Everyday Function in Alzheimer’s and Parkinson’s Patients with Mild Cognitive Impairment

INTRODUCTION

In both Alzheimer’s (AD) and Parkinson’s (PD) diseases, there is a spectrum of cognitive dysfunctions ranging from mild cognitive impairment (MCI) to dementia [1, 2]. MCI represents a prodromal stage of dementia in both diseases, and is defined as an undue decline from a premorbid cognitive level not normal for age but verified through cognitive testing, with preserved daily functioning [1–3]. In contrast, AD and PD dementia are characterized by a loss of independence indicated by problems with activities of daily living (ADL), in addition to cognitive decline. Although the presence of MCI is a risk factor for both types of dementia, not all MCI patients will develop dementia, as some individuals revert back to normal cognition or remain cognitively stable [4, 5]. Therefore, the investigation of factors associated with cognitive decline is crucial for identifying those MCI patients at a higher risk for developing dementia.

Research demonstrating the presence of ADL impairments in MCI is accumulating [6], specifically in complex instrumental activities of daily living (IADL), which include managing finances or using public transportation. These impairments in ADL emerge in the transition from MCI to dementia [7], and mild IADL changes can even be predictive of future cognitive decline [8]. Furthermore, there is increasing evidence for the presence of mild IADL impairment in AD-related amnestic MCI (AD-MCI) [9, 10], and individuals with AD-MCI and mild IADL deficits are at higher risk for progression to dementia [11, 12]. In PD-MCI, deficits in ADL function have also been shown to be related to worsening cognition and increased risk for PD dementia (PDD) in cross-sectional [13–18] as well as longitudinal studies [19]. However, assessing the impact of cognitive dysfunction on ADL is a challenge in PD, due to the interacting effect of motor impairment on daily function [14, 19–22]. Motor symptoms may therefore alter IADL function in PD-MCI differently than in AD-MCI [18, 23]. This highlights a need to examine both cognitive and motor influences on ADL to determine to which degree they contribute to deficits in both AD and PD. A better understanding of patients’ IADL characteristics, especially in the early stages of dementia, can facilitate personalized interventions in patients with different neurodegenerative disorders.

While IADL performance can aid in evaluating the progression of both PD and AD [24], previous research has mainly focused on the comparison of IADL impairment between AD dementia and PDD [14, 25–28]. Patients with PDD are characterized by a more severe progression of IADL dysfunction over time than AD dementia patients [28]. Importantly, motor disability was shown to be a significant contributor to IADL impairment in PDD, highlighting its role as a confounder [29]. However, a notable limitation of these studies is that patient groups were not matched for age, gender, ethnicity, or severity of cognitive impairment [28, 31]. One study controlling for these factors by comparing homogeneous patient groups with AD dementia and PDD was not able to detect differences in overall severity of IADL dysfunction, but identified different behavior and error profiles associated with this impairment in both groups [25].

Little is known about differences or similarities of the IADL profiles associated with MCI in AD and PD. One of the earliest IADL changes in AD-MCI may be the ability to manage finances [8], while difficulties remembering appointments was often the most impaired IADL [6]. These two activities best discriminated between cognitively normal and MCI patients [7]. Financial abilities have also been shown to be impaired in PD-MCI patients [17]. Moreover, keeping appointments, following recent events, managing finances, and using a telephone were specific ADL items that were unaffected by motor dysfunction, but able to identify PDD [14]. Only a few studies have evaluated IADL function in the prodromal dementia stage in AD and PD, reporting no differences in overall severity of IADL impairment between AD-MCI and PD-MCI [30, 32].

The focus of this study was therefore to explore the ADL profiles associated with the prodromal stage of AD and PD patients, as both of these neurodegenerative diseases have a well-characterized MCI stage. IADL dysfunction in AD is driven by memory impairment, which is a specific diagnostic feature occurring early in the disease. Therefore, items evaluating memory loss in everyday function might be more sensitive for assessing IADL within the frame of AD than for other diseases that have a more heterogenic profile of cognitive impairment in the prodromal stage of dementia, such as PD-MCI [33]. First, we aimed to investigate the association between cognitive dysfunction in the context of AD and PD and impairment on IADL using the Functional Activities Questionnaire (FAQ) [34]. In line with previous studies, we hypothesized that either AD-MCI and PD-MCI patients would demonstrate a similar overall degree of IADL dysfunctions assessed by the FAQ total score [30, 32], or patients with PD-MCI would score higher than AD-MCI, primarily due to the influence of motor performance in PD. Secondly, we aimed to specifically compare the task-related IADL profiles between AD-MCI and PD-MCI as, to the best of our knowledge, this comparison has not yet been performed. We hypothesized that both groups would be impaired in financial activities, with AD-MCI performing worse on items associated with memory impairment and PD-MCI showing more impairment in items related to non-memory domains, as well as tasks more prone to be affected by motor impairment. Lastly, we aimed to compare diagnostic merits of the FAQ for MCI due to AD and PD. The FAQ demonstrates good discriminative abilities in terms of distinguishing normal individuals and those with cognitive decline [34]. Therefore, we hypothesize that FAQ would have similar discriminant power for detecting IADL dysfunction in AD-MCI and PD-MCI patients, with a suspected higher cut-off for PD-MCI patients due to the potential influence of motor impairment.

METHODS

RESULTS

Classification accuracy for AD-MCI and PD-MCI according to FAQ scores

Figure 1 shows the ROC curves of the FAQ total score for diagnosis of MCI in both AD and PD, comparing MCI versus controls for each respective cohort. For classifying AD-MCI, an optimal FAQ cut-off score ≥1 was defined, showing a sensitivity and specificity of 69% and 90.5%, respectively. The optimal cut-off of the FAQ total score for distinguishing PD-MCI from PD-CN was ≥3 points. While the sensitivity of PD-MCI diagnosis was low (38.1%), specificity was excellent (92.9%). In addition, 9.5% of the HC and 7.1% of the PD-CN group scored above the pre-defined group cut-offs. An overview of diagnostic parameters, including the likelihood ratios, is displayed in Table 4.

OPEN IN VIEWER

Fig. 1

ROC curves of the FAQ total score for (A) HC versus AD-MCI and (B) PD-CN versus PD-MCI, including sensitivity and specificity.

OPEN IN VIEWER

Table 4

Diagnostic values of the FAQ total score for differentiating HC from AD-MCI and PD-CN from PD-MCI

	HC versus	PD-CN versus
	AD-MCI	PD-MCI
Cut-Off	>0	>2
Youden Index	0.60	0.31
Sensitivity, % (95% CI)	69.05 (52.9–82.4)	38.10 (23.6–54.4)
Specificity, % (95% CI)	90.48 (77.4–97.3)	92.86 (80.5–98.5)
Positive likelihood ratio	7.25	5.33
Negative likelihood ratio	0.34	0.67
Area under the curve, (95% CI)	0.82 (0.72–0.90)	0.65 (0.54–0.75)
p-value (AUC)	<0.001*	0.007*

AD-MCI, Alzheimer’s disease-related amnestic mild cognitive impairment; AUC, area under the Curve; CI, confidence interval; HC, healthy controls; FAQ, Functional Activities Questionnaire; PD-CN, Parkinson’s disease normal cognition; PD-MCI, Parkinson’s disease with mild cognitive impairment. ^*p < 0.05

DISCUSSION

This study explored IADL profiles associated with MCI in AD and PD. The present data analyses revealed no significant differences in global functional impairment expressed by the FAQ total score between AD-MCI and PD-MCI. However, task-related IADL profiles were different between the MCI groups. Most importantly, the number of patients with IADL impairment according to the calculated disease-specific cut-offs was nearly twice as high in AD-MCI (69.0%) as in PD-MCI (38.1%).

Our results support the hypothesis that severity of IADL impairment assessed by the FAQ total score does not differentiate between AD-MCI and PD-MCI patients when matched for global cognitive status, gender, age, and education. This is in accordance with previous research reporting no difference in IADL scores between amnestic MCI and PD-MCI [30, 32], or between AD and PDD patients when comparing overall performance-based ADL functioning [25]. In contrast to these reports, Farlow and colleagues [28] reported higher ADL impairment referring to both basic ADL and IADL tasks in PDD compared to AD. However, their study groups were not matched according to the level of global cognitive impairment, which was shown in our sample to have a strong influence on the severity of ADL impairment. Our findings highlight the importance of matching groups to avoid confounding effects, to accurately compare IADL profiles in different diseases.

The present analyses revealed differences in task-related IADL impairment between the MCI groups. We had hypothesized that both groups will be impaired in financial activities, with AD-MCI performing worse on remembering appointments and PD-MCI showing more impairment on tasks prone to be affected by motor impairment. Our results showed AD-MCI patients were more impaired than PD-MCI patients in IADL related to memory (remembering appointments), and handling finances, reflecting complex skills in major life areas and interpersonal interactions [7, 11]. Interestingly, patients with PD-MCI showed more problems with assembling taxes than AD-MCI patients. Previous studies have shown that impairment in assembling tax records distinguishes HC from AD-MCI patients [7], while handling finances are impaired in PD-MCI compared to PD-CN patients [17]; these findings were reversed in our results. It is possible that general financial capabilities are impaired with increasing cognitive deficits, regardless of the underlying neurodegenerative disease. We also cannot rule out that there are cultural differences between the US and Germany affecting financial and tax abilities that may affect our cohorts. However, previous studies in AD patients have shown that the FAQ item handling finances predicted progression from HC to AD-MCI and from MCI to AD dementia [7, 47]. This item may be of more value as a prognostic factor, than a differentiating factor. Moreover, more PD-MCI patients required assistance or were dependent on another person for assembling tax records. It has been previously postulated that assembling tax records is a complex task that involves executive functioning [7]. These executive functions contribute to ADL deficits, which are affected early in the disease course of PD [48, 49], possibly translating into a poorer ability to manage tax records. Future studies should examine associations between cognitive domains and the individual FAQ items.

PD-MCI patients also showed greater problems with mobility, which has been reported to be impaired in PD patients (walking in neighborhood, going by car, going by bus, driving a car) [50], and is one of the earliest deficits able to differentiate between pre-diagnostic PD and controls [51]. While our ANCOVA analysis showed that global cognitive functioning, and not motor severity, was the primary predictor of the FAQ, an influence of motor impairment cannot be excluded. It should also be noted that informants rated ADL dysfunction higher than PD patients did, which has been shown in previous studies where PD patients were shown to rate themselves as less impaired on measures of ADL than their caregivers [37, 52–54]. However, the interaction term between PD cognitive group and informant status was not significant, suggesting distribution across the groups was the same. It cannot be ruled out that some over-reporting of ADL dysfunction may have still occurred, as both patients and informants may have rated based on motor abilities instead of cognitive function. As other studies have been unable to verify this effect [55, 56], future studies should examine how informant ratings in PD reflect cognition or motor impairments.

All AD-MCI subjects had amnestic cognitive impairment, in accordance with the ADNI inclusion criteria. In AD, amnestic multi-domain MCI is usually found to be the most affected subtype on a functional level [57–59], and varying MCI subtypes differ in their profile of IADL impairment [58, 61]. To date, little is known about functional levels among subtypes of patients with PD-MCI, which can be classified as both amnestic and non-amnestic multi-domain MCI, the latter of which is probably associated with more marked executive dysfunction [62]. Compared to AD-related MCI, PD-MCI seems to be associated with more pronounced attentional and visuo-constructive deficits [32, 62]. Moreover, attention and language skills have been identified to be related to both IADL impairment and diagnosis of dementia in PD [63–65]. Therefore, we hypothesize that the differences in IADL task profiles of AD-MCI and PD-MCI can be explained by different cognitive deficit patterns that are associated with AD and PD, respectively.

As the FAQ is a measure originally conceptualized to detect functional dependence in normal aging and mild dementia, it may be more heavily weighted towards memory-dependent IADL [33]. In our sample, the discriminant ability of the FAQ for AD-MCI was good compared to controls, but the suggested cut-off (≥1 points) might overestimate IADL impairment in MCI. Previous studies demonstrated good diagnostic accuracy values of the FAQ for screening cognitive decline within AD [33, 67] and suggest a cut-off value of 1 for differentiating normal cognition from MCI or dementia [66], similar to our findings. For PD, we found an FAQ cut-off value ≥3 to distinguish cognitively impaired patients, with sensitivity and specificity of 38.1% and 92.9%, respectively. This is lower than the cut-off found in a previous study for differentiating between PD patients with normal cognition and those with PD-MCI or PDD [23]. Their cut-off of 3.5 had a sensitivity of 47.4% and specificity of 88.1%; however, it should be noted that the authors applied a shortened 8-item version of the FAQ where items 5 and 6 were found to be influenced by motor impairment and were therefore excluded [23]. The total 10-item FAQ score, which we used in our analyses, may thus not be comparable with the cut-off values reported previously.

Differences in cut-offs and previous reports of the discriminative ability of the FAQ for MCI might be at least partly caused by the high prevalence of MCI (50%) and the comparison of individuals matched for social-demographic and global cognitive status. As our intention was to compare the discriminative abilities of the FAQ for MCI in individuals with and without additional motor impairment like PD, our analysis might be not valid for the clinical diagnosis in a more heterogenic population. A solution might be incorporating Aβ_1–42 status as a biomarker when defining MCI [68], to make sure that AD pathologic change underlies AD-MCI. Low Aβ_1–42 values have also been identified as a potential risk marker for a more rapid disease course in PD [69, 70], but the predictive value of a dichotomized abnormal (low) Aβ_1–42 value is still under discussion. To compare both MCI groups according their diagnostic criteria, we therefore decided to use the clinical consensus criteria, excluding additional information gained by adding biomarkers. Future studies should aim to validate cut-offs for AD-MCI, as well as define accurate cut-offs for PD-MCI that take into account motor influences on ADL function.

The lower classification accuracy of the FAQ for our PD-MCI sample might be explained by the fact that only a subgroup of patients presented with IADL problems (38.1% compared to 69% of AD-MCI). These results are in accordance with the literature, showing a similar prevalence of mild IADL impairment among PD-MCI patients [13, 71]. In line with previous research [45], the present study also confirms that age is positively associated with IADL impairment when measured by the FAQ. For future research, we suggest developing separate normative standards for IADL impairments within PD-MCI, while also correcting for age effects.

This study faces some limitations. First, despite controlling for PD-specific motor impairment, we were not able to completely exclude this influence on the FAQ ratings. Second, the matched groups comprised only 42 participants each; however, we identified a distinct IADL profile in our homogenous MCI groups despite low sample size. Future studies in larger samples are needed to confirm our findings and to monitor change in IADL profiles over time in disease-specific MCI groups. For the identification of first symptoms of IADL impairment in PD, more sensitive assessments, such as performance-based tests or scales targeting the assessment of more complex IADL tasks, should be evaluated. It should also be noted that, due to the uneven distribution of informant ratings across the groups, the null effects of the interaction term (Informant Status x Group) in the ANOVA analyses should be interpreted with caution. Lastly, it must be noted that cross-cultural differences may have contributed to between-group differences. While little research has examined cultural and racial differences in the context of dementia, a recent study was not able to detect an ADL item bias across data from eight Western countries [72]. Therefore, we do not conclude that our differences are influenced by culture, although we cannot completely rule this out.