To our knowledge, neuroimaging study to explore changes in brain activation associated with memantine anti-dementia therapy in the patients with PD-MCI has never been done. The present study is the first pilot study that aimed to investigate whether memantine can alter brain function concerning visuospatial working memory of the patients with PD-MCI. Using a combination of visuospatial n-back test and fMRI, we revealed that memantine made local brain function within right LG and left SFG worse, comparing with placebo. In contrast, we did not find specific findings enhanced by memantine.
Memantine reduce brain activations associated with working memory in superior frontal gyrus and lingual gyrus
There were no significant regions enhanced by memantine compared with placebo at any load of the n-back tests. In contrast, fMRI findings of the 2-back versus 0-back test at memantine intervention showed reduction of brain activations in right LG and left SFG. While we did not find any specific regions in the results of 0-back, 1-back, and 1-back versus 0-back test. In behavioral analysis, the number of correct answers of 2-back test during memantine intervention was fewer than that of placebo.
Generally, the changes of task-related brain activation were most prominent for the 2-back condition; subtracting the activated regions in the 0-back test (the easiest condition) from those in the 2-back test is adequate for the purpose of searching brain regions associated with working memory. The subtracting condition of the 0-back test from those of the 1-back test had limited statistical power to detect brain areas associated with working memory because the 1-back test requires low-grade visuospatial working memory. Considering these, the deactivation within right LG and left SFG of the patients with PD-MCI represents the impaired visuospatial working memory at memantine intervention.
Concerning the observed regions, it was reported that impaired visual memory was related to either damage to the region or disconnections between the LG and other brain structures . Mangun et al. reported that activation of the LG has been shown in association with selective visual spatial attention in their fMRI study , and Machielsen et al. reported that LG has been linked to encoding of complex images . Therefore, deactivation of the LG in the present study may be related to worse visuo-spatial working memory of the patients with PD-MCI during memantine intervention, compared with placebo.
Second, concerning the role of SFG, Carlson et al. reported that this region is dependent on the memory load of visuospatial n-back task in fMRI . In their study which targeted for normal volunteers, comparison of the 2-back versus 0-back tasks revealed bilateral activation in the medial frontal gyrus, superior frontal sulcus, and SFG. They also reported that comparison between the 1-back versus 0-back tasks showed activation only in a few brain areas. However, unlike their result, there was no significant difference in the activation within medial frontal gyrus, comparing between the 2-back versus 0-back test in our study. We considered that fMRI findings associated with working memory were likely to be influenced by daytime sleepiness of the patients with PD-MCI.
In neuropsychological findings, the correct answer rate of 2-back test at memantine intervention was worse than that of placebo. In addition, although it did not reach statistical significance, there was a trend that the score of ESS at memantine was higher than placebo, suggesting that memantine may worsen daytime sleepiness. Focusing on the association between sleepiness and working memory, several fMRI studies have reported that sleepiness decreased performance of working memory [33, 34]. Interestingly, Li et al. reported that functional connectivity between the superior parietal lobule and pre-motor cortex, supplementary motor cortex, and left dorsolateral prefrontal cortex was observed in the primary insomnia group and the control group using resting state fMRI, and they reported that connectivity in the primary insomnia group was weaker between the superior parietal lobule and right superior frontal gyrus (dorsolateral prefrontal cortex) . Considering these, memantine is likely to worsen sleepiness during visuo-spatial n-back task, and it may result in a reduced activation of the SFG.
Clinical implications for the treatment of PD-MCI
The development of dementia in the patients with PD is caused by dual pathology, that is, Diffuse Lewy body and AD pathology . First, considering the Diffuse Lewy body pathology, therapeutic choice for cognitive impairment in PD is L-DOPA and DA, because dopamine may improve the neuro-transmission of cortico-striatal circuit. Some studies have reported efficacy of L-DOPA or DA for cognitive impairments of the patients with PD [36, 37]. Next, concerning the AD pathology, several studies have reported efficacy of cholinesterase inhibitors for PDD [16, 17], and memantine for PDD . However, neuropathological research for PD-MCI is limited, and therapeutic evidence of PD-MCI has unestablished other than cholinesterase inhibitors.
Focusing on the cognitive influence of memantine for the normal elderly, a double-blind pilot study which targeted for the participants of age-associated memory impairment (no dementia or MCI), reported positive effect, suggesting that memantine has a possibility to improve attentional processes if the memory impairment is in subtle stage . Whereas, in the present study which targeted for PD-MCI, we found a negative influence. A time to complete TMT-A and number of correct answers of the 2-back test were worse at memantine intervention. We considered that those results might be caused by both heterogenous pathology of PD-MCI and increase of daytime sleepiness.
It was well known that dopamine agonists tend to increase daytime sleepiness and the risk of sleep attacks for the patients with PD . In the present study, all patients received maximum dose of memantine on study period, and six of ten patients were treated with both dopamine agonists and L-DOPA. Therefore, it was plausible that combination of dopamine agonists and maximum dose of memantine might enhance daytime sleepiness which resulted in reduction of psychomotor speed and attention. A recent review which studies the efficacy of anti-dementia drug for PDD, reported that cholinesterase inhibitors improved cognitive function but memantine did not . This study focusing on PD-MCI revealed that memantine did not improve visuospatial working memory and brain activations. The results of this study suggested the importance of careful consideration of daytime sleepiness and/or cognitive function for the treatment of PD. Considering these, maximum dose of memantine is not beneficial for the patients with PD-MCI.
Limitations and strengths
This fMRI study has some limitations. First, the main limitation was the small sample size. The statistical power to search differences in behavioral performance and neuroimaging findings was restricted. As for the smoothing kernel in fMRI analysis, although there is no easy answer regarding how much we should smooth imaging data, the main downside to higher smoothing is the loss of spatial specificity. Because of the small sample size, we had to reduce noise as much as we can using relatively high smoothing kernel compared with voxel dimension. Therefore, we used the 10 mm smoothing kernel instead of the reduction of spatial resolution. Because of these reasons, we could not find significant voxels in imaging analysis with the family-wise error correction. Second, the results of n-back test were likely to be influenced by variety of patients’ profile such as the doses of L-DOPA and LEDD of DA. We cannot rule out the possibility that variety of treatment may have had differing impacts on behavioural performance. Third, there is also another possibility that neuroimaging results of n-back tests were affected by negative emotion including anxiety and depression, because negative emotion has a particularly worsen selectivity of attention as well as motivating action and behavior. To reduce the difference of negative emotion in each patient, we performed fMRI scanning during the ‘On’ state, because it was predicted that anxiety arising from being in the ‘Off’ state might influence more compared with ‘On’ state. However, there is no specific way to quantify real influence of emotions during cognitive task, and it may be a fundamental technical limitation of the activation study using fMRI.
However, the present study has some strengths and novel aspects. First, this is the first neuroimaging study which identified the changes of brain activation associated with memantine for the patients with PD-MCI. Second, we stress that fMRI is useful to explore the pharmacological influence for central nervous system noninvasively. To clarify association between another therapy for the patients with PD and therapeutic efficacy, further study is needed to be targeted for larger population.