Using Resting-State Functional Connectivity Within the Basal Ganglia as a Biomarker for Symptoms of Parkinson’s Disease

Neal Modi*, Tianlin Yuan*, Dr. Todd Parrish**, Dr. Xue Wang**, Dr. Darren Gitelman** * Illinois Mathematics and Science Academy ** Northwestern University, Feinberg School of Medicine, Department of Radiology Abstract Parkinson’s disease (PD) is a chronic movement disorder that occurs when nigrostriatal dopaminergic neurons within the Basal Ganglia Network (BGN) are damaged or destroyed, causing motor and cognitive disabilities. Motor disabilities can be measured by the Unified Parkinson’s Disease Rating Scale (UPDRS-III). PD currently has no cure, so the discovery of a biomarker is vital for successful clinical trials and accurate early diagnosis. This study looks at the potential of using functional connectivity in the BGN as a biomarker for symptoms of PD. Data from 73 subjects were downloaded from the Parkinson’s Progression Markers Initiative (PPMI) and analyzed for functional connectivity values using a seed based method analysis. The values were correlated with UPDRS-III scores, taking into consideration the effects of age, site, education, and gender. There was a significant correlation between functional connectivity between the left and right caudate and the UPDRS-III sub-score for rigidity. Patients with reduced connectivity received higher rigidity scores, demonstrating a more severe case of PD. Thus, results from this study indicate potential for functional connectivity within the basal ganglia to serve as a biomarker for the progression of specific symptoms of PD. Introduction Parkinson’s disease (PD) is a chronic movement condition affecting more than six million people worldwide, making it the world’s second most common neurodegenerative disease (Obeso et al., 2010).  The disease targets nigrostriatal dopaminergic neurons in the brainstem and extends upward to involve cortical and subcortical structures, such as the thalamus and basal ganglia, as the disease progresses. PD’s effect on movement is thought to arise from dopaminergic neuron damage in the substantia nigra and basal ganglia, which includes the caudate and putamen nucleus (Calne et al., 1985). When nigrostriatal dopaminergic neurons within the basal ganglia are damaged or are destroyed, the brain’s supply of dopamine becomes depleted, which induces motor impairments such as tremors, muscular rigidity, lessened balance and coordination, and bradykinesia (Alam & Schmidt, 2002). However, PD patients also demonstrate decreased cognitive aptitude than healthy subjects, which shows that basal ganglia dysfunction not only impacts motor function, but also mental ability (Cools et al., 1984). As the severity of PD progresses, the disease may induce in patients cognitive disorders such as depression, dementia, and severe anxiety (Jin et al., 2014). Motor symptoms are measured by the third part of the Unified Parkinson’s Disease Rating Scale (UPDRS-III) and cognitive symptoms are measured by the Montreal Cognitive Assessment (MoCA). However, by the time symptoms of PD are detectable by these tests, critical nigrostriatal dopaminergic neurons will have already been damaged (Jin et al., 2014). Existing treatments involve the use of levodopa, which is converted into dopamine, and temporarily alleviate the severity of some motor symptoms (Otsuka et al., 1996). Long term use of levodopa, however, can lead to the development of motor impairment such as dystonia and dyskinesia as well as mental disorders of hallucinations and illusions (Michael J. Fox Foundation for Parkinson’s Research). There is currently no way to repair neurons once they have been destroyed, thus making the discovery of a biomarker crucial. Biomarkers allow an early and accurate diagnosis as well as the objective measuring of disease progression as a response to treatments, which would aid greatly in the process to find a cure (Michael J. Fox Foundation for Parkinson’s Research). Studies verify that the amount of brain activity in the basal ganglia has a direct impact on the progression of PD (Fox & Raichle, 2007). Functional connectivity, the amount of activity between brain regions, can be measured through functional magnetic resonance imaging (fMRI), which measures the change in blood oxygen levels as it is used by neurons through the blood oxygen level dependent (BOLD) signal (Fox & Raichle, 2007). Because PD patients often suffer from tremors, it is difficult to get an accurate scan using normal fMRI. However, an investigation found that even when a patient is at resting state (i.e. lying completely still), the brain continues to send signals between brain regions, making resting state fMRI (rs-fMRI) an accurate way to measure functional connectivity between brain regions (Krolikowski et al., 2014). Because the space between cortico-striatal loops fluctuates due to dopamine depletion, a change in the amount of cerebral connectivity occurs (Helmich et al., 2009). Thus, a lack of dopamine directly influences brain activity, allowing the amount of connectivity within the basal ganglia to differentiate PD patients from healthy subjects (Krolikowski et al., 2014). However, the validity of using functional connectivity as a biomarker for specific symptoms of PD has not yet been investigated. In this experiment, we analyzed the effect of functional connectivity within regions of the basal ganglia on definite attributes and symptoms of Parkinson’s Disease. The null hypothesis is that an increase or decrease in functional connectivity between regions of the basal ganglia would have no effect on the manifestation of PD symptoms. Alternatively, we hypothesized that a decreased amount of functional connectivity between certain regions of the basal ganglia is likely to indicate the onset and severity of different symptoms of PD. This raises the question: can resting state functional connectivity in the basal ganglia region be used as a biomarker for specific symptoms of PD? Materials and Methods Subjects Data from 70 subjects with PD (49 male, aged > 38; 21 female, aged >38) and six healthy control subjects were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database. The data for each patient included T1 weighted structural MRI scans and functional MRI scans obtained while patients were at resting state in the scanner. 54 subjects were using anti-Parkinson medication at the time of their scan. The severity of each patient’s PD was assessed using the third part of the Unified Parkinson’s Disease Rating Scale (UPDRS-III). Average disease severity (UPDRS-III) was 21.5 points (maximum score is 75). Because subjects had previously consented to participation in the PPMI database and all data downloaded was de-identified, the study is considered exempt human subjects research and does not require Institutional Review Board approval. Imaging Data Preprocessing All data were analyzed and preprocessed through the Northwestern University Neuroimaging Data Archive (NUNDA). First, images were run through a Quality Assurance Analysis, which analyzed the data for the temporal Signal to Noise Ratio (tSNR), the Frame-wise Displacement (FD), and the derivative of the time course in scanner (DVARS) value. Data with tSNR values less than 120 were not included in the study. tSNR ratios of 120 or lower signify that the amount of noise in the images is greater than the amount of signal obtained from the image, thus making the data un-useable for the study. The data processing was done through a pipeline (algorithm) built by Xiaowei Song, a Northwestern researcher, and verified by Dr. Todd Parrish in the Radiology Department of Northwestern University. After slice timing correction and realignment, functional images were coregistered to the structural images and transformed into MNI standard space (Montreal Neurological Institute). The images were then smoothed using a 6 mm FWHM Gaussian kernel detrended, and bandpass filtered from 0.01 to 0.08 Hz. Images with greater than 0.5 mm of frame-wise displacement (FD) or larger than 50 volume to volume variation (DVARS) were discarded. Nuisance variables were regressed out of the time course, including 24 motion parameters, white matter signal, and cerebrospinal fluid signal. The pipeline consists of a series of three analyses. In the anatomical analysis, only animage of the head should remain in the image and no other artifacts or parts of the body. The skull strip analysis “cuts” out the brain tissue from the other tissues found in the head. The brain tissue was carefully outlined it so that when running other analyses, only data from the brain would be under consideration. Lastly, normalization/overlap was used to warp the patients’ brain images into a template brain, or an “average” human brain, created by averaging images from over 100 people with the same conditions as the patients used in this study. The results of the analyses in the pipeline can be seen in Figure 1. Analyses From each subject’s normalized anatomical rs-fMRI data, the left and right caudate and putamen were identified using the automated anatomical labeling (AAL) brain template and defined as seed regions of interest (ROI). The locations of these ROIs can be seen in Figure 2. Using seed-based voxel-wise analysis, an average time course was extracted for each seed ROI and correlated with the time courses of other ROIs to determine a correlation value, then Fisher Z transformed into Z values. Z values were correlated with UPDRS-III sub-scores taking into account the effect of the age, sex, on/off PD medication, MoCA, and site of acquisition using statistical functions in MATLAB®. Results Functional connectivity between the left and right caudate negatively correlated with rigidity sub-scores in the UPDRS-III (Figure 3), indicating that decreases in correlation between left and right caudates directly correlates with higher severity in rigidity. This trend was not observed in healthy control subjects. The linear regression analysis test yielded a statistical value of P = 6.76 E-6 (P < 0.05) and a correlation coefficient of R = 0.3630. These values were obtained while taking into consideration the effect of patient age and sex along with the medications those patients had during the image acquisition process, the site of image acquisition, and the Montreal Cognitive Assessment (MoCA) scores on the population studied. Functional connectivity maps from two subjects with different severity of PD are shown in Figure 4. The subject with more severe symptoms of rigidity (UPDRS-III subscore = 6) (A) showed lowered connectivity compared to the subject with a lesser degree of rigidity (UPDRS-III subscore = 4) (B). No correlation was observed between functional connectivity within the left and right putamen and disease severity from the UPDRS-III sub-scores (all P > 0.05). Figure 1. Shows a good (left) and a bad (right) example for each of the algorithms. The anatomical algorithm’s function is to show the anatomical structure of the patient’s brain from the MRI scanner. The Skull Strip algorithm outlined only the brain tissues, and the overlap algorithm fitted the patient’s brain (in red lines) onto a template brain created by an average of 150 brain scans. Figure 2. Highlights the main regions in the basal ganglia used as Regions of Interest (ROIs) in this study. The image is an fMRI image with regions labeled in color. The colors have no meaning to the fMRI results. They are for easy identification of ROIs. Figure 3. Shows the significant correlation between UPDRS-III rigidity score and level of connectivity between the left and right caudate regions of the brain.  There is a negative trend for this correlation. As the UPDRS-III rigidity score decreases, the connectivity between left caudate (CL) and right caudate (CR) increases. Figure 4. Brain maps displaying the amount of functional connectivity in different PD subjects. Warmer colors (red and orange) represent a larger amount of connectivity, while cooler colors (green and blue) represent lowered functional connectivity.  The subject with lowered connectivity (A) experienced more severe symptoms of rigidity (UPDRS-III subscore = 6) while the subject with a larger amount of functional connectivity (B) experienced a lesser degree of rigidity (UPDRS-III subscore = 4). Discussion In this study, we aimed to verify the validity of using functional connectivity as a biomarker for symptoms of PD. Results of our study show that there is a significant positive correlation between more severe symptoms of rigidity and a loss of functional connectivity within the caudate regions of the basal ganglia network in PD patients. This correlation suggests the possibility of using functional connectivity as a biomarker for certain symptoms of PD. A decrease in functional connectivity in patients with more severe symptoms of PD may be caused by a remapping of cerebral connectivity as a result of dopamine depletion. A recent study concluded that PD patients have altered connections between regions of the brain in the same pattern that dopamine is depleted as a result of PD (Helmich et al., 2009). As dopaminergic nigrostriatal neurons in the basal ganglia are destroyed, dopamine levels will decrease, which directly causes a change in brain connectivity. Brain activity levels have a direct impact on the severity of PD symptoms but the effect that activity has on specific symptoms of PD remains unknown (Krolikowski et al., 2014; Gottlich et al., 2013). Our results indicate that decreased connectivity in explicit regions may lead to the direct manifestation of certain PD symptoms. Dopamine supply influences functional connectivity amounts. Additional fluorodopa supplements in the caudate and putamen leads to a decrease in symptoms of rigidity and bradykinesia (Otsuka et al., 1996) but prolonged use of such medication results in further motor diseases and even hallucinations and illusions (Michael J. Fox Foundation for Parkinson’s Research). Patients with akinetic rigid PD symptoms are at a greater risk of developing cognitive impairment (Bunzeck et al., 2013), which also suggests that dopamine depletion as a result of PD not only affects motor function, but cognitive function as well. The caudate is typically considered a cognitive nucleus, so its correlation with the motor function of rigidity suggests a possible relationship between cognitive regions of the brain and various symptoms of PD, although further investigation is required to confirm this claim. Thus, in order to determine the validity of our conclusions, future novel studies will be conducted focusing on the change in cognitive performance symptoms in addition to motor symptoms in PD patients. Functional connectivity in brain regions will be correlated with subcomponents of MoCA to assess the effect of dopaminergic nigrostriatal neuron destruction on cognitive function. Additional regions of the brain associated with PD will be analyzed to find the correlations between functional connectivity and various disease symptoms. Furthermore, further study will require connectivity data from a large number of control subjects to determine a statistically significant difference between healthy and diseased individuals. This would confirm the suggestion of using resting-state functional connectivity in the basal ganglia as a biomarker for PD. Although the cause of neurodegeneration in PD remains unclear, experimental results imply that specific symptoms of the disease are directly correlated with a decrease in functional connectivity in regions of the brain. Further experimentation with MoCA and other brain regions will result in a greater understanding on the effect of the disease on cognitive and motor symptoms of PD. Our results indicate that the functional connectivity between brain regions directly correlates with severity in specific PD symptoms. Therefore, functional connectivity has potential to be a biomarker for individual symptoms of PD, which will allow us develop better methods to treat people afflicted with this neurodegenerative disease. Conclusion We have concluded that there is a high correlation between the connectivity of the left caudate (ROI) and the right caudate (another ROI) and the UPDRS-III rigidity scores. This makes sense because there is a negative trend in our graph in figure 3 12. This negative trend is seen because as the connectivity in the ROIs increase, the UPDRS-III scores decrease, meaning there is a less severe form of PD or no PD present. If the connectivity in the ROIs decrease, the UPDRS-III scores increase, meaning there is a more severe form of PD present in those patients in our study. Acknowledgements We would first like to thank Dr. Todd Parrish, Dr. Darren Gitelman and Dr. Xue Wang for mentoring us throughout the study. We would also like to thank Dr. Scheppler, Dr. Fischer, and Ms. Magana from the Illinois Mathematics and Science Academy for providing transportation to and from Northwestern University and for allowing us the opportunity of working through the Student Inquiry and Research Program to pursue our investigation.  

Literature Cited

1. Alam, M., & Schmidt, W. J. (2002, October). Rotenone destroys dopaminergic neurons and induces parkinsonian symptoms in rats. Behavioural Brain Research, 136(1), 317-324. doi:10.1016/S0166-4328(02)00180-8.
2. Bunzeck, N., Singh-Curry, V., Eckart, C., Weiskopf, N., Perry, R. J., Bain, P. G., & Duzel, E. (2013, December). Motor phenotype and magnetic resonance measures of basal ganglia iron levels in Parkinson’s disease. Parkinsonism and Related Disorders, 19(12), 1136-1142. doi: 10.1016/j.parkreldis.2013.08.011.
3. Calne, D. B., Langston, J. W., Martin, W. W., Stoessl, A. J., Ruth, T. J., Adam, M. J., & Pate, B. D. (1985, September 19). Positron emission tomography after MPTP: observations relating to the cause of Parkinson’s disease. Nature Publishing Group, 317, 246-248. doi:10.1038/317246a0.
4. Cools, A., Van Den Bercken, J., Horstink, M., Van Spaendonck, K., & Berger, H. (1984). Cognitive and motor shifting aptitude disorder in Parkinson’s disease. Journal of Neurology, Neurosurgery, and Psychiatry, 47, 443-453. doi:10.1136/jnnp.47.5.443.
5. Fox, M. D., & Raichle, M. E. (2007, September). Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nature Reviews Neuroscience, 8(1), 700-711. doi:10.1038/nrn2201.
6. Gottlich, M., Munte, T. F., Heldmann, M., Kasten, M., Hagenah, J., & Kramer, U. M. (2013, October 28). Altered Resting State Brain Networks in Parkinson’s Disease. Public Library of Science ONE, 8(10). doi:10.1371/journal.pone.0077336.
7. Helmich, R. C., Derikx, L. C., Bakker, M., Scheeringa, R., Bloem, B. R., & Toni, I. (2009, August 26). Spatial Remapping of Cortico-striatal Connectivity in Parkinson’s Disease. Cerebral Cortex- Oxford University Press, 1175-1186. doi:10.1095/cercor/bhpl178.
8. Jin, H., Kanthasamy, A., Anantharam, V., & Kanthasamy, A. G. (2014). Chapter 49- Biomarkers of Parkinson’s Disease.Biomarkers in Toxicology, 817-831. doi:10.1016/B978-0-12-404630-6.00049-X.
9. Krolikowski, K. S., Menke, R. A., Rolinski, M., Duff, E., Khorshidi, G. S., Filippini, N., & Zamboni, G. (2014, July 15). Functional connectivity in the basal ganglia network differentiates PD patients from controls. Neurology, 83(3), 208-214. doi:10.1212/WNL.0000000000000592.
10. Michael J. Fox Foundation for Parkinson’s Research. (2014). Parkinson’s Disease Medications. In Understanding Parkinson’s- Living With Parkinson’s. Retrieved from https://www.michaeljfox.org/understanding-parkinsons/living-with-pd/topic.php?medication.
11. Obeso, J. A., Rodriguez-Oroz, M. C., Goetz, C. G., Marin, C., Kordower, J. H., Rodriguez, M., & Hirsch, E. C. (2010, May 23). Missing pieces in the Parkinson’s disease puzzle. Nature Medicine Review, 16, 653-661. doi:10.1038/nm.2165.
12. Otsuka, M., Ichiya, Y., Kuwabara, T., Hosokawa, S., Sasaki, M., Yoshida, T., & Fukumura, T. (1996, April). Differences in the reduced 18F-Dopa uptakes of the caudate and the putamen in Parkinson’s disease: correlations with the three main symptoms. Journal of the Neurological Science, 136(1-2), 169-73. doi: 10.1016/0022-510X(95)00316-T.