The present study was the first to examine the neural correlates of semantic fluency deficits in PD-MCI. At a behavioral level, patients with PD-MCI performed significantly worse than PD-NC and HC on semantic and semantic switching fluency, whereas PD-NC performed comparably to HC. Our findings therefore support previous literature suggesting that verbal fluency deficits as a defining feature of PD-MCI and thus a potential predictor of PDD (Galtier et al, 2017; Williams-Gray et al, 2007). The results also support suggestions that verbal fluency deficits are associated with an increased risk of dementia more broadly (e.g., Sutin et al (2019)).
PD-MCI patients demonstrated increased activation in the right angular gyrus compared to PD-NC and HC groups during the semantic switching task. Interestingly, no differences in brain activation were revealed during the semantic fluency task across groups. This may be due to the fact that the semantic fluency task is less cognitively demanding than the semantic switching task, which places more substantial demands on executive functions compared to semantic fluency (Baddeley et al, 2001). It is perhaps the taxing combination of executive control and semantic memory that results in the increased activation of the right angular gyrus and decreased task performance in PD-MCI compared to PD-NC and HC. Increased activation was also revealed in the right angular gyrus of PD-MCI patients during semantic switching compared to both the rest and automatic speech conditions, further indicating task-specific hyperactivation in PD-MCI during semantic switching.
When the relationship of the right angular gyrus brain activity to broader fluency performance was considered, there was a significant negative correlation between the activation of the right angular gyrus and verbal fluency tasks including letter (F, A, S), semantic (e.g. animals, boys’ names) and semantic switching (e.g. fruits and furniture). That is, the greater the number of correct responses produced for each task, the less activity in the right angular gyrus was observed. Therefore, it is apparent that the increased right angular gyrus activity is a predictor of increased verbal fluency deficits identified outside the scanner.
The angular gyrus is located at the junction between the occipital, temporal and parietal lobes, and is interconnected bilaterally via the dorsal region of the corpus callosum (Park et al, 2008; Seghier, 2013). It is known to play a significant role in semantic processing, language, complex information integration, memory retrieval, default mode network (DMN) and social cognition (Binder et al, 2009; Seghier, 2013). Semantic processing, in particular, is a function consistently associated with the angular gyrus, with stronger activation typically observed in the left hemisphere (Binder et al, 2009). However, in the present study an increased activation was observed in the right, but not left, angular gyrus with impaired semantic switching in PD-MCI.
While our finding of right as opposed to left angular gyrus activity was unexpected, the involvement of the right angular gyrus in PD-related changes in neural activity is not without precedent. Of note, a recent meta-analysis of whole-brain resting state connectivity in PD found that the most consistent disturbances relative to healthy controls were evident within the posterior inferior parietal lobe (IPL) (Tahmasian et al, 2017). This finding was driven by observations of increased functional connectivity within the bilateral IPL for patients off dopaminergic medication, and decreased connectivity for those on medication. The authors suggested the findings may reflect compensatory mechanisms relating to basal ganglia dysfunction associated with the disease. Similarly, Manes et al (2018) observed increased resting state functional connectivity between the internal globus-pallidus and the angular gyrus bilaterally for PD patients with speech impairments relative to those without speech impairments. The authors attributed these findings to a potential compensatory mechanism, and also suggested that the findings may reflect increased semantic processing deficits in those with speech impairment given that impaired semantic verbal fluency has previously been shown to correlate with impaired speech (Gago et al, 2009). In a similar way, our findings may reflect a compensatory up-regulation of activity within the right angular gyrus in order to compensate for the increasing compromise of cognitive-linguistic function in patients with MCI.
Our findings of increased recruitment of a contralateral non-task-dominant region is also supported by previous lesion, aging, and clinical studies (Isaacs et al, 2019; Meinzer et al, 2009; Tinaz et al, 2008). Meinzer et al (2009) reported increased right inferior/middle frontal activity with worse semantic performance in older adults compared to young adults. Isaacs et al (2019) reported that PD patients compared to HC showed increased right hemisphere activity during a verbal production task related to verbal fluency. Another study demonstrated an increased recruitment of the right hemisphere in PD patients compared to age-matched and younger healthy controls during the semantic sequencing task (Peran et al, 2009). Compensation for the progressive loss of language function may therefore be reflected by the switch from the left to the right hemisphere, as observed in the present study. This switch may be exacerbated by increased task demands, particularly during the semantic switching task which requires both semantic memory and executive control. Considering the language function is typically dominated in the left hemisphere, the increased activation for additional recruitment can be related to the inefficient activation of the region in the dominant hemisphere (Braver et al, 2001).
The angular gyrus is also reported to be a part of the DMN which is highly active while at rest and decreases its activity during task performance (Buckner et al, 2008; Vatansever et al, 2017). DMN dysfunction in PD, particularly in patients with MCI, has been reported previously (Baggio et al, 2015; Gorges et al, 2015; Hou et al, 2016). Therefore, over-activated right angular gyrus in PD-MCI compared to decreased activity in PD-NC and HC may suggest disrupted connectivity of the DMN in PD-MCI.
A longitudinal study from the CamPaIGN cohort highlighted that impairment in semantic fluency is a key predictors of PDD, which was assumed to be related to posterior temporal-parietal regions (Williams-Gray et al, 2009). The present study confirmed that the increased activity in the parietal region was associated with worse semantic switching fluency in PD-MCI patients. Therefore, our results in PD-MCI patients indicate that the activation of the right angular gyrus may be a critical indicator of predicting PDD, and a possible target for future therapeutic interventions.
Strengths and Limitations
Stratifying PD-MCI from PD-NC was a great strength of this study, as it allowed for a more in-depth exploration of verbal fluency deficits related to cognitive impairment in PD. For example, Henry and Crawford (2004) found in their meta-analysis that PD patients were impaired on semantic fluency compared to HC. However, the results from the present investigation suggest that only PD-MCI patients exhibit impaired semantic fluency compared to HC, while verbal fluency in PD-NC is unimpaired. Furthermore, PD-MCI showed greater deficits during letter fluency compared to semantic fluency, which is contrary to Henry and Crawford (2004) meta-analysis, which reported the opposite pattern. This finding highlights the importance of looking at PD-MCI when exploring cognition in PD, particularly for the identification of those at risk of dementia. Another strength was the use of the automated speech as a control. The automatic speech condition requires over-learned responses without difficulty, therefore, this is used as a baseline for language production and to control for motion artifacts (Birn et al, 2010). Comparing the automatic speech condition to the semantic switching condition enabled us to isolate lexical-semantic processing from speech production.
It is important to consider that the verbal fluency assessments (outside the scanner) used to compare between groups were also used to separate patients into PD-MCI and PD-NC. However, PD-MCI is not defined solely by the performance of verbal fluency but rather is defined on a comprehensive cognitive testing. There were also many PD-MCI patients who were not impaired on semantic fluency.