|Home | About | Journals | Submit | Contact Us | Français|
While mutations in glucocerebrosidase (GBA1) are associated with an increased risk for Parkinson disease (PD), it is important to establish whether such mutations are also a common risk factor for other Lewy body disorders.
To establish whether GBA1 mutations are a risk factor for dementia with Lewy bodies (DLB).
We compared genotype data on patients and controls from 11 centers. Data concerning demographics, age at onset, disease duration, and clinical and pathological features were collected when available. We conducted pooled analyses using logistic regression to investigate GBA1 mutation carrier status as predicting DLB or PD with dementia status, using common control subjects as a reference group. Random-effects meta-analyses were conducted to account for additional heterogeneity.
Eleven centers from sites around the world performing genotyping.
Seven hundred twenty-one cases met diagnostic criteria for DLB and 151 had PD with dementia. We compared these cases with 1962 controls from the same centers matched for age, sex, and ethnicity.
Frequency of GBA1 mutations in cases and controls.
We found a significant association between GBA1 mutation carrier status and DLB, with an odds ratio of 8.28 (95% CI, 4.78–14.88). The odds ratio for PD with dementia was 6.48 (95% CI, 2.53–15.37). The mean age at diagnosis of DLB was earlier in GBA1 mutation carriers than in noncarriers (63.5 vs 68.9 years; P<.001), with higher disease severity scores.
Mutations in GBA1 are a significant risk factor for DLB. GBA1 mutations likely play an even larger role in the genetic etiology of DLB than in PD, providing insight into the role of glucocerebrosidase in Lewy body disease.
Mutations in the gene encoding the lysosomal enzyme glucocerebrosidase (GBA1 [OMIM 606463]) are an important and common genetic risk factor for Parkinson disease (PD). Multiple individual studies1–8 and an international multicenter collaborative study9 of more than 5600 patients with PD and an equal number of controls demonstrated that among patients with PD, the odds ratio (OR) for carrying a mutation in GBA1 was greater than 5.00. Overall, patients with PD who carried GBA1 mutations had an earlier disease onset (mean, 4 years) and a higher incidence of cognitive changes.
These differences led to the suggestion that GBA1 mutation carriers might have more Lewy body (LB) disease, addressed in further studies by evaluating GBA1 in patients diagnosed as having other LB disorders with deposition of fibrillated α-synuclein in the brainstem or as cortical inclusions.10
Results from these studies varied considerably. Focusing on cases with pathologically confirmed dementia with LB (DLB), GBA1 mutations were first identified in 8 of 35 cases (23%)11 and then in 3 of 50 samples (6%).12 Another study, screening only for the mutations N370S and L444P in 57 patients with DLB, detected GBA1 alterations in 2 of 57 patients (3.5%) compared with 2 of 554 control subjects (0.4%) (P=.045),4 whereas Clark et al13 identified GBA1 mutations in 27 of 95 cases (28%) with LB disorders and 3.5% of control cases (P<.001). More recently, a multisite study14 of autopsy samples from subjects with different forms of dementia reported GBA1 mutations in 6 of 79 patients (7.6%) with pure DLB (OR, 7.6 [95% CI, 1.8–31.9]) and 8 of 222 subjects (3.6%) with LB disease and Alzheimer disease neuropathology (4.6 [1.2–17.6]). A clinical study15 evaluating the PD phenotype in a cohort of 225 patients from Spain concluded that GBA1 mutations strongly influence the course of PD with respect to the development of dementia.
The overall impression that GBA1 mutation carriers were more likely to have a pathological finding of LBs prompted this multicenter analysis of GBA1 mutations in DLB (defined according to criteria by McKeith et al10 [hereinafter referred to as the McKeith criteria]), similar to a previous international study of GBA1 mutations in PD.9 Because patients with the diagnosis of DLB are not as common as those with PD, we believed that a multicenter analysis was essential to collect adequate numbers of subjects to draw meaningful conclusions. Discovering the extent of neuropathology associated with GBA1 mutations could provide insight into disease pathogenesis. Also, ascertainment of the full phenotypic spectrum of GBA1-associated parkinsonism is of clinical importance because the disease prognosis and level of impairment encountered in PD and DLB differ significantly. For this reason, we also included for evaluation patients considered to have PD with dementia (PDD),16 collected from 6 of the participating sites.
Each center that participated in the previous collaborative study in PD9 and several additional PD or dementia research centers known to be performing GBA1 genotyping were contacted regarding their interest in sharing genotyping data on subjects with other LB disorders. Eleven centers had genotyped GBA1 in such samples, and all data contributed were assembled and analyzed at the National Institutes of Health (Table 1). Unlike the previous study,9 all participating centers were in North America, Europe, or Australia, and most of the subjects were reported to be of European descent. Written informed consent was obtained from all subjects under the supervision of their respective institutional review boards.
Each center was instructed to include only subjects who met the McKeith criteria10 for DLB or the Movement Disorders Task Force criteria16 for PDD. Subjects with other concurrent diagnoses were excluded. All participating centers were provided with a uniform spreadsheet for compiling the requested data and specific instructions regarding how to list demographic and genotype data. Each center was also asked to respond to specific questions regarding clinical and pathological findings, when available. The analysis included patient-level data from published and unpublished studies.
Mutation detection methods varied considerably among the participating centers. Previously, centers that had participated in the PD study9 had genotyped a standard panel of 12 DNA samples to confirm reliability in genotyping. In the present study, full sequencing of all exons of GBA1 was performed in 681 patient samples; screening for 5 to 10 specific variants, in 47 patients; and screening for only 2 to 4 mutations, in 144 patients (Table 1). Because most cases underwent sequencing of all exons, we primarily relied on genotypes established by sequencing but also included the other samples as confirmation. The analysis was therefore stratified on the basis of the genotyping method. Samples from subjects who underwent sequencing were compared with controls who underwent sequencing. All centers screened for the mutations L444P and N370S. The GBA1 variants E326K and T369M were evaluated separately. Synonymous variants and those in the upstream untranslated region were omitted from the analysis.
All available genotyping data for GBA1 in cases with the diagnosis of DLB or PDD and neurologically healthy controls from the 11 different sites were aggregated9 (Table 2 and Table 3). Pooled analyses were conducted using basic logistic regression, investigating GBA1 mutation carrier status as predicting DLB or PDD case status using a common control series for both disease outcomes. A full list of variants detected in cases and controls is listed in Table 4.
A priori analyses were to be stratified by mutation detection (sequencing of all GBA1 exons, screening for 5–10 specific mutations, or screening for ≤4 mutations), by sequencing center, and by whether the diagnosis was based on pathological or clinical assessments (for PDD, only clinical data were available). For all predefined subsets for which adequate model fit was achieved from logistic regression models, the Der-Simonian and Laird procedure for random-effects meta-analysis was conducted by combining summary statistics from the logistic regression models for each subset (Table 5 and Figure). Heterogeneity of effect variances (I2) were quantified in these models as well. Owing to sample size restraints for studies of rare mutations, meta-analyses of subgroups were conducted only for the larger, better-characterized DLB cohort. Means and frequencies of intermediate phenotypes were compared among cases with and without GBA1 mutations using standard 2-sided unpaired t tests to compare quantitative traits or z-based tests of proportions for discrete traits (Table 2 and Table 3). In addition, basic logistic regression was used to quantify the risk of DLB or PDD due to GBA1 variants E326K and T369M. We used additional logistic regression models adjusting for study site to generate further estimates of risk for DLB and PDD separately as a means of accounting for study heterogeneity while focusing only on samples that underwent full exon sequencing.
This study demonstrated a significant association between GBA1 mutation carrier status and DLB. Subjects with DLB were more than 8 times more likely to carry a GBA1 mutation than were controls (from logistic regression using all pooled samples based on raw count data, OR, 8.28 [95% CI, 4.78–14.88]) (Table 5 and Figure). This study differed from the previous collaborative study of PD in that very few (<1%) of the subjects included had Ashkenazi Jewish or Asian ancestry. In the regression analysis data from the PD study, the OR for non-Ashkenazi cases and controls with full GBA1 sequencing was reported as 6.51 (95% CI, 3.62–11.74).9 Because a subset of 151 cases who did not meet the McKeith criteria for DLB10 were considered to have PDD,16 we evaluated this subset separately, appreciating the limitation in sample size. We found that the association was slightly attenuated compared with DLB, with similar pooled analyses showing an OR of 6.48 (95%CI, 2.53–15.37) (Table 5) yet very similar to those reported for PD.9
To maximize the sample size available, logistic regression models adjusting for site were run for DLB and PDD. These models allowed us to use samples from all sites, including those that provided only cases without controls, which would be prohibitive factors in count-based analyses. These models showed slightly different results from the unadjusted pooled analyses or meta-analyses by site, likely owing to the inclusion of site adjustment and larger effective sample size in the models. When accounting for study center, these regression models showed that the OR estimate for DLB was 8.66 (95% CI, 4.49–17.71; P = 5.11 × 10−10) and the OR for PDD was 3.82 (95% CI, 1.45–10.05; P = 5.77 × 10−3).
We also attempted to address issues of recruitment and experimental biases by conducting additional analyses focusing only on the 4 centers where full exon sequencing was performed on cases and controls (Antwerp, Belgium; New York, New York; Bethesda, Maryland; and Sydney, Australia). When we analyzed data limited only to these 4 sites, the estimated OR for DLB was increased to 14.20 (95% CI, 4.75–61.12; P = 2.48 × 10−5). This OR, which is elevated compared with the values reported in Table 5, could be due to the increased ability of exonic sequencing to detect more mutations or to the attenuated sample size in this statistical model. This analysis of only fully sequenced cases and controls from the 4 sites where this was performed (eTable; http://www.jamaneuro.com) suggests that rarer variants are likely overrepresented in cases compared with equivalent controls.
In subset analyses of DLB by the method of genotyping, all 2683 samples were included in the meta-analysis, and the results suggest a higher estimate of risk than in our pooled analyses (OR, 10.06 [95% CI, 3.39–29.87]). However, owing to limitations in data, including certain centers lacking screened controls or centers genotyping cases and controls differently, confounding by site existed in the pooled analyses. Meta-analysis by study center showed an attenuated OR compared with pooled or genotyping method–stratified analyses and comparatively increased effect heterogeneity (OR, 5.24 [95% CI, 1.82–15.08; I2 = 0.9]), possibly due to these confounding effects. The OR varied quite markedly by subgroup when stratified by study center, by genotyping method, or by a clinical or a pathological diagnosis, with all heterogeneity P values for meta-analyses of the subsets ranging from .04 to .08 (I2, 0–0.90), indicating moderate, but only suggestive, heterogeneity of effects in some analyses (Table 5).
Additional analyses indicated that GBA1 mutations were also significantly associated with earlier ages at onset and death in DLB cases carrying mutations compared with noncarriers (P <.001; Table 2 and Table 3). Similar to the previous study of PD, the age at disease onset in DLB was approximately 5 years earlier in cases with GBA1 mutations than in those without, with mean ages of 63.5 and 68.9 years, respectively. The mean period from diagnosis to death in DLB cases carrying mutations vs those without was quite similar (10.5 and 9.2 years, respectively). Hoehn and Yahr scale and Unified Parkinson’s Disease Rating Scale scores tended to be higher in DLB with GBA1 mutations (P = .028 and P = .045, respectively). Treatment with antipsychotics and the presence of visual hallucinations were slightly more common in mutation carriers (P = .024 and P = .053 respectively). In the subset of subjects for whom pathological descriptions were provided, the LB burden and the presence of Lewy neurites did not differ significantly between mutation carriers and noncarriers.
This study demonstrated that the GBA1 variant E326K, but not T369M, was associated with DLB (OR, 2.72 [95% CI, 1.38–5.54; P = .006]) and PDD (3.91 [1.41–10.86; P = .009). This finding differed from the results of the study of GBA1 in PD in which the ORs for E326K in the non-Ashkenazi and Ashkenazi groups were 0.66 and 0.40, respectively.9 However, the finding is consistent with a previous genome-wide meta-analysis of 7976 PD cases and 6350 controls for E326K (OR, 1.71 [P = 5 × 10−8]) and T369M(1.15 [P = .49]).17 If our combined data are reanalyzed considering E326K as a mutation for DLB, the OR is 8.38 (95% CI, 5.47–12.84; P < 2 × 10−16).
Our finding that mutations in GBA1 are a significant risk factor for DLB confirms the overall impression that GBA1-associated parkinsonism is characterized by an increased incidence of dementia. In fact, GBA1 mutations appear to have a larger contribution to the genetic etiology of DLB compared with PD and PDD. Evidence of a limited genetic overlap between different LB disorders has been reported previously.18 In addition, as was seen in PD, GBA1 mutations appear to affect the age at onset of DLB and its overall severity and rate of symptom progression. This finding correlates with recent positron emission tomography studies of the neurobiology of GBA1-associated PD examining brain dopamine synthesis with fluorodopa F 18 and resting regional cerebral blood flow with oxygen 15–labeled water (H215O) in subjects with Gaucher disease and PD. Although the distribution and extent of dopamine synthesis does not differ between PD subjects with or without GBA1 mutations, the H215O positron emission tomography studies indicated that patients with Gaucher diseases and parkinsonism have decreased resting activity in a pattern characteristic of diffuse LB disease.19 Neuropathological evaluation of GBA1-associated parkinsonism also indicates considerable LB disease in confirmed mutation carriers.20,21
Although different results were obtained from the various participating groups, combining data from multiple centers using meta-analytic techniques and including previously unpublished results maximized the sample size. To attempt to address possible sources of confounding, we decided a priori to subset and then combine the data in this analysis to make more specific and finely tuned generalizations. We believe that this method provides a better assessment of the frequency of mutations overall across centers in a more real-world research setting.
One weakness of this study is that it included subjects undergoing genotyping by different methods. On the other hand, a strength is that GBA1 exons were fully sequenced in 80.6% of subjects with DLB and 66.2% with PDD. Although we accepted data from each of the centers regardless of the genotyping methods used, we were careful to analyze the sequenced samples separately and to include the other samples as additional validation or support. Although in some instances patients and controls underwent different methods of genotyping, we addressed this difference by stratifying samples and adjusting for covariates in pooled regression models as shown in the forest plots (Figure), in which we compared controls and subjects who underwent sequencing. One center in Antwerp reported a higher frequency of GBA1 mutations among subjects with PDD than among those with DLB. Although that center adhered strictly to the 1-year rule in addition to the McKeith criteria, because no cases were pathologically confirmed, some cases may actually have had LB disease. Also, because 3 subjects with PDD and 1 with DLB all carried the L444P mutation, the finding could reflect a previously described founder effect in Belgium.22 The higher OR in subjects with clinically diagnosed disease compared with those with pathologically confirmed disease remains perplexing.
Although mechanisms underlying GBA1-associated parkinsonism are still not totally understood, accumulating evidence suggests that impairment of the aging lysosome, enhanced by deficient or mutant glucocerebrosidase, can affect α-synuclein degradation.23 Studies in a murine model of Gaucher disease generated using the specific glucocerebrosidase inhibitor conduritol B epoxide24 and in knock-in mice with specific Gba1 point mutations demonstrate enhanced α-synuclein pathological findings.25–27 Another study in primary or human-induced pluripotent stem cell–derived neurons suggests that the functional loss of glucocerebrosidase causes accumulation and aggregation of α-synuclein, which in turn inhibits the trafficking and lysosomal activity of glucocerebrosidase leading to self-propagating disease.28 Disruption of autophagy could also play a role,29,30 especially because lysosomal dysfunction has been implicated in the pathology of DLB.31 Ultimately, understanding the basis for this observation will provide insight into the role of glucocerebrosidase in LB disease, with important prognostic and therapeutic implications.
Funding/Support: Certain tissue for this study was provided by the Newcastle Brain Tissue Resource, which is funded in part by grant G1100540 from the UK Medical Research Council (MRC) and awards from the UK National Institute for Health Research (NIHR) Biomedical Research Centre in Ageing and Age Related Diseases and Biomedical Research Unit in Lewy Body Disease to the Newcastle upon Tyne Hospitals National Health Service Foundation Trust, and by a grant from the Alzheimer’s Society and Alzheimer’s Research Trust as part of the Brains for Dementia Research Project (Drs Kurzawa-Akanbi, McKeith, Chinnery, and Morris). The clinical sample collection was partly supported by Parkinson’s Disease UK grant K-0901. Human brain tissue samples for DNA extraction were received from the Sydney Brain Bank, which has approval to collect and distribute human brain tissue for research studies by the Human Research Ethics Committee of the University of New South Wales under the Human Tissue Act 1983. The French project was supported by contract ANR-08-NEUR-004-01 from the National Research Agency in the ERANET NEURON framework, and the research leading to these results has received funding from the program Investissements d’avenir ANR-10-IAIHU-06. Some of the human brain tissue samples for DNA extraction were received from the Massachusetts Alzheimer Disease Research Center and the Harvard Brain Tissue Research Center. In addition, this study was supported by the UK Lewy Body Disease Society (Dr Kurzawa-Akanbi); by the UK Health Protection Agency (Dr Morris); by the Methusalem excellence program of the Flemish government, a Centre of Excellence grant from the University Research Fund of the University of Antwerp, the Research Foundation Flanders, the Agency for Innovation by Science and Technology Flanders, the Interuniversity Attraction Poles program of the Belgian Science Policy Office, the Foundation for Alzheimer Research Belgium, and the Belgian Parkinson Foundation (Drs Theuns, Crosiers, Cras, Engelborghs, De Deyn, and Van Broeckhoven); by Alzheimers Research UK and Alzheimer’s Society through their funding of the Manchester Brain Bank under the Brains for Dementia Research initiative (Drs Mann, Snowden, and Pickering-Brown and Mss Halliwell, Davidson, Gibbons, and Harris); by the MRC and Wellcome Trust (Drs Mann and Pickering-Brown); by award WT089698 from the Wellcome Trust/MRC Joint Call in Neurodegeneration to the UK Parkinson’s Disease Consortium whose members are from the Institute of Neurology, University College London, the University of Sheffield, and the MRC Protein Phosphorylation Unit at the University of Dundee (Drs Duran, Sheerin, Bras, and Hardy); by a postdoctoral fellowship from Alfonso Martin Escudero Foundation (Dr Duran); by the Reta Lilla Trust (Dr Hardy); by grants 5R01NS060113 (Dr Clark) and R56NS036630 (Dr Marder) from the National Institute of Neurological Disorders and Stroke, NIH; by grant P50AG08702 (Drs Honig and Clark) from the National Institute on Aging, NIH; by the Panasci Fund for Lewy Body Research (Dr Honig), Alzheimer Disease Drug Discovery Foundation (Dr Honig), and the Parkinson’s Disease Foundation (Drs Clark and Dr Marder); by European Project on Mendelian Forms of Parkinson’s Disease grant agreements 241791 and FP7/2007-2013 from the European Community’s Seventh Framework Programme (Drs Brockmann and Gasser); by research fund grants from the Canadian Institutes of Health Research (Drs Rogaeva and Black); by a New Investigator Award from the Parkinson Society Canada (Dr Masellis); by grants 2R56NS037167 and 2P30AG010133 from the National Institute on Aging, NIH (Dr Ghetti); by fellowship 630434 and project support 1008307 from the National Health and Medical Research Council of Australia (Dr Halliday); by grants from the Agence Nationale pour la Recherche (Drs Lesage, Klebe, Durr, Duyckaerts, and Brice); by High Q (Dr Durr); by grant NS053488 from the National Institute of Neurological Disorders and Stroke (Dr Ghetti); by the Intramural Research Program of the National Human Genome Research Institute, NIH (Drs Lopez, Tayebi, Knight, Wolfsberg, and Sidransky and Ms Landazabal); and by National Institute on Aging project Z01 AG000949-02 from the Intramural Research Program of the NIH (Drs Nalls and Singleton and Ms Keller).
Author Contributions: Dr Sidransky had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Nalls, Mann, Hardy, Honig, Trojanowski, Wolfsberg, and Sidransky. Acquisition of data: Duran, Kurzawa-Akanbi, McKeith, Chinnery, Morris, Theuns, Crosiers, Cras, Engelborghs, De Deyn, Van Broeckhoven, Mann, Snowden, Halliwell, Davidson, Gibbons, Harris, Sheerin, Bras, Clark, Marder, Honig, Berg, Maetzler, Brockmann, Gasser, Novellino, Quattrone, Annesi, DeMarco, Rogaeva, Masellis, Black, Bilbao, Ghetti, Nichols, Halliday, Lesage, Klebe, Durr, Duyckaerts, Brice, Giasson, Trojanowski, Landazabal, Knight, Singleton, Wolfsberg, and Sidransky. Analysis and interpretation of data: Nalls, Duran, Lopez, Theuns, Croziers, Van Broeckhoven, Pickering-Brown, Clark, Honig, Brockmann, Masellis, Foroud, Pankratz, Trojanowski, Hurtig, Tayebi, Keller, and Sidransky. Drafting of the manuscript: Nalls, Chinnery, Theuns, Van Broeckhoven, Mann, Pickering-Brown, Gibbons, Foroud, Klebe, Giasson, Trojanowski, Landazabal, and Sidransky. Critical revision of the manuscript for important intellectual content: Nalls, Duran, Lopez, Kurzawa-Akanbi, McKeith, Morris, Theuns, Crosiers, Engelborghs, Van Broeckhoven, Snowden, Pickering-Brown, Halliwell, Davidson, Harris, Sheerin, Bras, Hardy, Clark, Marder, Honig, Berg, Maetzler, Brockmann, Gasser, Novellino, Quattrone, Annesi, De Marco, Rogaeva, Masellis, Black, Bilbao, Ghetti, Nichols, Pankratz, Halliday, Lesage, Durr, Duyckaerts, Brice, Trojanowski, Hurtig, Tayebi, Knight, Keller, Singleton, Wolfsberg, and Sidransky. Statistical analysis: Nalls, Foroud, Trojanowski, Keller, and Sidransky. Obtained funding: McKeith, Chinnery, Theuns, De Deyn, Van Broeckhoven, Pickering-Brown, Hardy, Clark, Black, Nichols, Brice, Giasson, Tayebi, and Sidransky. Administrative, technical, and material support: Nalls, Lopez, Kurzawa-Akanbi, Morris, Theuns, Van Broeckhoven, Mann, Halliwell, Davidson, Gibbons, Harris, Sheerin, Bras, Clark, Honig, Berg, Maetzler, Gasser, Rogaeva, Ghetti, Pankratz, Lesage, Duyckaerts, Giasson, Landazabal, Knight, Wolfsberg, and Sidransky. Study supervision: Nalls, McKeith, Chinnery, Morris, Theuns, Van Broeckhoven, Pickering-Brown, Clark, Bilbao, Trojanowski, Singleton, and Sidransky.
Conflict of Interest Disclosures: Dr Crosiers reports receiving travel support from Abbott and Boehringer Ingelheim. Dr Brockmann reports receiving speaker honoraria from Deutsche Gesellschaft für klinische Neurophysiologie, GlaxoSmithKline, and Orion Corporation. Dr Berg reports receiving consultation fees from UCB Schwarz Pharma, Merz, and Novartis; payment for lectures from Teva, GSC, Lundbeck, and UCB Schwarz Pharma; and grant support from Janssen Pharmaceutica, Teva, Michael J. Fox Foundation, the Federal Ministry of Education and Research, Abbott, Boehringer, German Parkinson’s Disease Association, and the Center for Integrative Neurosciences. Dr Gasser reports receiving consultation fees from Cefalon Pharma and Merck-Serono; lecture fees from Valeant Pharma, Merck-Serono, UCB-Pharma, and Boehringer; and grant support from Novartis Pharma, European Union, Helmholtz Association, and the Federal Ministry of Education and Research. Dr Maetzler reports receiving lecture fees from GlaxoSmithKline and grant support from the Robert Bosch Foundation. Dr Masellis reports receiving speaker honoraria from Novartis and EMD Serono, Inc; serving as an associate editor for Current Pharmacogenomics and Personalized Medicine; receiving publishing royalties from Henry Stewart Talks; serving as a consultant for Bioscape Medical Imaging CRO; and receiving research support from the Canadian Institutes of Health Research, Parkinson Society Canada, an Early Researcher Award from the Ministry of Economic Development and Innovation of Ontario, Teva Pharmaceutical Industries Ltd, and the Department of Medicine, Sunnybrook Health Sciences Centre. Dr Black reports receiving contract research funds to the Cognitive Neurology Research and Stroke Research Units from Roche, GlaxoSmithKline, Novartis, Myriad, Pfizer, sanofi-aventis, Boehringer Ingelheim, Lundbeck, Novo Nordisk, and AstraZeneca and honoraria from Janssen-Ortho, Novartis, Lundbeck, Pfizer, Eisai, Myriad, GlaxoSmithKline, Roche, Schering-Plough, Elan, Wyeth, and Bristol-Myers Squibb. Dr Ghetti reports receiving support from Elan and Bayer Schering Pharma AG. Dr Nichols reports receiving support from the National Institute of Neurological Disorders and Stroke, National Institutes of Health (NIH).