|Other Abstract||The ability to maintain and manipulate sequential information in a short period of time is crucial in our daily life, namely sequential working memory. As a common neurodegenerative disease in elderly adults, Parkinson's disease (PD) has motor symptoms as well as non-motor symptoms such as cognitive impairment. Previous studies found that PD patients were impaired in information manipulation in sequential working memory, even when information maintenance is well preserved. This study is aimed to investigate the neural mechanism of information manipulation in sequential working memory in PD patients, using neuropsychological tests, neuroimaging, and pharmacological intervention.
Study 1 demonstrated the characteristics of PD patients’ deficits in sequential working memory (Experiment 1a and Experiment 1b). In Experiment 1a, 48 newly diagnosed and untreated PD patients and 50 healthy controls completed the digit span forward and adaptive digit ordering tests. Participants were asked to recall the digits in the original and ascending order in these two neuropsychological tests, respectively. The contrast between the digit span forward and digit ordering tests emphasizes the manipulation of sequences (ordering). PD patients scored lower than healthy controls in the adaptive digit ordering test, but not in the digit span forward test. We classified the error responses into different types. In the adaptive digit ordering test, PD patients produced more transposition errors than healthy controls particularly anticipation errors. It suggests that PD patients are more likely to recall items too early in sequential ordering. In PD patients, the rate of transposition errors positively correlated with the severity of non-motor symptoms in daily life. In Experiment 1b, 47 newly diagnosed and untreated PD patients and 47 healthy controls completed a computerized digit ordering task during functional magnetic resonance imaging (fMRI) scanning. They were asked to memorize the digits in ascending order. In “pure recall” trials, the digits were presented in ascending order and there was no need to reorder them. In “reorder & recall” trials, the digits were randomized and participants had to reorder them. PD patients performed worse than healthy controls with lower accuracy and longer response times. In PD patients, ordering-related accuracy/reaction time cost (reorder & recall vs pure recall) positively correlated with the severity of non-motor symptoms in daily life. Regional activations were greater for “reorder & recall” than “pure recall” trials in the posterior parietal cortex, prefrontal cortex, premotor cortex, subthalamic nucleus, thalamus, caudate nucleus and cerebellum. Regional deactivations were greater for “pure recall” than “reorder & recall” trials in the medial prefrontal cortex and posterior cingulate cortex (default mode network). Compared with “pure recall” trials, the subthalamic nucleus and dorsolateral prefrontal cortex were more activated in “reorder & recall” trials in PD patients than healthy controls. The accuracy of “pure recall” trials could be predicted by the functional connectivity of subthalamic nucleus. However, ordering-related accuracy change (reorder & recall vs pure recall) could not be predicted by the activation or functional connectivity of subthalamic nucleus. It suggests that the dysfunction of subthalamic nucleus may be the neural markers of sequential working memory deficits in mild PD.
Study 2 investigated the effects of dopaminergic drugs on behavior and related brain activities of sequential working memory (Experiment 2a and Experiment 2b). In Experiment 2a, 40 patients with mild PD and 40 healthy controls completed a computerized digit ordering task. The task used in Experiment 2a was combined with the mouse tracking technique to separate the cognitive processes and cognitive aspects of motor programming before the actual movement and the execution processes of the actual movement. The initiation time, ordering time, movement time, and area under the movement trajectory curve were estimated. PD patients showed longer initiation time, longer movement time, and more constrained movement trajectories than healthy controls. The initiation time of “pure recall” trials negatively correlated with the daily exposure to actual levodopa dose. The ordering time negatively correlated with levodopa equivalent dose for dopamine D2/3 receptor agonists. It suggests that stimulating dopamine D1 and D2/3 receptors might speed up the maintenance and manipulation of sequences, respectively. In Experiment 2b, 21 newly diagnosed and untreated PD patients completed the computerized digit ordering task combined with fMRI before and after 4-6 weeks of medical therapy with a selective D2/3 receptor agonist (pramipexole). After taking pramipexole, the severity of motor symptom of PD patients was decreased and task accuracy was improved, in addition, the deactivation in the default mode network was greater in “reorder & recall” trials. There are significant individual differences that patients with more improved motor symptom and more deactivated default mode network (particularly precuneus) showed higher accuracy improvement in “reorder & recall” trials. It suggests that dopamine D2/3 receptor agonist may improve the ordering accuracy by hyper-deactivation of the default mode network. In conclusion, this study found that the manipulation of sequences was impaired even in mild PD patients with lower score and accuracy in digit ordering test/task. The impairment positively correlated with the severity of non-motor symptoms in daily life. At brain level, it correlated with the hyper-activity of subthalamic nucleus and dorsolateral prefrontal cortex. After treating with pramipexole, the patients’ task performances were enhanced. It may due to hyper-deactivation of the default mode network.|