N, Sheehan J, Kim J, Cina S, Malatesta C, Gerber J, JC, Dale AM, Fischl B. Thinning of the cerebral cortex in aging. Cereb McManus C, Mezzacappa P, Morse LR, Audette J, Napadow V. Altered Cortex 2004;14:721–30. brain morphometry in carpal tunnel syndrome is associated with median [46] Segonne F, Dale AM, Busa E, Glessner M, Salat D, Hahn HK, Fischl B. A nerve pathology. Neuroimage Clin 2013;2:313–319. hybrid approach to the skull stripping problem in MRI. Neuroimage 2004; [31] May A. Chronic pain may change the structure of the brain. PAIN 2008; 22:1060–75. 137:7–15. [47] Segonne F, Pacheco J, Fischl B. Geometrically accurate topology- [32] Metz AE, Yau HJ, Centeno MV, Apkarian AV, Martina M. Morphological correction of cortical surfaces using nonseparating loops. IEEE Trans and functional reorganization of rat medial in Med Imaging 2007;26:518–29. neuropathic pain. Proc Natl Acad Sci U S A 2009;106:2423–8. [48] Seminowicz DA, Wideman TH, Naso L, Hatami-Khoroushahi Z, Fallatah [33] Moayedi M, Weissman-Fogel I, Salomons TV, Crawley AP, Goldberg MB, S, Ware MA, Jarzem P, Bushnell MC, Shir Y, Ouellet JA, Stone LS. Freeman BV, Tenenbaum HC, Davis KD. Abnormal gray matter aging in Effective treatment of chronic low back pain in humans reverses abnormal chronic pain patients. Brain Res 2012;1456:82–93. brain anatomy and function. J Neurosci 2011;31:7540–50. [34] Mogyoros I, Bostock H, Burke D. Mechanisms of paresthesias arising [49] Sled JG, Zijdenbos AP, Evans AC. A nonparametric method for automatic from healthy axons. Muscle Nerve 2000;23:310–20. correction of intensity nonuniformity in MRI data. IEEE Trans Med Imaging [35] Mondelli M, Reale F, Sicurelli F, Padua L. Relationship between the self- 1998;17:87–97. administered Boston questionnaire and electrophysiological findings in [50] Smallwood RF, Laird AR, Ramage AE, Parkinson AL, Lewis J, Clauw DJ, follow-up of surgically-treated carpal tunnel syndrome. J Hand Surg Br Williams DA, Schmidt-Wilcke T, Farrell MJ, Eickhoff SB, Robin DA. 2000;25:128–34. Structural brain anomalies and chronic pain: a quantitative meta-analysis [36] Moore CI, Stern CE, Corkin S, Fischl B, Gray AC, Rosen BR, Dale AM. of gray matter volume. J Pain 2013;14:663–75. Segregation of somatosensory activation in the human rolandic cortex [51] Sud V, Tucci MA, Freeland AE, Smith WT, Grinspun K. Absorptive using fMRI. J Neurophysiol 2000;84:558–69. properties of synovium harvested from the carpal tunnel. Microsurgery [37] Napadow V, Kettner N, Ryan A, Kwong KK, Audette J, Hui KK. 2002;22:316–9. Somatosensory cortical plasticity in carpal tunnel syndrome–a cross- [52] Tamburin S, Cacciatori C, Praitano ML, Cazzarolli C, Foscato C, Fiaschi sectional fMRI evaluation. Neuroimage 2006;31:520–30. A, Zanette G. Median nerve small- and large-fiber damage in carpal tunnel [38] Nora DB, Becker J, Ehlers JA, Gomes I. What symptoms are truly caused syndrome: a quantitative sensory testing study. J Pain 2011;12:205–12. by median nerve compression in carpal tunnel syndrome? Clin [53] Taylor KS, Anastakis DJ, Davis KD. Cutting your nerve changes your Neurophysiol 2005;116:275–83. brain. Brain 2009;132:3122–33. [39] Ochoa JL, Torebjork HE. Paraesthesiae from ectopic impulse generation [54] Truini A, Garcia-Larrea L, Cruccu G. Reappraising neuropathic pain in in human sensory nerves. Brain 1980;103:835–53. humans–how symptoms help disclose mechanisms. Nat Rev Neurol [40] Ortiz-Corredor F, Calambas N, Mendoza-Pulido C, Galeano J, Diaz-Ruiz J, 2013;9:572–82. Delgado O. Factor analysis of carpal tunnel syndrome questionnaire in [55] Truini A, Padua L, Biasiotta A, Caliandro P, Pazzaglia C, Galeotti F, relation to nerve conduction studies. Clin Neurophysiol 2011;122:2067–70. Inghilleri M, Cruccu G. Differential involvement of A-delta and A-beta [41] Padua L, Padua R, Aprile I, Pasqualetti P, Tonali P. Multiperspective fibres in neuropathic pain related to carpal tunnel syndrome. PAIN 2009; follow-up of untreated carpal tunnel syndrome: a multicenter study. 145:105–9. Neurology 2001;56:1459–66. [56] You H, Simmons Z, Freivalds A, Kothari MJ, Naidu SH. Relationships [42] Papanicolaou GD, McCabe SJ, Firrell J. The prevalence and between clinical symptom severity scales and nerve conduction characteristics of nerve compression symptoms in the general measures in carpal tunnel syndrome. Muscle Nerve 1999;22: population. J Hand Surg Am 2001;26:460–6. 497–501. [43] Pardoe HR, Abbott DF, Jackson GD, Alzheimer’s Disease Neuroimaging, [57] Zatorre RJ, Fields RD, Johansen-Berg H. Plasticity in gray and white: Initiative. Sample size estimates for well-powered cross-sectional cortical neuroimaging changes in brain structure during learning. Nat Neurosci thickness studies. Hum Brain Mapp 2013;34:3000–9. 2012;15:528–36.

Copyright Ó 2016 by the International Association for the Study of Pain. Unauthorized reproduction of this article is prohibited.