European Journal of Nuclear Medicine and Molecular Imaging

, Volume 39, Issue 11, pp 1767–1777

Posterior parietooccipital hypometabolism may differentiate mild cognitive impairment from dementia in Parkinson’s disease

Authors

  • David Garcia-Garcia
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
    • Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED)
  • Pedro Clavero
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
  • Carmen Gasca Salas
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
    • Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED)
  • Isabel Lamet
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
  • Javier Arbizu
    • Department of Nuclear Medicine, ClínicaUniversity of Navarra
  • Rafael Gonzalez-Redondo
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
    • Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED)
  • Jose A. Obeso
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
    • Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED)
    • Neurosciences Area, CIMA, Department of Neurology and Neurosurgery, Clinica Universidad de NavarraUniversity of Navarra
    • Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED)
    • Department of Neurology and NeuroscienceUniversity Hospital Donostia, BioDonostia Research Institute
    • Ikerbasque, Basque Foundation for Science
Original Article

DOI: 10.1007/s00259-012-2198-5

Cite this article as:
Garcia-Garcia, D., Clavero, P., Gasca Salas, C. et al. Eur J Nucl Med Mol Imaging (2012) 39: 1767. doi:10.1007/s00259-012-2198-5

Abstract

Purpose

Patients with Parkinson’s disease (PD) may have normal cognition, mild cognitive impairment (MCI) or dementia. We investigated differences in cerebral metabolism associated with these three cognitive states and the relationship between metabolism and cognitive dysfunction.

Methods

FDG PET and a battery of neuropsychological tests were used to study PD patients with dementia (n = 19), MCI (n = 28) and normal cognition (n = 21), and control subjects (n = 20). Regional glucose metabolism in patients and controls was analysed using statistical parametric mapping (SPM8) corrected for age, motor severity and depression. Correlations between the mini-mental state examination score and Z-score values of the different cognitive domains with respect to cerebral FDG uptake were assessed using SPM8.

Results

PD patients with MCI (PD-MCI patients) exhibited decreased FDG uptake in the frontal lobe, and to a lesser extent in parietal areas compared with cognitively normal patients. Patients with dementia showed reduced metabolism in the parietal, occipital and temporal areas and a less extensive reduction in the frontal lobe compared with PD-MCI patients, while widespread hypometabolism was seen in comparison with patients with normal cognition. PD-MCI patients exhibited reduced FDG uptake in the parietal and occipital lobes and in localized areas of the frontal and temporal lobes compared with controls, whereas patients with dementia showed a widespread reduction of cortical metabolism. Mini-mental state examination score correlated positively with metabolism in several lobes, executive function with metabolism in the parietooccipitotemporal junction and frontal lobe, memory with temporoparietal metabolism, visuospatial function with occipitoparietal and temporal metabolism, and language with frontal metabolism.

Conclusion

PD patients with MCI exhibited hypometabolism in several cortical regions compared with controls, and in the frontal and parietal regions compared with cognitively normal patients. Hypometabolism was higher in patients with dementia than in those with MCI, mainly in the posterior cortical areas where it was correlated with visuospatial, memory and executive functions.

Keywords

Parkinson’s diseaseMild cognitive impairmentPETCerebral metabolismDementia

Introduction

Cognitive impairment is a frequent comorbidity in Parkinson’s disease (PD), with a reported dementia prevalence of up to 80 % in long-term longitudinal studies [1, 2]. Mild cognitive impairment (MCI) is defined as a cognitive decline that is not normal for age but in which essentially normal functional activities can be maintained [36]. This condition is also common in PD and is considered a risk factor for the development of dementia [7]. As yet, the pattern of progression of the cognitive decline from MCI to dementia in PD patients has not been well defined, and longitudinal studies addressing the neuropsychological predictors of dementia in PD have yielded inconsistent results [69]. However, a longitudinal study on early PD concluded that patients with deficits in tasks with a more temporal and parietal lobe involvement (“posterior cortical” dysfunction) have a higher risk of developing dementia [10]. Similar results were found in a cross-sectional study assessing the cognitive changes characterizing the transition from MCI to dementia in PD [11]. In keeping with this, a recent longitudinal study involving FDG PET showed that patients who develop dementia have reduced baseline FDG uptake in the visual association area and posterior cingulate cortex [12].

Cross-sectional studies with FDG PET have revealed that dementia is associated with widespread areas of cortical hypometabolism [1319], while in PD patients with MCI (PD-MCI patients), hypometabolism appears to be more localized to the temporoparietooccipital junction and the frontal cortex [1820] compared with healthy controls. In addition, PD-MCI patients show reduced FDG uptake in the frontal and parietal regions with respect to cognitively normal PD (PDCN) patients [19, 21]. However, the metabolic changes that distinguish PD-MCI patients from PD patients with dementia (PDD) have not been studied as yet.

We hypothesized that PDD patients would have greater hypometabolism in posterior cerebral areas than PD-MCI patients. Here, we describe patterns of cerebral metabolism in PD-MCI patients compared with PDD patients and with PDCN patients. Our aim was to identify metabolic differences between the cognitive states in PD, specifically between dementia and MCI. We also report the correlations between cerebral metabolism and cognitive status in specific cognitive domains.

Material and methods

Subjects

A cross-sectional study was conducted in patients with PD diagnosed according to the UK Parkinson’s Disease Society Brain Bank criteria [22] who were consecutively recruited from the Movement Disorders Unit of the Clinica Universidad de Navarra. Patients over 60 years of age and with a disease duration of at least 10 years were included, as this profile best represents the PD population with the highest risk of cognitive decline [23]. Exclusion criteria were other brain disorders, abnormal findings on MRI (i.e. tumour, hydrocephalus or severe vascular lesions), severe systemic disease, major psychiatric illness, prior cerebral surgery, abnormalities in thyroid function, positive VDRL test and low levels of vitamin B12 or folic acid. Healthy controls were recruited from among members of the Association of Blood Donors of Navarra (Spain). Controls with any history of neurological, psychiatric or major medical illness, memory complaints, scores below normal in the neuropsychological assessment or with MRI abnormalities were ruled out. The Ethics Committee for Medical Research of the University of Navarra approved the study, and all patients, or their legal representatives, and controls provided informed consent to participate in the study.

Motor assessments

The motor state in PD patients was assessed using the Hoehn and Yahr scale and the motor section of the unified Parkinson’s disease rating scale (UPDRS-III) in the “off” (minimum of 12 h without anti-parkinsonian medication) and “on” states. Drug intake was recorded and dopaminergic treatment calculated in levodopa equivalents (Table 1).
Table 1

General features of the study groups

 

Control (n = 20)

PD (n = 68)

PDCN (n = 21)

PD-MCI (n = 28)

PDD (n = 19)

Age (years), mean (SD)

67.9 (3.1)

70.6 (6.4)

67 (7.1)

71.5 (3.8)b

73.1 (7.1)a,b

Male gender, n (%)

11 (55)

37(54.4)

15 (71.4)

14 (50)

8 (42.1)

Disease evolution (years)

13,6 (5.1)

12.4 (3.8)

14.1 (6)

14.3 (5.1)

UPDRS III “on”, mean (SD)

20.8 (10.6)

16.4 (7.1)

17.7 (9.1)

30.8 (10.2) c,e

UPDRS III “off”, mean (SD)

37.9 (12.4)

32.3 (8.4)

33.2 (13.3)

49.4 (10.3) b,d

Levodopa equivalents (mg/day), mean (SD)

1147 (585.7)

1062 (347.2)

1249 (700.8)

1088 (616.5)

GDS score, mean (SD)

4.4 (4.1)

9.9 (5.2)a

7.8 (5.2)

9.9 (4.9)a

12.8 (5.9)a,b

Hallucinations, n (%)

18 (26.5)

2 (9.5)

5 (17.8)

11 (57.9)c,e

Hoehn and Yahr scale score, mean (SD)

3 (0.8)

2.6 (0.6)

2.9 (0.7)

3.7 (0.7) c,e

Education (years), mean (SD)

9.8 (3)

10.2 (3.2)

11.7 (3.6)

9.9 (3.1)

9 (2.3)

ap < 0.001 vs. control group

bp < 0.05 vs. PDCN

cp < 0.001 vs. PDCN

dp < 0.05 vs. PD-MCI

ep < 0.001 vs. PD-MCI

Neuropsychological assessment

Global cognitive function was evaluated with the mini-mental state examination (MMSE) [24]. The Interview for Deterioration in Daily Living in Dementia (IDDD) scale [25] was used to assess functional independence. Depression was rated using the Geriatric Depression Rating Scale (GDS) of Yesavage et al. [26]. Different cognitive domains (verbal and visual memory, attention and executive function, language and visuospatial function) were evaluated using a battery of neuropsychological tests [27]. Memory was assessed using the Free and Cue Selective Reminding test of Buschke [28], the Cerad word list, and the delayed recall of two simple figures (Massachusetts General Hospital, Boston). Other tests used were the Raven’s Progressive Matrices, semantic (animals) and phonetic (words starting with “p”) verbal fluency [29], Trail Making Test parts A and B, the Stroop test and Digit Span Forward and Backwards task for attention and executive functions. The Boston naming test and verbal fluency were evaluated for language, and the copying of two simple figures and the two intersecting pentagons of the MMSE were used for testing visuospatial function. All tests were applied by two members of the team to control subjects and patients under treatment, and were used alongside the diagnostic criteria to diagnose PD patients as being cognitively normal, as having MCI or as having dementia.

Criteria for diagnosing cognitive status

The clinical diagnostic criteria for dementia in PD [30] were applied to diagnose dementia in the present study. MCI was diagnosed in nondemented patients when the following two features were present: (1) cognitive decline was reported by either the patient or informant, or observed by the neurologist, but the decline did not interfere significantly with the functional independence of the patient; (2) the patient scored more than 1.5 standard deviations below control values in at least two tests in the neuropsychological battery, either within a single cognitive domain or across different cognitive domains [31]. Values used to determine test score deviations in PD patients were taken from a sample of 20 age- and education-matched healthy control subjects. Individual neuropsychological test scores were transformed into Z-scores using the mean and standard deviation of the control sample according to the following formula: (test score − median score from control sample)/standard deviation from control sample. Single-domain PD-MCI was diagnosed when abnormalities (the two abnormal tests) were present in a single cognitive domain. Multiple-domain PD-MCI was diagnosed when abnormalities were present in at least one test in two or more cognitive domains. In addition, to correlate the cognitive state with FDG uptake, the Z-score for the different domains was calculated from the average of the Z-scores of the tests assessing each domain. Patients not fulfilling criteria for MCI or dementia were considered to have cognitively normal PD.

FDG PET

Image data acquisition

Patients were studied in the “on” pharmacological condition (i.e. under the effect of their usual anti-parkinsonian dopaminergic medication). Central nervous system depressant drugs such as benzodiazepines, neuroleptics or antidepressive treatments were withdrawn according to their pharmacological kinetics. Additionally, subjects fasted overnight before PET scanning. Before injection of the radiopharmaceutical, blood glucose was checked and was <120 mg/dL in all patients. After a few minutes of rest in silence and with dimmed lighting, 18F-FDG (370 MBq) was injected intravenously, and subjects were required to rest for 40 min in the supine position on the PET scanner bed with their eyes closed. Then, 74 planes (128 × 128 matrix) were acquired with a voxel size of 2.06 × 2.06 × 2.06 mm during a 20-min scan using a Siemens ECAT EXAT HR+ scanner (Siemens, Knoxville, TN). A transmission scan in 3D mode for attenuation correction was performed at the end of the acquisition period [32]. Images were reconstructed by means of a filtered back-projection method using ECAT software (version 7.2; Siemens).

Data analysis

Data were processed using statistical parametric mapping (SPM8) software (Wellcome Department of Neurology, London, UK) implemented in Matlab 7.13 (MathWorks Inc. Sherborn, MA). First, we created a customized FDG PET template using data from the control sample (n = 20). For this purpose, all control subjects were scanned with a 1.5-T Siemens Symphony system using a three-dimensional T1-weighted gradient-echo sequence (acquisition parameters: coronal acquisition, TR/TE/TI 1,900/3.36/1,100 ms, flip angle 15°, 144 slices, FOV 187 × 250 mm, matrix 192 × 256, voxel size 0.98 × 1.6 × 0.98 mm). Thus, control FDG PET images were coregistered with their corresponding MR images. MR images were segmented using the SPM8 segmentation tool [33] in MATLAB 7.0. Grey matter (GM) and white matter templates were generated from the entire image dataset using the DARTEL technique [34]. After an initial affine registration of the GM DARTEL templates to the tissue probability maps in Montreal Neurological Institute (MNI) space [35], nonlinear warping of the GM images was performed to normalize them onto the MNI space. The spatial normalization parameters of each MR image were then applied to each corresponding coregistered FDG PET image. The FDG PET template was obtained by averaging the spatially normalized PET images and smoothing using an isotropic gaussian filter with a full-width at half-maximum of 8 mm.

All FDG PET images were spatially normalized into a standard stereotactic MNI space using the customized FDG template. For every spatially normalized PET image, voxel values were normalized to pons activity (becquerels per centimetre cubed) using the pons volume of interest (Nifti format) from WFU PickAtlas v3.0 [12, 3638]. Finally, the resulting PET scans were smoothed with an isotropic gaussian filter with a full-width at half-maximum of 8 mm. Changes in metabolism were assessed by analysis of the preprocessed images using one-way analysis of variance. Age and GDS score were included as covariates for the metabolism comparison between controls and patients, while for the metabolism comparison between the different groups of patients, age, UPDRS-III and GDS scores were included as covariates. Significance was set at p < 0.05 and corrected for multiple comparisons, i.e. a false discovery rate (FDR) with a cluster size of >20 voxels. In the comparison between PD-MCI and PDCN patients in which less significant differences would be expected, significance was set to p < 0.001 uncorrected, similar to previous works in the field [18, 19]. The correlation between the MMSE, UPDRS-III, GDS scores and Z-scores of the different cognitive domains and glucose metabolism was assessed in PD patients using regression analysis and significance set at p < 0.001 uncorrected, with a cluster size of >20 voxels.

The coordinates of the voxel peaks were transformed into Talairach space using the mni2tal program by Dr. M. Brett (http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach) and their anatomical locations were found using Talairach Daemon Client [39].

Statistics

Differences in the demographic and clinical characteristics between the PD groups and controls were analysed using Fisher’s exact test in cases of categorical variables, analysis of variance with post-hoc Bonferroni’s multiple comparison in cases of continuous, normally distributed variables, and the Kruskal-Wallis and Mann-Whitney U tests for continuous, nonparametric variables. The normality of the distributions of clinical and demographic variables was assessed using the Kolmogorov-Smirnov test. A value of p < 0.05 was considered to indicate statistical significance.

Results

Clinical data

The subjects included 20 controls and 68 PD patients (21 PDCN, 28 PD-MCI, and 19 PDD). The demographic and clinical characteristics of all groups are summarized in Table 1. With the exception of a higher GDS score, PD patients did not differ from control subjects. PDD patients were older than controls and PDCN patients. They also had higher GDS scores than PDCN patients, and had more severe parkinsonism (UPDRS and Hoehn and Yahr scores) and more hallucinations than PDCN and PD-MCI patients. The PD-MCI patients were older than PDCN patients, with no other differences in clinical features.

Compared with the controls and PDCN patients, PDD patients had poorer scores in all neuropsychological tests, while with respect to PD-MCI patients, they had poorer scores in all but the recall of figures and word delayed recall tests (Supplementary Table 1). PD-MCI patients in turn had lower scores than PDCN patients for all tests, with the exception of the Buschke and the copying of simple figures and intersecting pentagon tests. The cognitive domains affected in PD-MCI patients were as follows: three patients (10.7 %) had only the executive domain affected; 14 patients (50 %) had two domains affected (executive and memory in nine patients, executive and visuospatial in four patients, executive and language in one patient); six patients (21.4 %) had three domains affected (executive, memory and language in five patients, and executive, visuospatial and memory in one patient); and five patients (17.4 %) had four domains affected. No differences were found between controls and PDCN patients (Supplementary Table 1).

Regional differences in FDG PET

Comparison between PD groups

PDD patients had extensive bilateral areas of reduced FDG uptake in the frontal, parietal, occipital and temporal lobes and in the posterior cingulate cortex compared with PDCN patients (Fig. 1A). PDD patients had a lower metabolism mainly in posterior brain areas (parietal, occipital and temporal lobes) than PD-MCI patients, and also, albeit to a lesser extent, in the right frontal lobe (Fig. 1B; Supplementary Table 2). Compared with PDCN patients, PD-MCI patients did not exhibit regions of reduced metabolism. However, using a relatively lower conservative threshold (p < 0.001 uncorrected), PD-MCI patients showed hypometabolism that was mainly localized in the left frontal lobe and to a lesser extent in the left parietal lobe (Fig. 1C; Supplementary Table 2). PDCN patients did not show reduced FDG uptake in any region compared with PDD and PD-MCI patients. Likewise, PD-MCI patients did not show reduced FDG uptake in any region compared with PDD patients.
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-012-2198-5/MediaObjects/259_2012_2198_Fig1_HTML.gif
Fig. 1

Regions with reduced metabolism comparing PDD, PD-MCI and PDCN patients: A PDD<PDCN, B PDD<PD-MCI, C PD-MCI<PDCN (p < 0.05 FDR corrected for A and B; p < 0.001 uncorrected for C; age, GDS and UPDRS III score as covariates in all comparisons)

Comparison between PD groups and controls

As expected, PDD patients showed an extensive bilateral reduction in FDG uptake in the frontal, parietal, occipital and temporal lobes, in the anterior cingulate cortex, and in the caudate nucleus compared with controls (Fig. 2A; Supplementary Table 3). In PD-MCI patients, more localized hypometabolic areas were identified in the parietal (mainly in the angular gyrus) and occipital lobes, and to a lesser extent in the frontal and temporal lobes (Fig. 2B; Supplementary Table 3). PDCN patients did not show hypometabolic areas compared with controls. No regions of reduced metabolism were identified in the control subjects compared with the PD patients.
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-012-2198-5/MediaObjects/259_2012_2198_Fig2_HTML.gif
Fig. 2

Regions with reduced metabolism comparing PDD patients and PD-MCI patients with respect to control subjects: A PDD<controls, B PD-MCI<controls (p < 0.05 FDR corrected; age and GDS score as covariates)

Correlation between cerebral metabolism and cognitive state in PD patients

A positive correlation between FDG uptake and MMSE score in all PD patients was observed for uptake in the parietal, occipital, temporal and frontal lobes, and in the anterior cingulate cortex and caudate nucleus using GDS and UPDRS-III scores and age as nuisance variables (Fig. 3). In addition, there were positive correlations between the Z-score of cognitive domains and FDG uptake as follows: executive function in the parietal, frontal and occipitotemporal junction; memory with temporal and parietal regions; visuospatial function with posterior areas (occipitoparietal and temporal) uptake; and language with anterior areas mainly the frontal lobe (Fig. 4). No correlation between the GDS score and FDG uptake was observed.
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-012-2198-5/MediaObjects/259_2012_2198_Fig3_HTML.gif
Fig. 3

Positive correlations between MMSE score and FDG uptake in all PD patients (p < 0.001 uncorrected; age, GDS and UPDRS-III as covariates)

https://static-content.springer.com/image/art%3A10.1007%2Fs00259-012-2198-5/MediaObjects/259_2012_2198_Fig4_HTML.gif
Fig. 4

Positive correlations between the Z-score of cognitive domains altered in PD patients and FDG uptake

Regions with hypermetabolism and clinical correlation

Compared with control subjects, PD patients exhibited increased metabolism in the putamen, thalamus and cerebellum and in the motor cortical (paracentral gyrus) areas bilaterally, but there were no differences among the PD groups (Fig. 5A–C). Moreover, FDG uptake in cortical areas was positively correlated with the UPDRS-III score but there was no correlation with the MMSE score (Fig. 5D).
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-012-2198-5/MediaObjects/259_2012_2198_Fig5_HTML.gif
Fig. 5

Regional increases in metabolism comparing PDD, PD-MCI and PDCN patients with respect to control subjects: A PDCN>controls; B PD-MCI>controls; C PDD>controls (p < 0.05 FDR corrected; age and GDS score as covariates). D Positive correlation between UPDRS III score and FDG uptake (p < 0.001 uncorrected)

Discussion

We compared cerebral FDG uptake in PDCN, PD-MCI and PDD patients, and control subjects. A major finding was that with respect to PDCN patients, PD-MCI patients showed a reduction in metabolism that predominated in the frontal lobe and to a lesser extent in the parietal lobe. In contrast, hypometabolism in PDD patients compared with PD-MCI patients was mainly located in posterior brain regions (parietal, occipital and posterior temporal areas) and to a lesser extent in the frontal lobe. We also found that, compared with controls, PDD and PD-MCI patients shared a common pattern of reduced metabolism in the parietal and occipital lobes, and to a lesser extent in the frontal and temporal lobes. However, in PDD patients the cortical hypometabolism was more widespread affecting wider cortical and subcortical areas compared to that seen in control subjects. Taken together, our data suggest that dementia in PD is characterized by a more intense and widespread cerebral hypometabolism than MCI in PD patients and that this hypometabolism predominates in posterior cortical areas. We also found that deficits in different cognitive domains in PD were associated with reduced cerebral metabolism involving different brain regions. Thus, our results further define aspects of cerebral metabolism associated with MCI and dementia in PD.

Most previous studies have focused on comparing the cerebral metabolism of PD patients with that of control subjects [1317, 20], but putative distinctive features among the different cognitive states in PD have been poorly elucidated. While a reduction in FDG uptake in the posterior and frontal cortices of PD-MCI patients compared with PDCN patients has been identified in some studies [18, 19, 21], no study has been carried out to compare differences between PD-MCI and PDD patients. A cross-sectional study [11] demonstrated that dementia differs from MCI by both a generalized failure in executive function and the addition of “posterior cortical dysfunction” (naming and clock copy tests). Further to this, the Cambridge longitudinal study showed that PD patients with deficits in tasks revealing a predominant temporal and parietal lobe (“posterior cortical”) dysfunction have a higher risk of dementia than those with only a frontal executive dysfunction [10]. In keeping with this, a recent longitudinal study showed that patients who develop dementia after 3.9 years of follow-up have a low performance in delayed visual reproduction learning and a reduced FDG uptake in the visual association and posterior cingulate cortex at baseline compared to controls [12]. In addition, hypometabolism in the parietooccipitotemporal and medial temporal brain has been correlated with visuospatial and mnemonic functioning in PD patients [40]. Here we report that PDD patients have reduced FDG uptake compared with PD-MCI patients mainly in the parietal, occipital and posterior temporal areas, supporting the notion that a cognitive deficit based on posterior cortical function (i.e. visuospatial and memory) is the main difference between the two cognitive states.

We also found that PD-MCI patients exhibited a reduction in metabolism in the frontal and parietal lobes compared with PDCN patients. Previous FDG PET studies have shown similar results [19] or more extensive hypometabolism in the posterior cortex [18, 21]. The discrepancy is probably due to the different criteria used for the diagnosis of MCI. For example, when a clinical scale (the clinical dementia rating scale) for staging the severity of dementia was used and MCI was diagnosed for a score of 0.5, corresponding to very mild dementia, patients with MCI had more extensive hypometabolism in the posterior cortex than PDCN patients [18]. In a similar manner to Huang et al. [19], we used a more stringent diagnostic criterion based on 1.5 standard deviations in neuropsychological test scores with respect to control subjects’ scores. This approach probably resulted in the classification of PD-MCI patients with lower levels of cognitive deficits. Consequently, these data also reinforce the association between more severe cognitive deficits and a higher level of posterior cortical hypometabolism.

It should be noted that the diagnostic criteria for MCI are still under development, and MCI in PD requires further definition [31]. Currently, MCI in PD includes different cognitive abnormalities depending upon the number and type of cognitive domains affected. As yet, whether a given subtype of MCI in patients with PD might confer a higher risk of developing dementia has not been clarified. A single prospective study in a small number of patients showed that single and multiple-domain non-amnestic forms of MCI were more associated with the development of dementia 4 years later [7]. Although ongoing longitudinal studies will eventually clarify this point, it is probable that PD patients with multiple-domain MCI have a more severe or advanced cognitive decline than those with single-domain forms, and therefore are at a higher risk of developing dementia. In this sense, previous cerebral FDG PET studies did not highlight differences between PD patients with single-domain MCI and those with normal cognition. In contrast [19, 21], patients with multiple-domain MCI showed reduced metabolism in the frontal and parietal lobes, and less consistently in the temporal lobe than PDCN patients [18, 19, 21]. In the present study, all but three patients with PD-MCI had a multiple-domain type of MCI, giving more consistency to the differences found in cerebral metabolism underlying each cognitive state in patients with PD [10, 11].

We have also demonstrated a relationship between metabolic findings and cognitive state given the correlations observed between the severity of the global and cognitive domain deficits measured by MMSE and the corresponding Z-scores, respectively, and regions of reduced metabolism in PD patients after correcting for age, depression (GDS score) and motor severity (UPDRS-III). We found that executive function mainly correlated with metabolism in the parietooccipitotemporal junction and frontal lobe, while memory correlated with metabolism in the temporal and parietal regions, and visuospatial function correlated with posterior areas (occipitoparietal and temporal). In contrast, language was correlated with metabolism in anterior (mainly frontal) areas. These results are in keeping with those of previous studies in PD patients showing that bilateral hypometabolism in the frontal and parietal regions and in the parietooccipitotemporal and medial temporal brain correlate with executive dysfunction [4145] and visuospatial and mnemonic functioning, respectively [40]. Although partially limited by the lack of data in normal individuals, the correlations reported here also indicate that the higher posterior hypometabolism in PDD with respect to PD-MCI patients is related to a worsening of executive dysfunction and to the addition of visuospatial and memory deficits.

This was a cross-sectional study seeking to identify metabolic differences between PDD and PD-MCI patients. Considering that MCI forms part of the cognitive decline preceding the development of dementia, only large longitudinal studies will allow a better definition of clinical MCI and of the changes that characterize the transition from MCI to dementia. Our findings therefore require confirmation in prospective studies with a larger number of patients. Nevertheless, we have studied a large number of PD patients considered to be at high risk of developing dementia [23]. Indeed, it has been reported that the development of substantial abnormalities in the defined cognitive network expression in PD takes places by the end of the first decade following clinical onset [14]. Patients in our study were studied while under dopaminergic treatment, but the doses were not different between the different groups and it is known that dopaminergic drugs do not have a significant impact on the pattern of cognitive expression detected in FDG PET [14, 41]. Moreover, there were no differences in other treatments that could have interfered with the results; for example, only three PDD patients [46] were receiving treatment with cholinesterase inhibitors, which would have increased the FDG uptake [46].

The disease duration was not different among the groups, but one limitation of this study is that, due to the natural course of the disease and the age at which dementia most frequently occurs, PDD patients exhibited a greater degree of motor disability than patients in the other groups, were older on average and had higher depression scale scores than PDCN patients and controls [16, 17, 4750]. It should be noted that the metabolic pattern related to motor aspects is different from that related to cognition [14]. In addition, we did not identify any correlation between depression score and cerebral metabolism. However, the data were corrected for age, motor severity and depression, and therefore we believe the results would not have been significantly affected by these factors. Further to this, the validity of our data is reinforced by the fact that the metabolic patterns in the PD groups studied here in comparison with control subjects are similar to those reported in the literature [1318, 51]. Moreover, the subcortical (putamen and thalamus) and motor cortical (paracentral gyrus) hypermetabolic areas encountered in all groups of PD patients were also in keeping with the PD-related motor pattern [52, 53], and actually correlated positively with the UPDRS-III score. In addition, we used a more advanced version of the software for image analysis (i.e. SPM8) and corrected our data more thoroughly than in previous studies (FDR corrected at p < 0.05). Admittedly, as a result of the use of this conservative threshold, PD-MCI patients did not show regions of reduced metabolism compared with PDCN patients. Thus, differences between PD-MCI and PDCN patients were obtained only with the less-conservative analysis (uncorrected p < 0.001). However, the metabolic differences previously reported between these two groups of patients (PD-MCI and PDCN) have also been obtained with uncorrected data [18, 19, 21].

Conclusion

Our study demonstrated that PD-MCI patients exhibited hypometabolism (decreased FDG uptake) in numerous cerebral areas compared with controls, and in the frontal and parietal regions compared with PDCN patients. In contrast, dementia in PD was characterized by a more expansive cerebral hypometabolism than MCI, with predominance in the posterior cortical areas. The reduction in FDG uptake in these posterior areas correlated with poorer outcomes in visuospatial, memory and executive functions. Taken together, these results indicate that dementia in PD is associated with a worsening of executive dysfunction along with an impairment of visuospatial and memory function.

Acknowledgments

This study was supported by a grant from the Government of Navarra (32/2007), by a grant from FIS (ISCIII), and CIBERNED, Spain. We thank Ainara Estanga for her critical review of the article.

Conflicts of interest

Maria C. Rodriguez-Oroz is on the advisory board of UCB Spain. She has received payment for lectures, as well as travel and accommodation to attend scientific meetings from GlaxoSmithKline, UCB, Lundbeck and Medtronic. She has received research funding from national and regional government bodies in Spain. Jose Obeso has served previously on the Advisory Board of GSK (UK), and received honorarium for lectures given at meetings organized by GSK (Spain), Lundbeck-TEVA and UCB. Grants/Research: Funding from Spanish Science and Education Ministry and European Union (REPLACES). The other authors have no conflicts of interest to report concerning the research dealt with in this manuscript.

Supplementary material

259_2012_2198_MOESM1_ESM.docx (87 kb)
ESM 1(DOCX 87 kb)
259_2012_2198_MOESM2_ESM.doc (68 kb)
ESM 2(DOC 68 kb)
259_2012_2198_MOESM3_ESM.doc (100 kb)
ESM 3(DOC 99 kb)

Copyright information

© Springer-Verlag 2012