Introduction
Cognitive impairment is a common and disabling non-motor symptom of Parkinson's disease (PD). Cognitive deficits may be present in up to 24% of PD patients by the time of diagnosis, and this rate reaches over 80% in the long-term [1]. Although dysexecutive syndrome has long been considered the main hallmark of cognitive decline in PD, deficits in visuospatial, memory and attention functions may be also present. The rate and pattern of these deficits vary greatly among PD patients, and different biological mechanisms appear to play a role [2]. In this regard, the dual syndrome hypothesis was recently proposed, suggesting two facets of cognitive decline in PD: (i) changes in frontostriatal dopaminergic transmission, leading to deficits in planning, working memory, response inhibition and attentional control; and (ii) posterior cortical Lewy body pathology and secondary cholinergic loss, affecting visuospatial, mnemonic and semantic functions and leading to dementia [3].
It is possible to assess the state of the frontostriatal circuitry by imaging the striatal dopamine transporter (DAT) with [123I]FP-CIT SPECT. In this scan, dopamine depletion can first be observed in the putamen, which affects the motor loop, whilst dopamine depletion in the caudate usually occurs in later stages, affecting two well-defined frontostriatal loops: the cognitive and the limbic loops [4]. Therefore, the integrity of this latter pathway is essential to correct cognitive functioning, and a large number of studies have in fact found a correlation between cognitive performance (including executive and working memory tasks) and caudate dopamine levels in PD [5].
Several genetic loci have also been proposed as risk factors for cognitive decline in PD [6]. Some of these genetic loci have been linked to the dopaminergic pathway, such as the Val158Met polymorphism in the catechol-O-methyltransferase gene (COMT). This gene encodes the COMT enzyme, which contributes to the degradation of cortical dopamine. Met carriers show low enzyme activity in comparison to Val carriers. This genotype therefore modulates dopamine levels in the frontostriatal network, and in turn, executive function performance [7].
Other genetic loci have been linked to both the development of dementia and performance in tasks mediated by other non-dopaminergic mechanisms. Specifically, the apolipoprotein E gene (APOE) ε4 allele has been associated with an increased risk of dementia [8] and deficits in memory and verbal fluency [7, 9]. Furthermore, the microtubule-associated protein tau gene (MAPT) H1 haplotype has been linked with dementia [10, 11] and visuospatial deficits [7], although recent results have been controversial [9]. It is not yet known what role these two loci play in dopaminergic degeneration, and this needs to be addressed.
Another genetic locus of interest is the rs356219 polymorphism, located in the 3’UTR of the α-synuclein gene (SNCA). Mutations and repetitions in SNCA lead to a familial form of PD with prominent cognitive impairment and dementia. The rs356219 polymorphism has been linked to PD pathogenesis [12], although its role in cognition is far from clear. Its relationship with dopaminergic imaging has besides not yet been studied. Lastly, the glucocerebrosidase gene (GBA) is the most common genetic factor that has yet been identified for developing PD [13]. PD patients with mutations in GBA have earlier disease onset, and are at a higher risk of developing visual hallucinations, cognitive impairment and dementia [14]. Recent studies suggest that GBA-carriers have a more severe phenotype, with quicker disease progression [15]. As is the case with other genes, very little is known about the relationship between GBA and dopaminergic imaging.
Although other groups are thoroughly investigating the relationship between these genes and domain-specific neuropsychological tasks, using large cohorts of PD patients, not enough studies have yet been conducted to evaluate how these genetic loci contribute to dopaminergic degeneration through imaging. Furthermore, recent studies have produced contradictory results concerning the role that some loci, such as the H1 haplotype in MAPT, play in PD dementia. This study aims to investigate the role that these genes play in striatal dopaminergic denervation and PD dementia. To this end, we collected [123I]FP-CIT SPECT images and long-term clinical data on the developement of dementia, and genotyped APOE, MAPT, COMT and SNCA risk polymorphisms, as well as GBA screening in a cohort of 298 PD patients from our centre.
Materials and methods
Subjects
A total of 298 PD patients were included in this study (age at onset 55 ± 13 years, 60% males), recruited from the Movement Disorders Unit at Virgen del Rocío Hospital (Seville, Spain). The diagnosis of PD was made using the UK Parkinson's Disease Society Brain Bank clinical diagnostic criteria. All patients underwent [123I]FP-CIT SPECT (mean disease duration 6 ± 6 years, median Hoehn and Yahr 2 [1.5, 2.5]) and were clinically monitored during the course of the disease with periodic visits to our clinic. At SPECT exam, 17% of the patients had no medication, 17% on levodopa, 18% on dopaminergic agonists and 48% on both. The median levodopa equivalent daily dose (LEDD) for those under treatment was of 596 [300, 1063] mg/day. The influence of the genetic factors on the dopamine-mediated syndrome was investigated through the association between the genotypes and striatal DAT, whereas the influence on the posterior syndrome was investigated through the association between the genotypes and the onset of dementia. We identified patients who met diagnostic criteria for possible or probable dementia [16] in a long-term review of the medical records (mean disease duration at time of review: 11 years). The diagnosis of dementia was ascertained by a variety of screening tools including a medical interview to the patient and caregiver, or scores in standard scales such as Mini Mental State Examination (MMSE ≤ 24), and Parkinson’s disease Dementia Short Screen (PDD-SS ≤ 11) [17]. The disease duration at the visit when the patient met the criteria was used to perform survival analyses of the progression to dementia. The core features of these criteria include the presence of deficits (social, occupational, or personal care) impairing daily life and the presence of deficits in one or more cognitive domains such as attention, executive, visuo-spatial, memory and language, and behavioral symptoms. The diagnosis of possible dementia included an atypical profile of cognitive impairment in one or more domains such as prominent or receptive-type (fluent) aphasia, or pure storage-failure type amnesia. The diagnosis of probable dementia included impairment in at least two domains such as attention, executive and/or visuo-spatial functions, and free recall memory. Probable dementia diagnosis was also reinforced by the presence of behavioral symptoms such as apathy, changes in personality and mood, hallucinations and delusions, and excessive daytime sleepiness. All subjects provided informed written consent, and the Hospital Virgen del Rocío ethics committee approved this study.
Genetics
Genomic DNA was extracted from peripheral blood samples using the standard methods. All patients underwent genotyping for rs429358 and rs7412 (APOE ε2, ε3, and ε4), rs9468 (MAPT H1 vs. H2), rs4680 (COMT Val158Met), and rs356219 (SNCA). Genotyping was performed with TaqMan SNP Genotyping Assay in a LightCycler480 (Roche Applied Science), and the genotyping success rate was over 98%. All patients were also screened for variants in the entire gene GBA, using both high-resolution melting analysis for all exons and direct DNA resequencing for those samples showing abnormal melting profiles. Of the 298 subjects, we identified 62 GBA variants in a total of 48 carriers. Identified variants were classified as potential deleterious (n = 35) or potential benign (n = 27) (see S1 Table) based on in silico analyses using the bioinformatic tools Grantham score, Polyphen-2, MutPred v1.2, and Mutation Taster. PD patients were classified into the group of “benign” if all the carried variants were potentially benign (n = 17), and classified into the group of "deleterious" if at least one of the carried variants was potentially deleterious (n = 31). A more detailed description about the sequencing procedure, variants and criteria for assessing pathogenicity can be found in a recent work from our group [18].
SPECT imaging
The acquisition procedure and SPECT reconstruction can be found in a previous report [19]. SPECT images were processed with standard procedures in SPM8 using a homemade [123I]FP-CIT template (http://www.nitrc.org/projects/spmtemplates). Quantitative analyses were based on volumes of interest in the striatum manually drawn by expert nuclear-medicine specialists (https://www.nitrc.org/projects/striatalvoimap) following established methodology [20]. A volume in the occipital cortex was used as a reference region and, for each patient, [123I]FP-CIT binding potential (BP) for posterior putamen and head of caudate was calculated. The BP was expressed as the percentage of age-expected binding with respect to 184 normal scans (age range 18–90 years) [21], and since laterality can affect the statistics at the group level, the comparisons were made for the more affected side.
Statistical analysis
We investigated the role of APOE ε2 and ε4 alleles; MAPT H1 and H2 haplotypes; COMT Met allele; SNCA G allele; and deleterious and benign variants in GBA. Based on the previous reported genotypes of risk in the literature, the comparisons of interest in this study were: APOE: ε2+ vs. (ε2-, ε4-) and ε4+ vs (ε2-, ε4-); MAPT: H1/H1 vs. H2; COMT: Met/Met vs. Val; SNCA: G vs. A; GBA: GBA+ vs. non carriers. Other potential genetic models were also explored (allelic, dominant and recessive), but for the sake of simplicity only the contrasts of interest and/or the strongest associations are presented. We performed separate linear regression analyses for each gene and imaging variable to study their potential interaction with PLINK (http://pngu.mgh.harvard.edu/~purcell/plink/). We entered the quantitative imaging variables as dependent variables, the genetic factors as independent variables and the regression coefficients were calculated as a measure of effect size. The potential confounding factors sex, age and disease duration at time of scan were also introduced as covariates. Although previous studies have shown that dopaminergic medications do not alter DAT imaging [22, 23], we verified with exploratory analyses that medication was not confounding striatal uptake. The development of dementia was examined with survival analyses through Cox regression. For this analysis, we used as event variable the presence of dementia (yes/no), and as time variable the disease duration at dementia onset for the positive cases (yes) and the disease duration at the review of the records for the negative cases (no). We performed a separate regression analysis for each gene and Hazard ratios (HR) for each risk genotype were calculated adjusting for sex and age at disease onset as potential confounding factors for dementia. Analyses were done using IBM SPSS Statistics 22.0 and the statistical threshold for significance was set to P < 0.05. Given that this is an exploratory study, we did not apply multiple testing penalization.
Discussion
In this study, we found that striatal DAT availability levels in PD were influenced by APOE ε2 allele, COMT Val158Met, SNCA rs356219 and deleterious variants in GBA, whereas the development of dementia was influenced by the APOE ε4 allele and also by deleterious variants in GBA. Our results therefore suggest that APOE2, COMT and SNCA may be related to dopaminergic degeneration, while APOE4 may be related to other, non-dopaminergic degeneration mechanisms, and GBA may be implicated in both. Our findings support the dichotomy of the dual-syndrome hypothesis and provide new insights into the dissociation of the genetic factors which contribute to cognitive decline in PD.
The role of APOE2 in PD is controversial, some studies found a higher ratio of ε2 alleles in PD patients than in controls, although other studies do not share this finding [22, 24]. Similarly, the 'G' allele in SNCA polymorphism has been found overrepresented in PD [25]. There are no previous studies investigating the relationship between these genetic factors with striatal DAT, but our results suggest that both APOE2 and SNCA could have a negative effect on the dopaminergic pathway. We also observed a trend for reduced caudate BP fot H2/H2 carriers, although this trend should be further supported by other data sets since, to the best of our knowledge, no prior data on this relationship have been reported. Hence, our results suggest that APOE2 and SNCA may be implicated in PD pathogenesis and lead to a faster frontoexecutive impairment. No association with dementia onset was found, which is consistent with previuos data [11], and indicate that these loci do not play any role in the posterior cortical syndrome. We also found increased levels of caudate DAT in Met/Met carriers of COMT polymorphism. This is consistent with a 18F-DOPA PET study, which found higher presynaptic dopamine levels in frontal regions in Met/Met [26], and in controversy with a recent study that found higher levels of striatal FP-CIT BP in Val/Val [27]. However, this last result could be a false positive due to a small sample size (40 subjects in total, and only 3 Met/Met carriers).
We found APOE4 to be associated with a faster progression to dementia, but no such relationship was found for MAPT H1/H1. The observed effect size for APOE ε4 (HR = 1.90) was modest in comparison to that seen in AD but consistent with an existing meta-analysis in PD, which also suggests this allele has a moderate effect on PD dementia (OR = 1.74; 95% CI 1.36–2.23) [8]. Consistent with our data, Mata and colleagues' recent study noted the detrimental effect of APOE ε4 on cognition in PD, and no effect for MAPT H1/H1 [9]. Also, a previous study of PD in Spain discarded a relationship between MAPT H1/H1 and dementia [28]. On the other hand, a recent 10-year follow-up for the CamPaIGN cohort found a link between MAPT H1/H1 and dementia, and no link for APOE ε4 [11]. However, this discrepancy concerning APOE ε4 could arise from a lack of power, since only 38 demented PD patients and 35 non-demented PD patients were evaluated, and ε4 frequency was higher in the case of the demented (37% vs. 26%), although it did not reach a significant level.
Interestingly, we found that deleterious variants in GBA were associated to both reduced striatal BP and faster progression to dementia, possibly indicating that these variants play a role in both dopaminergic and non-dopaminergic degeneration processes. There are very few studies on dopaminergic imaging for PD GBA carriers, and these are limited to only a few cases;[29, 30] as of yet, no solid conclusions have therefore been drawn on the relationship between GBA and the dopaminergic system. Consistent with our observations, a recent study found a reduced glucocerebrosidase enzymatic activity in the substantia nigra of GBA carriers [31]. Clinical studies also support our results, having observed greater motor and cognitive impairment in PD patients with deleterious GBA variants (e.g. L444P, N370S), including a higher risk of progressing to Hoehn and Yahr stage 3 and dementia [15, 32]. Moreover, a recent study found executive and visuospatial deficits in these carriers, supporting our view that GBA might have a double-edged role in both dopaminergic and non-dopaminergic degeneration [33]. Also importantly, despite our bioinformatic analyses classified the variant E326K as benign, there are recent data suggesting the deleterious effect of E326K variant, including lower glucocerebrocidase activity [34] and worse performance in executive and visuospatial tasks in these carriers [33]. However, our data do not support the negative role of this variant. There were 5 patients heterozygous for E326K and 4 patients with compound heterozygosis with other deleterious variants, and none of them had reduced DAT binding in comparison with analogous non-carriers nor displayed signs of dementia after a mean follow-up of 14 years. We acknowledge that our sample size is limited to make conclusions about this variant but our observations on these 9 patients do not indicate that this variant should be classified as deleterious. Lastly, we also observed that the risk of GBA deleterious variants carriers to develop dementia was increased in combination with APOE4 allele. However, this is just an observation and should be interpreted with caution since we only had 6 patients having both risk genotypes (3 of them got demented, and two of them were L444P carriers). Indeed, the result is not significant due to the lack of power.
In summary, APOE2, COMT Met, 'G' allele at SNCA rs356219 and deleterious variants in GBA contribute to dopaminergic degeneration in PD. These loci may therefore contribute to frontostriatal dysfunction. APOE4 and variants in GBA contribute to the development of dementia, and are possibly related to other non-dopaminergic processes. Different genetic risk genotypes produce different outcomes of the dual syndromes of cognitive impariment in PD, and deleterious variants in GBA may play a double-edged role in both. We acknowledge that the lack of exhaustive clinical and neuropsychological assessments for dementia is a potential limitation in our study. However, this population-based study was designed to overcome limitations of sample size in genetic studies and provide reliable effect sizes. Futher research will be able to verify the findings of this discovery sample, and will allow for more convincing conclusions.