Specific Attentional Disorders and Freezing of Gait in Parkinson’s Disease

Objectives and hypotheses

The main objective of the present study was to investigate the nature of attentional disorders in PD patients with FoG by using a paradigm based on Van Zomeren and Brouwer’s model. We sought to compare performance levels in patients with and without FoG under different attentional conditions. On the basis of previous studies in which gait disturbances in PD patients were exacerbated during dual-task paradigms that involved sustained and divided attention [24], we hypothesized that patients with FoG (referred to henceforth as “FoG patients”) would perform less well in specific attentional modalities (namely in divided attention or mental flexibility conditions) than patients without FoG (referred to henceforth as non-FoG patients).

Experimental design and task

We assessed the patients’ performance in a computer-controlled reaction time (RT) paradigm designed to measure the different attentional subcomponents while controlling for visuospatial processing (via uniform central presentation) and motor participation (via a single-key response). The task has been described by Dujardin et al. [25]. The patients were assessed in an ecological “on” state after administration of their usual anti-parkinsonian medication(s).

Participants were seated in front of a 15-inch colour monitor. They were instructed to fix a grey square (2.5×2.5 cm) in the centre of the screen and to press the response key with their preferred hand as soon as the target stimulus appeared. The task comprised five levels:

A simple RT task, intended to measure processing speed.

The mean RT (in ms), the number of misses and the number of false alarms were recorded for each condition (except for the simple RT condition, in which false alarms were not possible). The test session lasted 45 minutes. The stimuli remained on the screen for 2 s at most or until a response was recorded during that time. Each level of the assessment was preceded by a practice block. The simple RT condition was always performed first. To limit order effects, the order of presentation of the four other conditions was counterbalanced.

All participants gave their informed consent to participation in the study. The study protocol was approved by the local independent ethics committee.

Population

Seventy-eight patients with PD (diagnosed according to accepted international criteria [26]) were recruited by the Department of Neurology and Movement Disorders. The FoG and non-FoG groups were constituted according to the patients’ responses to item 3 of Giladi’s FoG questionnaire [27]. All included patients suffered from ‘off-drug’ FoG [28], defined according the clinical response of FoG with levodopa administration. All patients performed a FoG trajectory [11] composed of gait initiation, turning, going through narrow passages and performing dual tasks. We determined a FoG trajectory score based on the duration of FoG episodes during the FoG trajectory: 1 point for a brief episode of FoG (<10 s), 2 points for a moderate episode (10–30 s) and 3 points for a long episode (>30 s).

Demographic and disease-related variables were collected during an ad hoc clinical interview. All patients were also assessed with the following instruments: part III of the Unified Parkinson’s Disease Rating Scale (UPDRS), the FoG questionnaire [27], the Mini Mental State Examination (MMSE) and the Mattis Dementia Rating Scale. The levodopa equivalent daily dose was calculated [29]. The main exclusion criteria were the presence of neurological or psychiatric disorders other than PD or an MMSE of 25 or less.

Statistical analyses

As distributions were not Gaussian, the demographic and clinical data from the FoG and non-FoG groups were compared by applying a Mann-Whitney test (for continuous variables) or a chi-squared test (for categorical variables). For RT data, the RTs’ coefficients of variance and accuracy (omissions and false alarms), that were normally distributed, analyses of covariance (ANCOVAs) were performed with group (FoG, non-FoG) as a between-group factor and with confounding factors as covariates (namely disease duration and deep brain stimulation (DBS), which are strongly associated with FoG [1]). Because disease duration and levodopa therapy were highly correlated, we used only one of these two parameters (the one that had the more robust association with RT data) as a covariable.

A Greenhouse-Geisser correction was applied when the assumption of sphericity did not hold. To explore putative associations between FoG characteristics and RTs, Spearman’s correlation coefficient was calculated.

The threshold for statistical significance was set to p <  0.05 in all cases. All analyses were performed with SPSS ^® for Windows ^® software (version 16.0, IBM Inc., Armonk, NY, USA).

Population

The demographic and clinical characteristics of the two PD groups are summarized in Table 1. Intergroup comparisons did not reveal any differences in terms of age, duration of formal education or global cognitive efficiency. The FoG group had a longer disease duration and a higher levodopa equivalent daily dose, relative to the non-FoG group (p <  0.05). The prevalence of DBS was also higher in the FoG group (p <  0.05). Thus, disease duration and the presence of DBS were included as covariates in subsequent statistical analyses. The levodopa equivalent daily dose was not included in the covariates because it correlated with disease duration (rho = 0.421, p <  0.05).

The two groups of patients did not differ in terms of the UPDRS motor score in the “on” state (p = 0.22). In contrast, the groups differed for variables that directly reflected the FoG phenomenon, such as the FoG questionnaire score (p <  0.05) and the FoG trajectory score (p <  0.05).

Attentional raw data

The mean (SD) performance levels of the two groups under the five conditions of the computer-controlled RT paradigm are shown in Table 1. In the absence of adjustment, RTs were higher in the FoG group than in the non-FoG group (p <  0.05) for all attentionalmodalities.

When disease duration and DBS were considered as covariates in ANCOVAs (Table 2), the only significant difference between the FoG and non-FoG groups concerned the RTs in the divided-attention condition (p = 0.02). There were no intergroup differences in RTs under the other conditions (p >  0.05). Furthermore, there were no intergroup differences in the accuracy of responses (p >  0.05 for all conditions).

Attentional RTs were correlated with FoG variables: FoG-questionnaire score and FOG-trajectory score (Table 3).

In Supplemental data, the specific effect of DBS in the subgroup of patient with FoG is also shown.

FoG: A failure of divided attention?

The presence of an impairment in divided attention distinguished between patients with FoG and those without. Importantly, our results clarified the role of attention in FoG and evidenced the specific role of divided-attention impairment in patients with FoG, which had not been possible with dual-task paradigms [9, 21]. Indeed, dual-task paradigms involve the combined use of working memory and executive functions. In our present task, the patient had to simultaneously select two features of the stimulus before taking a decision (rather than having to cope with two different motor tasks). In fact, the motor involvement in our paradigm was low since the subjects had only to press a single response button. Moreover, the divided attention condition did not constitute a dual task. Hence, in a simple task with a single encoding modality (a visual modality, in the present study), PD patients with FoG processed concomitant information more slowly than patients without FoG –highlighting an early failure in simultaneous processing. These results increase knowledge of the mechanisms involved in the dual-task deficit in freezers. Usually, the explanation for dual-task difficulties in PD is that previously automated procedures (generally for motor activity, such as gait) are impaired by basal ganglia deficiency and are replaced by components controlled by the supervisory attentional system [23]. Here, the basic impairment in simultaneous two-stimulus integration (with no involvement of complex motor programs) suggests that patients with FoG have difficulty processing several simultaneous streams of data (rather than choosing relevant data - as in the flexibility task, where two orders had to be memorised simultaneously). Our results in patients with FoG suggest that their impairment in divided-attention tasking is located upstream of the high planning/regulating functions that are usually involved in dual-task impairments. The partial respect of their capacities in the flexibility task plead for a effective central executive [32] and a location of their impairment in divided-attention tasking even further upstream than the previously described limitation of attentional allocation.

However, this early impairment in processing several information concomitantly is probably greater when the patients with FoG have to deal with two different tasks at the same time (as is often the case in activities of daily living), with regard to capacity-sharing theories and a spectrum of prioritization [33].

FoG: An impairment of attention distribution?

Focused attention and mental flexibility were relatively unaffected in patients with FoG. This finding suggests that the necessary attentional resources are available but are poorly recruited. By considering attention simply as the sum of distinct subunits or modules (thus explaining limited attentional capacity) [34] in which several components run distinct processes and are functionally distinct [35], freezers are able to perform each correctly in an alternating choice RT condition (in which attention is alternately allocated to the number of squares and to their colour). That means that the sum of units is unaffected in freezers. However, when patients with FoG had to recruit all resources at the same time (in order to integrate information concerning the number and the colour of the squares), an impairment was observed. Although patients with FoG appear to have the same resources as patients without FoG and to be able to allocate attention to each single task, there does appear to be a difference in distribution of resources when required. Difficulty coping with several features at the same time might be a central feature of FoG and might be related to failure of the supervisory attentional system [23] or an overload in parallel processing of stimuli [31].

It is interesting to consider why this impairment could be particularly deleterious for freezers patients. Indeed, divided attention usually enables better exploration of the environment than focalized attention because the former is thought to involve “zoom lens” visuospatial attention [36]. When the subject fixes a single location in space, signal discrimination at that location is more difficult because all the signals have to be treated simultaneously by the same channel. Hence, the discrimination of incongruent sensory signals at this single, monitored location is more difficult because separation of the stimulus’ independent components is more complex. In divided attention tasks, the presentation of several modalities or locations enables the subject to independently select stimuli “in parallel”, which thus increases the overall amount of available resources. Processing two instructions in parallel is a means of recruiting separate pools of attentional resources [37]; this is especially true for the simultaneous processing of stimuli from different modalities [38]. In this respect, the specific impairment inincreasing attentional resources to process concomitant data (which appears to be crux of the divided-attention problem in patients with FoG) is particularly disabling; patients with FoG are capable of focusing their attentional resources but not distributing them between several parallel input streams.

Prioritization, interference resistance and accuracy

In both groups, the trade-off between speed and accuracy tended to reflect a lengthening of the RT and the maintenance of accuracy under all attentional conditions. Indeed, there were few false alarms and omissions in either group. Interestingly, a parallel can be drawn with the “posture second” strategy described by Bloem et al. [39], in which PD patients’ behaviour was marked by a preference for cognitive accuracy over motor performance. In divided-attention tasks with different inputs, the patient spontaneously considers cognitive aspects and does not prioritize the motor aspect, what was also noted in our results although the present study’s motor task was not very challenging. It reflects the poor central processing capacity [32] in patients with PD. Indeed, patients with PD are known to process tasks sequentially, in order to avoid system overload [30]. Here, patients with and without FoG displayed the same prioritization strategy and resisted interference to a similar extent, as evidenced by relatively unaffected performance in a focused attention task. The patients with FoG did not appear to be more impaired in information selection or filtering because the ability to inhibit a motor response (low number of false alarms and results in a “go–no go” task) or neglect a distractor (in a focused attention task) was unaffected; this may be due to the up-regulation of prefrontal dopaminergic receptors involved in resistance interference [40]. Previously, conflict resolution impairment was noted in freezers patients [19, 41] when they had to cope simultaneously with incongruent and relevant stimuli. Our results suggest that executive control is deficient in freezers, but also simply processing several features on a single stimulus at a same time.

What about neural substrates?

Our results agree with current hypotheses on FoG. Indeed, neuroimaging data revealed increased activity in the posterior parietal cortex when monitoring two modalities at different locations, while no specific region was recruited in the focused attention conditions [37]. In the same manner, in an object recognition divided attention paradigm the bilateral intraparietal sulcus was involved and correlated with attentional performance [42]. This region allows in-parallel processing when attention is spatially divided, referring to the cortical hypotheses of the origin of the FoG phenomenon [36] and explaining why the integration of external stimuli to drive movement is so deficient in freezers [43]. The frontoparietal regions of the cognitive control network [44], involved in the FoG phenomenon [45] also seems a good candidate for this low-level attentional discrimination processing (see Supplemental Figure).

Limitations

We performed a descriptive, cross-sectional study; further prospective follow-up studies are needed to establish whether early failure of divided attention in PD could predict the occurrence of FoG. Attentional tasks were performed in the “on drug” state and so the mean levodopa equivalent daily dose was higher in the FoG group. This may explain the comparable processing speed but may have modified attentional performance. We deliberately chose to test patients in the “on-drug” condition, in order to reflect ecological processing conditions.

Furthermore, it is important to keep in mind that the two groups were different: even with adjustment of analyses for disease duration and DBS, freezers patients had generally more advanced disease and could suffer from more diffuse lesions whose attentional results could also reflect. Concerning the others potential bias, we did not include levodopa therapy as a covariate because of the redundancy with disease duration. However, it could partly explain our results in ecological situation [46, 47]. Finally, the subgroup analysis highlighted the weight of DBS in the impairment in divided attention inside the FoG group. All these factors related to PD progression could impact attentional performance and act as a confounding factor in the evaluation of FoG [48].

In previous research [25], a specific impairment in mental flexibility in patients with PD was described. This impairment is critical for attention failure. However, mental flexibility does not appear to be factor that most strongly discriminates between PD patients with FoG and those without.

Usually, neuropsychological tests are performed in on drug condition as in this study, in order to reduce interference due to the akinesia or slowness. Because the off-FoG mainly occurs when the levodopa level decreased, the specific evaluation of attentional performance at this time would be very helpful, for example related to the levodopa pharmacokinetics. Thelimiting factor would be the necessity to take into account non-motor fluctuations (depression, anxiety or apathy) frequent in such circumstances and that could also interfere with attentional performance and trigger the FoG phenomenon [49]. Further studies are necessary to better apprehend the interplay between therapies, cognitive and affective aspects in the FoG occurence [7].