Paired Studies Comparing Clinical Profiles of Lewy Body Dementia with Alzheimer’s and Parkinson’s Diseases

Study population

The research complied with the Declaration of Helsinki and was approved by the Ohio State University’s Biomedical Sciences Human Subject Institutional Review Board. The study used a matched-pair cross-sectional design.

Participants

All subjects met clinical diagnostic criteria for their specific neurodegenerative condition. Inclusion criteria included vision and English sufficient for compliance with testing, Mini-Mental State Examination (MMSE) score greater or equal to 15, and the ability to ambulate 15 feet independently or with the use of an assistive device (i.e., walker or cane). Subjects were excluded if they had additional conditions (other than the diagnoses of interest) that in the opinion of the investigators might contribute to dementia, gait, or balance impairment or complicate their assessment.

Individuals meeting inclusion and exclusion criteria and with an informed consent, or assent through a legally authorized representative consent, were enrolled. Permission from the subject was obtained to be able to interview a study partner who knew the subject well. Subjects were recruited from the Memory Disorder and Movement Disorder Clinics at The Ohio State University Wexner Medical Center. Subjects with LBD met standard clinical criteria for either DLB or PDD as established by McKeith or Emre, respectively [17, 32]. All AD subjects met clinical diagnosis of probable AD according to the National Institute of Neurologic and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association and DSM-IV TR criteria [33]. PD subjects met clinical diagnostic criteria according to the UK Parkinson’s Disease Society Brain Bank [34]. Strict adherence to clinical criteria for diagnosis was maintained and all subjects were carefully assessed and diagnosed as typical cases by both movement and cognitive experts.

Each DLB and PDD subject was matched by age, gender, education, race, and MMSE [35] score with one subject with AD. They were also matched by age, gender, education, race and Unified Parkinson’s Disease Rating Scale-Motor Examination (UPDRS III) [36] score with one subject with PD.

Outcome measures

Study subjects underwent a medical evaluation including a medical history, physical examination, orthostatic vital signs, and neurological examination including UPDRS III and Hoehn and Yahr (H&Y) staging [37]. We designated specific components of the UPDRS III into the following subsections that have been reported commonly in the literature: axial Postural Instability and Gait Difficulty (PIGD; arising from chair, posture, gait, and postural stability), axial bulbar (speech, facial expression), and bradykinesia (hand movement, rapid alternating movements of hands, leg agility, body bradykinesia) [25]. Each subject underwent cognitive, gait, and mobility evaluations. Cognitive assessment consisted of Self-Administered Gerocognitive Examination (SAGE), MMSE, FAS verbal fluency task (the sum of the number of words in one minute starting with “F”, then “A”, then “S”), Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), modified (15-word) Boston Naming Test, Wisconsin Card Sort Task-64 (WCST-64), forward digit span, and Alzheimer’s Disease Assessment Scale (ADAS) constructional task [38–40]. Motor measurements included Berg Balance Scale (BBS), Tinetti Mobility Test (TMT), Figure of 8 (FO8), and nine hole Pegboard Test (9HP) [41–44]. Study partners were interviewed for assessments of subject behavior and function using the following: Epworth Sleepiness Scale (ESS), Mayo Sleep Questionnaire-Informant (MSQ), Mayo Fluctuations Scale (MFS), Neuropsychiatric Inventory (NPI), and Activities of Daily Living (ADLs) interview [45–47]. Study partners and subjects were queried about duration of cognitive impairment (if any), prior diagnosis of sleep apnea, duration of Parkinsonism (if any) and falls history (if any). Subjects were asked to take all regularly scheduled medications during the study visit.

Statistical analyses

Summary statistics for continuous data are given by mean±SD, and the range, min-max. Paired differences were computed and tested for normality using Shapiro-Wilk test with a p-value of 0.10 as a threshold for normality. For normal data paired t-test was used and Wilcoxon signed rank test was used otherwise. For paired nominal data McNemar test and 95% confidence intervals (CI) were used. Level of significance was set at 0.05 and appropriate Bonferroni corrections were implemented to adjust for multiple tests. Spearman’s correlation coefficient was used to express the strength of association. Statistical software SAS JMP 11.0 was used for statistical analyses.

Subject exclusions

Seventy-two (72) subjects were screened and sixty-three (63) subjects, all Caucasian, meeting inclusion and exclusion criteria were enrolled for the study. Nine subjects were excluded due to low score MMSE (n = 2), uncertain diagnosis (n = 3), diabetic neuropathy affecting ambulation (n = 1), and severe knee pain affecting ambulation (n = 1). Two subjects could not be matched: one subject with 10 years of education and one subject with a low UPDRS III score. Twenty-one subjects met clinical criteria for either DLB (n = 11) or PDD (n = 10), were pooled into one group designated as LBD, and were individually matched with 21 subjects who met clinical criteria for AD and 21 subjects who met clinical criteria for PD.

Demographics (Table 1)

Demographic data and clinical characteristics for the three groups are presented in Table 1. As expected, there were significant differences between LBD and PD subjects for MMSE (mean difference (MD): –5.24±3.48, p < 0.0001), SAGE (MD: –7.19±5.33, p < 0.0001), and CERAD (MD: –13.91±11.0, p < 0.0001). PD subjects had lower H&Y scores (MD: –0.31±0.46, p = 0.0112) and longer duration in years of motor symptoms than LBD subjects (MD: –4.13±7.82, p = 0.025). Also as expected, there were significant differences between LBD and AD for UPDRS III scores (MD: 22.05±7.10, p < 0.0001) and H&Y scores (p < 0.0001). Eighty-one percent of LBD group used anti-Parkinsonian medication (95% CI = 60% –92%).

Motor comparisons between LBD versus PD and AD (Table 2)

Despite being matched by total UPDRS III scores, PD subjects performed significantly better than LBD subjects on the following measures of motor function: UPDRS III sum of axial subscores (PIGD; arising from chair, posture, gait, postural stability), BBS, TMT, non-dominant hand 9HP score, and number of missteps in FO8 after accounting for multiple tests. No significant differences were found between the groups in the other UPDRS III subscores. LBD subjects demonstrated more motor impairment than AD subjects on all measures (p-values not reported).

Cognitive comparisons between LBD versus AD and LBD versus PD (Table 3)

Despite being matched by MMSE score, AD subjects were significantly more impaired than LBD subjects with orientation questions of MMSE. AD subjects also had significantly worse delayed recall, more false positive intrusions and poorer recognition discrimination memory on the CERAD memory list. This implies a retrieval memory disturbance for LBD subjects and an amnestic memory impairment for AD subjects.

On the other hand, LBD subjects showed significantly more deficits compared to AD subjects on the executive domain tasks of SAGE, FAS verbal fluency, and the ADAS constructional task, with near significance on the visuospatial domain tasks of SAGE, and the MMSE interlocking pentagon construction task.

There were no significant differences noted between LBD and AD subjects on CERAD word list learning or intrusions, forward digit span, modified Boston Naming Test, WCST, or the language, memory, or reasoning/computation domains of SAGE. LBD subjects demonstrated more cognitive impairment than PD subjects on all measures.

Autonomic, behavioral, and other comparisons between LBD and PD and between LBD and AD (Table 3)

Compared to either PD or AD subjects, LBD subjects were significantly more impaired in ADLs and had significantly more daytime sleepiness (ESS score), decreased level of alertness (question 8 of the MSQ), cognitive/behavioral fluctuations (MFS composite total score), hallucinations (NPI subscore), and increased incidence of sleep apnea (prior diagnosis). Compared with AD subjects, LBD subjects had significantly more symptoms of orthostatic hypotension (vital signs on study visit), falls, and marginally significantly more restless leg syndrome (question 3 of the MSQ) but there was no significant increase in the presence of dream enactment behaviors (question 1 of MSQ).

Correlations between motor and cognitive performance in subjects with LBD and PD (Table 4)

We examined the Spearman correlations between seven cognitive scores (SAGE total, SAGE executive, SAGE visuospatial, MMSE total, CERAD total, CERAD memory, and FAS verbal fluency) and nine motor tasks (Berg Balance Scale, Tinetti Mobility Test, missteps in the FO8, non-dominant 9HP, and the following five subscores of the UPDRS III: axial PIGD items, axial bulbar items, rigidity, tremor (resting & action), and bradykinesia) for the LBD and PD subjects. Except for the axial PIGD items, correlations between the seven cognitive scores and UPDRS III subscores were small (under 0.29, p-value >0.05) and are not reported in Table 4. However, accounting for the 63 correlations considered and using the Bonferroni correction, only the seven with p-values under 0.000793 are taken to be significant. These are numerically high as well (ranging from 0.498 to 0.582) representing a strong clinical association between the concerned variables.

Table 4 shows strong associations, many statistically significant, between SAGE total, executive, and visuospatial scores and motor tasks. Among other cognition test scores, only the CERAD total score correlates with the BBS total score significantly.

Method and design

Currently accepted diagnostic criteria for DLB and PDD are based on having one year or less between onset of dementia and parkinsonism for DLB diagnosis, the so called “one year rule”. Identifying the onset of dementia is subjective and often suspect in its accuracy [12, 17–20]. Therefore, we decided to combine our DLB and PDD subjects into one group, called LBD.

Our average MMSE of 22.6 allowed us to investigate early cognitive pattern differences and more accurately differentiate and diagnose these disorders. As dementia conditions become more severe, differences in cognitive profiles are harder to evaluate due to floor effects of cognitive tests.

Our inclusion criteria also required our study subjects to be able to ambulate 15 feet. Comparing ambulatory subjects defines early gait pattern differences and provides clues to more accurately differentiate and diagnose these disorders. As Parkinsonian symptoms become more severe, patients become non-ambulatory and gait differences can no longer be assessed.

The standard clinical criteria for DLB considers whether the participant has fluctuating attention, REM sleep behavior disorder, and visual hallucinations. These features are not required as part of the standard clinical criteria to diagnose PDD. In our study we had 11 DLB subjects and 10 PDD subjects that were pooled together to make up our LBD group. Therefore, our comparisons of cognitive/behavioral fluctuations, REM sleep behavior disorder, and visual hallucinations between the LBD group and the AD and PD groups may reflect in large part the impact of the diagnostic clinical criteria used for our 11 DLB subjects.

Unlike many studies using only group comparison or age matched samples, the strength of our method is the gender, age, education, race, and MMSE or UPDRS III scores matched comparison.

Differentiation of PD and LBD subjects based on motor characteristics

The items of the UPDRS III that include arising from a chair, posture, gait, and postural stability have been referred to as axial PIGD items [25]. Some investigators include speech and facial expression (bulbar features) or neck rigidity as additional axial motor items [25–27]. In our study, LBD subjects, while matched for UPDRS III total score, did significantly worse than the PD subjects on only PIGD items (Table 2). LBD subjects also had significantly worse gait and balance compared to PD subjects, who tended to have more resting tremor. Our results are consistent with other studies showing that patients presenting with more prominent resting tremor compared to PIGD and balance disturbances are more likely to have PD, while those with more prominent PIGD and balance issues compared to resting tremor may be more likely developing a LBD condition[1, 24–30].

The prominent non-dominant hand 9HP score differences between LBD and PD subjects likely reflect the more impaired visuospatial impairments seen in our LBD group [48].

Differentiation of AD and LBD subjects based on cognitive characteristics

Despite being matched for MMSE (global cognitive) scores, LBD subjects were significantly more impaired than AD subjects in executive and visuospatial domains and significantly less impaired in orientation, delayed memory recall, and recognition discrimination memory. The progression of pathology in LBD starts in the brainstem and deep grey matter before spreading to limbic regions and parietal-occipital, posterior temporal-occipital and frontal cortical regions. This corresponds to more retrieval memory (frontal-subcortical pathways), executive (frontal), and visual processing (parietal-temporal-occipital) deficits [49]. The progression of pathology in AD starts in the mesial temporal lobe before spreading to frontal and parietal cortical regions resulting in an amnestic pattern of memory loss (hippocampus), orientation impairments due to memory deficits, and eventually aphasia (peri-Sylvian), visuospatial (parietal), and executive (frontal) deficits [50]. The timing of the cognitive deficits is distinctly different in AD and LBD and therefore can be very useful in differentiating those conditions.

Although our LBD subjects did significantly worse than our AD subjects on the executive measures we employed, the pattern of dysfunction in LBD and AD subjects appear to be similar, involving more dorsolateral frontal (verbal fluency, problem solving, modified Trails B) and less orbitofrontal (disinhibition or impulsivity) impairments. Visual processing and executive planning impairments in our LBD subjects combine to create more constructional deficits than in our AD subjects matched to the same MMSE score.

Differentiation of PD and AD from LBD subjects based on autonomic, behavioral, and other characteristics

Since some ADLs require both cognitive and physical skills it is not surprising that our LBD subjects with both physical and mental impairments did worse than matched AD (predominantly cognitive disabilities) or matched PD (predominantly physical disabilities) subjects. Alpha-synuclein aggregates are commonly found in the peripheral autonomic nervous system in both non-demented PD and LBD patients [51], which may explain why orthostatic hypotension was more commonly seen in our PD and LBD subjects than in the AD subjects. Orthostatic hypotension likely contributes to increased falls. Central autonomic control pathways outside the hypothalamus or brain stem structures are differentially impacted by cortical Lewy bodies in LBD patients compared to PD patients [52]. While further study is warranted, the alterations in those pathways in LBD subjects might be correlated to the excessive daytime somnolence, impaired alertness, cognitive/behavioral fluctuations, and sleep apnea seen significantly more often in our LBD subjects than either our matched PD or AD subjects [52–54]. Other autonomic-influenced sleep impairments including symptoms of restless leg syndrome and REM sleep behavioral disorder were seen similarly in PD and LBD subjects and more commonly than in our matched AD subjects [53, 54]. Visual hallucinations occurred in 62% of our LBD subjects but only in 9.5% of our PD and AD subjects. Visual hallucinations are thought to be more common in LBD subjects due to alpha-synuclein proteinopathy involving the visual processing posterior cortical regions and the limbic areas [55]. More research is required to fully understand the pathophysiology of the behavioral disturbances.

Correlations between motor and cognitive features

SAGE scores were significantly correlated with BBS, TMT, missteps in the FO8, and non-dominant 9HP scores (Table 4). Some of these high correlations may be due to the strong association between the four motor tasks (the correlations between them ranged from 0.45 to 0.81). Among other cognitive tests performed, only CERAD total score exhibited strong association with BBS. Executive and visuospatial cognitive domains had higher correlations than temporal (memory) domains. Frontal and parietal cognitive deficits might predictably correlate with the motor tasks and spatial processing involved in balance, walking in FO8, and performing spatial tasks utilizing a pegboard. Walking and balance relies on several cognitive skills, particularly executive and visuospatial abilities. Intact attentional skills may also play a role, and in our study, PD subjects performed better than matched LBD subjects on serial 7 subtractions.

The PIGD features did not reach significance but correlated the best with global cognitive results (CERAD and SAGE) and executive domain assessments (SAGE and FAS verbal fluency), and correlated less well to memory (CERAD) or visuospatial (SAGE) domain scores. Therefore, frontal more than temporal or parietal cognitive deficits correlate with axial motor impairments. It is interesting to note that the cognitive deficits and the axial motor impairments that correlated the best are both associated with frontal regional impairments. Those cognitive findings prominently involve frontally located executive domain deficits and those axial motor findings suggest impairments in the supplementary motor area, also located in the frontal lobes. The supplementary motor area is thought to be involved in control of postural stability with standing or walking, initiation of internally generated motor activities like arising from a chair, and coordinating temporal sequences of action like walking [25–28].

The non-PIGD items that contribute to 85% of the UPDRS III score, i.e., axial bulbar items of facial expression and speech, tremor, rigidity, bradykinesia, and finger taps, did not correlate with cognitive measures. These motor measures also did not differentiate PD from LBD subjects. This suggests that these motor findings are more related to damage in subcortical and brainstem regions than from cortical regions [25–28]. Therefore, patients presenting with subtle evidence of PIGD and cognitive impairment might alert clinicians to possible LBD.

Limitations of the research

Limitations include small sample size, particularly with the PDD and DLB subjects, diminishing meaningful comparisons between these groups. In order to reduce the chance of having subjects with the pathological features of both AD and dementia with Lewy bodies (the Lewy body variant of AD), strict adherence to clinical criteria for diagnosis was maintained and all subjects were carefully assessed and diagnosed as typical cases by both movement and cognitive experts. By these methods, subjects with atypical symptoms at the onset of their condition and those with uncharacteristic clinical courses were avoided. Nevertheless, this clinical study did not include the use of DaTscan, amyloid PET imaging, cerebrospinal fluid analysis, polysomnography, or pathology to substantiate or negate the clinical diagnoses given. Going forward, more research in establishing the criteria of when to use these other biomarker oriented techniques and which combinations are most useful and cost effective will be a great help in aiding the diagnosis of these neurodegenerative conditions.

Our cognitive tool for matching, the MMSE, exhibits ceiling effects in more educated patients and is not sensitive for mild cognitive impairment or executive deficits. LBD subjects have prominent executive impairment so the MMSE may not accurately reflect their global cognitive impairments. Despite its shortcomings, we used the MMSE due to its familiarity and long tradition as a brief global cognitive assessment instrument. Our motor tool for matching, the UPDRS III, has floor effects that may limit sensitivity in early disease, concealing important differences between PD and LBD subjects.

While the milder cognitive impairment (average MMSE = 22.6) and mild motor dysfunction (requirement for ambulating 15 feet) of our subjects limits its generalizability, our focus was on distinguishing characteristics in early disease. Our LBD subjects also had more falls than our matched PD subjects in the last 6 months but significance was not achieved. This trend might become significant if more subjects were evaluated or if we had subjects with more severe cognitive impairments. Some studies have shown that falls may predict cognitive decline [30]. We excluded subjects who had undergone deep brain stimulation, which may have inadvertently excluded PD subjects with severe tremor.