Surface-based Map Plasticity of the Sensorimotor Cortex in Hip Disorder at Local and Extensive Levels: A Resting-state fMRI Study

Background: Pain is one of the manifestations of hip disorder and has been proven to lead to the remodeling of somatotopic map plasticity in the cortex. However, it’s not clear whether hip disorder with pain induces somatotopic map plasticity in the cortex. We aimed to evaluate the surface-based map plasticity of the somatotopic cortex in hip disorder at local and extensive levels by resting-state functional magnetic resonance imaging (rs-fMRI). Methods: 20 patients with osteonecrosis of the femoral head (ONFH) (12 males and 8 females, age= 56.80±13.60 years) with Visual Analogue Scale (VAS) scores ≥ 4 and 20 healthy controls (9 males and 11 females, age= 54.56±10.23 years) were enrolled in this study. rs-fMRI data and T1 imaging data were collected, and surface-based regional homogeneity (ReHo), seed-based functional connectivity (FC), cortical thickness and the volume of subcortical gray nuclei were calculated. Results: Compared with the healthy controls, the ONFH patients showed signi ﬁ cantly increased surface-based ReHo in areas distributed mainly in the left dorsolateral prefrontal cortex and frontal eye eld, the right frontal eye eld and the premotor cortex and decreased surface-based ReHo in the right primary motor cortex and primary sensory cortex. When the area with decreased surface-based ReHo in the frontal eye eld and right premotor cortex was used as the regions of interest (ROI), compared with the controls, the ONFH patients displayed increased FC in the right middle frontal cortex and right inferior parietal cortex and decreased FC in the right precentral cortex and right middle occipital cortex. ONFH patients also showed signi ﬁ cantly decreased cortical thickness in the para-insular area, supplementary motor cortex area and frontal eye eld and decreased volume of subcortical gray matter nuclei in the right nucleus accumbens (479.32±88.26 vs 539.44±68.36, P=0.026). Conclusions: Hip disorder patients showed cortical plasticity changes, mainly in sensorimotor and pain-related

collected, and surface-based regional homogeneity (ReHo), seed-based functional connectivity (FC), cortical thickness and the volume of subcortical gray nuclei were calculated.
Results: Compared with the healthy controls, the ONFH patients showed significantly increased surfacebased ReHo in areas distributed mainly in the left dorsolateral prefrontal cortex and frontal eye eld, the right frontal eye eld and the premotor cortex and decreased surface-based ReHo in the right primary motor cortex and primary sensory cortex. When the area with decreased surface-based ReHo in the frontal eye eld and right premotor cortex was used as the regions of interest (ROI), compared with the controls, the ONFH patients displayed increased FC in the right middle frontal cortex and right inferior parietal cortex and decreased FC in the right precentral cortex and right middle occipital cortex. ONFH patients also showed significantly decreased cortical thickness in the para-insular area, supplementary motor cortex area and frontal eye eld and decreased volume of subcortical gray matter nuclei in the right nucleus accumbens (479.32±88.26 vs 539.44±68.36, P=0.026).
Conclusions: Hip disorder patients showed cortical plasticity changes, mainly in sensorimotor and painrelated regions.

Background
Osteonecrosis of the femoral head (ONFH), also known as avascular or ischemic necrosis of femoral head, is a hip disorder and challenging orthopedic disease that severely diminishes patient quality of life.
The risk factors of ONFH are varied and include physical trauma, smoking, excessive alcohol consumption, autoimmune diseases, coagulation disorders, hemoglobinopathies, corticosteroid therapy and other factors 1 . ONFH is a progressive pathological process characterized by ischemia, necrosis, and eventually collapse. Despite no predominant pattern of a clinical presentation, pain and dysfunction of the joint seem occur frequently in ONFH 2 . Pain presents in only half of patients with no limitation of activity in the early stages but almost always appears in later stages 2 . The disease generally progresses to collapse and eventually painful osteoarthritis if symptomatic ONFH has not been treated 3 . The treatment strategy of ONFH depends on the disease stage. Reducing pain is one of the main goals of early treatment 4 . Total hip arthroplasty (THA) in patients with osteonecrosis is widely used to improve patient quality of life. However, revision surgeries after THA in young patients are common, and thigh pain is a common complaint after THA 5 . This pain is a primary factor in uencing patients' dissatisfaction after THA 6 .
As observed in ONFH, joint diseases cause pain, and the impaired joints and surrounding muscles, ligaments and tendons can cause proprioception abnormalities. Such abnormal sensory perception leads to the remodeling of the sensory cortex. The mature human primary somatosensory cortex can reorganize itself in response to changes in sensory input, displaying strong plasticity capacity. Following the elimination of afferent return, there is a well-known 'invasion' of the deafferented region of the brain by the cortical representation zones of still-intact portions of the brain adjacent to it 7 . Human and animal studies have con rmed that sensory deprivation can induce somatotopic map plasticity in the primary somatosensory cortex 7,8 . Prolonged and chronic pain can both alter neural plasticity at the cortical level 9 .
It has gradually been recognized that cerebral processes contribute to pain beyond the level of nociceptive input and contact behavioral and psychological in uences 10 . Pain is a complex sensory and emotional experience that is shaped by psychobiology, expectations from past and learned pain experiences and attention processes 11 . In a study of intestinal diseases related to abdominal pain, researchers found altered neural modulation of pain in affective and cognitive brain regions, including the cingulate cortex 12 . Brain regions of headache patients showed signi cant activations in areas responsible for the processing of cognitive empathy 13 . Current theories describe the brain activity of pain as abnormal functioning in large-scale networks that include non-nociceptive regions 14,15 . Chronic cervical spondylotic pain, chronic back pain and knee osteoarthritis show functional anomaly during the pain process 16,17 . Subjects with cruciate ligament de ciency also exhibit increased activation in the secondary somatosensory area (where pain and sensory processing occur) 18 .
The remodeling of somatosensory cortex evidences the plasticity of the somatotopic map 19 . Recently, somatosensory cortex stimulation has been proposed as a possible treatment for deafferentation pain after amputation 20 . Exploring the map plasticity in patients with ONFH can provide insight into its possible treatment. Therefore, based on previous studies that have discovered abnormal brain activity in many kinds of diseases related to sensory abnormality, we hypothesized that ONFH with pain may induce somatotopic map plasticity in cortex and yield a characteristic pattern of brain neural activity. Volumebased normalization may introduce inaccuracies in anatomical positioning of functional data 21,22 , and it is di cult to account for inter-subject variability in gyrus size, shape or position in a 3D referential; furthermore, such differences may correspond to the displacement of a functional focus to a different gyrus in some subjects 22 . As a large part of the data originates from the cortex, methods that work on the cortical surface may be more sensitive than those using the full brain volume and thus be more suitable for map plasticity study.
Diverse approaches can be used to detect map plasticity, including genetics analysis, synaptic and in vivo physiology analysis, optical imaging, and ultrastructural analysis 19 . Although these methods show excellent validity and reliability, they are not simple or convenient to use in the human body. An article published in the New England Journal of Medicine reported increased activity in the stable patterns across regions that are associated with experimentally induced acute pain and tonic pain in healthy people. They proposed that a functional magnetic resonance imaging (fMRI) could be used to assess pain 23 . Resting-state functional magnetic resonance imaging rs-fMRI (rs-fMRI), as a convenient and noninvasive method, has been widely used in the evaluation of central remodeling in various diseases related to pain. Therefore, in our study, we chose fMRI as the assessment tool to test our hypothesis. To comprehensively understand the characteristics of somatotopic map plasticity in patients with ONFH, an understanding of the roles of pain-related brain regions and functional connections is needed. Therefore, among several functional measures of fMRI, surface-based regional homogeneity (ReHo) 24 and seedbased functional connectivity (FC) 25 were chosen to analyze localized and remote changes in brain functions. Compared with 3D ReHo, surface-based ReHo has higher test-retest reliability and can more clearly reveal the intrinsic functional organization of cortex 26 , which could be extremely useful for integrating various measures of both the structure and function of the cortical surface 27 . Cortical thickness and the volume of subcortical gray nuclei were also analyzed in the study.

Clinical Samples
A total of 20 right-handed ONFH inpatients without surgery (12 males and 8 females, age= 56.80±13.60 years) who were consecutively recruited and 20 right-handed healthy controls (9 males and 11 females, age= 54.56±10.23 years) were enrolled in this study. Diagnosis of ONFH was made based on magnetic resonance imaging (MRI) and clinical demonstrations. Patients were included if they had joint pains in groin, buttock and thigh areas and Visual Analogue Scale (VAS) scores for pain 28 ≥ 4 over 1 year. Individuals were excluded from the study if they had a history of cardiovascular or cerebrovascular disease, ankylosing spondylitis, rheumatoid arthritis, hip dysplasia, metabolic disorders, or bone tumor.
Informed consent was obtained from all participants included in the study. There was no statistical difference in background data between ONFH patients and healthy controls, including education, past history and operation history.

MRI Data Acquisition
MRI data were acquired using a MAGNETOM Verio 3.0 T scanner (Siemens Healthcare, Germany).

Data processing
Functional and structural images of each subject were preprocessed by using DPABI 29  Statistical analysis Two-sample t-tests were performed by using SPSS to evaluate differences in the volume of subcortical gray nuclei between ONFH patients and healthy controls. The level of two-tailed statistical signi cance was set at P < 0.05. Two-sample t-tests were used for analyses of surface-based ReHo and cortical thickness differences between the ONFH patients and healthy controls. The results were corrected for multiple comparisons, with a combined threshold of a single voxel (P < 0.05). A more strict threshold (P < 0.01, cluster size> 50 voxels) was also used for each cohort to reduce the possibility of false negative results. Two-sample t-tests were used to analyses FC differences between the ONFH patients and healthy controls. We employed a cluster-level family-wise error (FWE) correction for multiple comparisons (P 0.05).

Sample size calculation
This is a cross-sectional study. The essence of fMRI image is digit, fMRI data is measurement data. We planned a study of a continuous variable from independent control subjects (healthy controls) and experimental subjects (ONFH). Considering a true difference in the experimental and control means of 0.07 in ReHo, we needed to study at least 18 experimental subjects and 18 control subjects to be able to reject the null hypothesis that the population means of the experimental and control were equal with probability (power) 0.80. The type I error probability associated with this test of this null hypothesis was 0.05 (α).

Results
Comparison of functional images between ONFH patients and healthy controls Compared to healthy controls, ONFH patients showed significantly increased surface-based ReHo in areas distributed mainly in the left dorsolateral prefrontal cortex and frontal eye eld, the right frontal eye eld and premotor cortex and decreased surface-based ReHo in the right primary motor cortex and primary sensory cortex ( Figure 1 and Table 1).
Regions showing signi cant differences in surface-based ReHo values between the two groups were de ned as ROIs for seed-based FC analysis. As shown in Figure 2, when the area with decreased surfacebased ReHo in the right frontal eye eld and premotor cortex was used as ROI, the ONFH patients displayed increased FC in the right middle frontal cortex and right inferior parietal cortex and decreased FC in the right precentral cortex and right middle occipital cortex ( Figure 2 and Table 2). FC values of another two ROIs did not signi cantly differ between the two groups.

Comparison of structural images between ONFH patients and healthy controls
Compared to healthy controls, ONFH patients showed significantly decreased cortical thickness in areas mainly distributed in the para-insular area, posterior insular area, anterior superior temporal area, frontal eye eld and supplementary motor cortex. In addition, comparison of the volume of subcortical gray matter nuclei between the two groups revealed significantly decreased values in the right nucleus accumbens in ONFH patients compared with healthy controls (479.32±88.26 vs 539.44±68.36, P=0.026; Figure 3 and Table 3).

Discussion
In this study, we revealed differences in functional and structural imaging between ONFH patients and healthy controls that may reveal the characteristics of altered central pain processing in ONFH patients. The brain regions exhibiting differences between the two groups included the dorsolateral prefrontal cortex, frontal eye eld, premotor and supplementary motor cortex, primary motor cortex, primary sensory cortex, middle frontal cortex, inferior parietal cortex, precentral cortex, middle occipital cortex, insula and nucleus accumbens. These regions are mainly sensorimotor and pain-related regions. Our functional and structural analyses both support previous neuroimaging ndings regarding pain and provide novel ndings that may provide the foundation for future larger studies in ONFH.
Pain is one symptom of ONFH and is usually con ned to the groin area, occasionally involving the ipsilateral hip and knee or greater trochanteric area 32 . There are two complementary pathways related to pain processing: the medial and lateral pain pathways 33 . In the present study, ONFH patients showed functional decline in the primary sensory cortex, which is compatible with the existence of complementary pathway related to pain processing and align with the results of brain function studies in other non-central nervous system diseases with pain 34,35 . Interestingly, both of these pathways involve the insular cortex. In the pain matrix, the insular cortex is mainly involved in discriminative sensory and motivative emotion. Abnormal signal transmission from the injury site causes neuropathic pain, which generates enhanced synaptic plasticity 36 . Our study showed decreased cortical thickness of the insular cortex, which is consistent with the ndings of other pain studies in chronic migraine, cervical spondylosis with neck pain and irritable bowel syndrome 37-39 . Usui C et al 40 observed that patients with bromyalgia showed a signi cant difference in connectivity between the insular cortex and other brain regions. After analgesic treatment, the insular cortex is activated in low back pain 41 and bromyalgia 40 .
We also observed that the volume of the nucleus accumbens was signi cantly decreased in the ONFH patients compared with the healthy controls. Makary MM et al 42 provided evidence that a lower nucleus accumbens volume confers risk for developing chronic pain and that altered nucleus accumbens activity is a signature of the state of chronic pain.
Numerous studies have demonstrated that pain processing can shift from nociceptive somatosensory pathways to emotional brain circuits 43,44 . Increased activation of the prefrontal cortex (PFC) is related to decreased pain and inhibits the functional connectivity between the midbrain and the medial thalamus 45 . Further projections from the anterior cingulate cortex to the PFC may also be involved in cognitive appraisal of the stimulus. In addition, prefrontal responses to pain depend on the psychological state of the subject, who may expect worse pain or a reduction in pain 46 . Neuroimaging studies indicate that upregulation and downregulation of negative emotions are associated with increased activation of prefrontal regions 47 . These observations demonstrate that the PFC plays an important role in pain processing 48 . We found that ONFH patients showed signi cantly increased ReHo in the dorsolateral prefrontal cortex. Notably, the areas with signi cantly increased ReHo are in the central part of the PFC.
The dorsolateral prefrontal cortex is functionally linked to the descending pain modulation system and has been implicated in top-down pain inhibition, including placebo analgesia 49 . In another orthopedic disease, poor recovery of upper limb pain after surgical interventions for cervical spondylotic myelopathy was found to be associated with the dorsolateral prefrontal cortex 50 . Similarly, during pain onset, a higher BOLD signal response in the dorsolateral prefrontal cortices was observed in bromyalgia patients than in control subjects 51 . Therefore, brain hyperactivation may be a mechanism underlying the generalized hypervigilance to salient stimuli in pain 51 .
The sensorimotor cortex is another brain region closely related to the sensory cortex. The sensorimotor cortex includes somatosensory and motor regions and extends to the supplementary motor area (SMA) 52 . The FC analysis revealed differences in sensorimotor regions between ONFH patients and controls.
These ndings are consistent with a previous pain study that showed abnormal functional connectivity of sensorimotor cortex in primary dysmenorrhea patients 53 . Furthermore, some studies have con rmed a relationship between pain persistence and aberrant sensorimotor cortex activity 54 . Interestingly, in the present study, the ONFH patients showed decreased cortical thickness in the supplementary motor cortex. Consistent with this result, a study of diabetic peripheral neuropathy (DPN) with focus on painful DPN revealed impaired motor gray matter in DPN patients 55 . In addition, the duration and frequency of migraine attacks have been found to have strong effects on cortical thickness in the sensorimotor cortex 56 . The increased surface-based ReHo in the premotor cortex observed in the ONFH patients in this study might represent a compensatory increase. In our future work, we aim to explore this topic with a larger sample size. The result that ONFH patients showed signi cantly increased frontal eye eld may indicate the abnormality of visual movement and visual attention network.

Conclusions
In conclusion, patients with ONFH appear to exhibit cortical and subcortical thinning and abnormal functional activity in speci c brain regions associated with sensorimotor and pain processing. This is a pioneer study of the brain mechanisms in ONFH patients, as revealed by comparisons with healthy controls. There are several limitations to note. First, the sample size was small, limiting the statistical power to detect differences. Second, the different stages and types of ONFH were not considered.  Figure 1 Comparison of surface-based ReHo values between ONFH patients and healthy controls. Compared to healthy controls, ONFH patients showed significantly increased surface-based ReHo in areas distributed mainly in the left dorsolateral prefrontal cortex and frontal eye eld, the right frontal eye eld and premotor cortex and decreased surface-based ReHo in the right primary motor cortex and primary sensory cortex.

Figure 2
Comparison of FC values between ONFH patients and healthy controls When the area with decreased surface-based ReHo in the frontal eye eld and right premotor cortex was used as ROI, the ONFH patients displayed increased FC in the right middle frontal cortex and right inferior parietal cortex and decreased FC in the right precentral cortex and right middle occipital cortex. Frontal_Mid_R: right middle frontal gyrus; Parietal_Inf_R: right inferior parietal gyrus; Precentral_R: right precentral gyrus; Occipital_Mid_R: right middle occipital gyrus Figure 3 Comparison of structural images between ONFH patients and healthy controls. a: Comparison of cortical thickness; b: Comparison of volume of subcortical gray matter nuclei. ONFH patients showed significantly decreased cortical thickness in areas mainly distributed in the para-insular area, posterior insular area, anterior superior temporal area, frontal eye eld and supplementary motor cortex. In addition, comparison of the volume of subcortical gray matter nuclei between the two groups revealed significantly decreased values in the right nucleus accumbens in ONFH patients compared with healthy controls. NAc: nucleus accumbens