1. Introduction

In the past decade, literally hundreds of reports have been published claiming or refuting genetic association between putative PD genes and disease risk, onset age variation, or other phenotypic variables. Currently, 60 to 90 PD genetic association studies are being published annually from research groups worldwide. For the public but also for the PD genetics research community this wealth of information is becoming increasingly more difficult to follow, evaluate, and much less to interpret.

2. Database Organization and Methods

Overview

The goal of the PDGene database is to serve as a comprehensive, unbiased, publicly available and regularly updated collection of published genetic association studies performed on PD. Eligible publications are identified following systematic searches of scientific literature databases, as well as the table of contents of journals in genetics, neurology, and psychiatry. Data selected for display summarize key characteristics of the investigated study cohorts (e.g., gene overview), as well as genotype distributions in cases and controls (e.g., polymorphism details). For polymorphisms with genotype data in at least four case-control samples, continuously updated random-effects meta-analyses are presented (see meta-analysis methods). Note that data obtained from family-based studies are not included in the meta-analyses, as crude odds ratios cannot be readily calculated from overall genotype distributions. However, these studies and their qualitative results are still listed on the gene-summary pages of the PDGene website (see Table 2 for example).

To ensure the highest degree of scientific objectivity, only studies published in peer-reviewed journals available in English are considered for inclusion into the database. In particular, this precludes the inclusion of data presented only in abstracted form, e.g. at scientific meetings. We encourage authors of original reports fulfilling the above criteria to submit their data as soon as their work is accepted for publication.

Meta-Analysis Methods

For all polymorphisms with minor allele frequencies in healthy controls >1%, and for which case-control genotype data are available in four or more independent samples, crude odds ratios (ORs) and 95 percent confidence intervals (CIs) are calculated from the reported allele distributions for each study. Summary ORs and 95 percent CIs are calculated using the DerSimonian and Laird (1986) random-effects model, which utilizes weights that incorporate both within-study and between-study variance. This procedure is done including all studies irrespective of ethnicity (denoted by "All studies" on the meta-analysis figures), and repeated after exclusion of the initial study ("All excl initial"), after exclusion of studies in which a deviation of Hardy-Weinberg Equilibrium (HWE) was detected in controls ("All excl HWE deviations"). In addition, ethnicity-specific meta-analyses are performed whenever genotype data is available from at least three independent case-control populations. Visually, the results of these meta-analyses are displayed for each polymorphism in form of forest plots. Overlapping samples (of which usually only the largest is included), studies with missing data, or control samples deviating from HWE are indicated on these graphs. Note that when only a few studies are included in the meta-analyses (i.e. less than ~10), the random effects model may yield summary ORs and confidence bounds that are slightly anti-conservative.

Along with the forest plots, which depict summary ORs only for the most current set of available data, the results of cumulative meta-analyses are displayed for each polymorphism with data available in at least four independent case-control datasets. These graphs display summary ORs (for the "All studies" paradigm using the same allele-based random-effects analyses as outlined above) recalculated after each study. Thus, these graphs allow an evaluation of the estimated summary effects over time. Note that the most current summary OR (listed at the very top of the cumulative meta-analysis plots) is identical to the corresponding summary OR depicted on the forest plots (see above).

Inclusion of Genome-wide Association Studies (GWAS)

GWAS summaries: Data from GWAS and other large-scale studies are summarized in a dedicated section on PDGene. This section distinguishes between "GWAS" (≥10,000 independent markers) and "Other large-scale association studies" (≥1,000 independent genetic markers; this latter section also includes re-analyses or meta-analyses of previous GWAS). In addition to providing the main characteristics of each study, this section also provides a hyperlinked list of "featured genes", i.e. those loci or pathways highlighted by the primary authors as the main outcome of their study after having completed all analyses. Note that the criteria used to define of "featured genes" varies across publications.

GWAS genotype data: The extent of data integration from large-scale genotyping studies (e.g. GWAS) is based on the availability of the data. Whenever possible, individual-level data is obtained (e.g. via dbGaP), and study-specific ORs determined after data-cleaning and adjustment for age, gender and population stratification. In cases where only summary-level genotype data were available, we either calculate crude ORs from the available data or use the allelic ORs and 95% CI's supplied by the original investigators. If neither individual-level or summary-level genome-wide genotype data are available, we include as much of the data possible reported in the primary publications; these are usually restricted to a set of "featured genes" or other loci of interest. Generally, preference is given to include allelic ORs adjusted for population stratification as reported in the primary publications. If these are not available or not eligible, crude ORs are calculated from the provided genotype/allele summary data. Finally, ORs from GWAS and other large-scale studies are then combined by meta-analysis as outlined above.

***Please note that GWAS genotype data and allele frequencies cannot be displayed online due to data protection policy (see ref. Homer et al., 2008, for explanation), unless provided in the original publications. The respective SNP entries are labelled as "Either no data provided, or data otherwise not eligible for inclusion". However, GWAS genotype data are included in the PDGene meta-analyses and rounded ORs and CIs are displayed on the respective graphs where applicable.***

Association studies on mitochondrial genes

Studies assessing a potential association between PD and genetic variants in the mitochondrial (mt) genome are subject to the same inclusion criteria as studies investigating markers from the nuclear genome, and are displayed on a separate "chromosome graph" (which is adapted from imagery on the "Mito Map" website [http://www.mitomap.org/]). Owing to the specific characteristics of human mt-inheritance (e.g. its multicopy nature and the high frequency of somatic mutation events) and the innate heterogeneity of mt-association studies, however, genotype data from these studies are not included on PDGene and therefore not subject to meta-analysis.

For more details on inclusion criteria, literature searches, data-management procedures, statistical analyses, and online database structure, please see Bertram et al. (2007).

3. The "Top Results" List

In an effort to facilitate the identification of the most promising meta-analysis results available in PDGene, a continuously updated list displaying the most strongly associated genes ("Top Results") has been added to the PDGene homepage. The list includes genes/loci which contain at least one variant showing a nominally significant summary OR in the analysis of all studies (“All”), or those limited to samples of a specific ethnicity (e.g. “Caucasian”). The nominally significant meta-analyses are then graded based on interim guidelines "Venice critera" for the grading of the epidemiological credibility of genetic association studies recently developed by the Human Genome Epidemiology Network (HuGENet; Ioannidis et al, 2008).

In the "Top Results" list, genes are ranked based on statistical significance (P value). For genes with more than one polymorphism showing nominally significant association, ranking is based on the best statistical meta-analysis result per gene.

While we believe that this list represents an up-to-date summary of particularly promising PD candidate genes that warrant follow-up with high priority, we note that many of these may still represent false-positive findings.

HuGENet "Venice criteria"

We rate overall epidemiological credibility as ‘strong’ if associations received three A grades, ‘moderate’ if they received at least one B grade but no C grades, and ‘weak’ if they received a C grade in any of the three assessment criteria. Current Venice rating of the PDGene top results can be found here.

Briefly, each meta-analyzed association in PDGene is graded on the basis of the amount of evidence, consistency of replication, and protection from bias. For amount of evidence, we assign the grade ‘A’ when the total number of minor alleles of cases and controls combined in the meta-analyses exceeds 1,000, ‘B’ when it is between 100 and 1,000, and ‘C’ when it is less than 100. For consistency of replication, we assign the grade ‘A’ for I2 point estimates <25%, ‘B’ for I2 values of 25–50%, and ‘C’ for I2 values >50%. Note that this criterion does not apply to meta-analyses with a P-value <1x10-7 after exclusion of the initial studie(s), as described in Khoury et al, 2009. For protection from bias, the guidelines propose consideration of various potential sources of bias, including errors in phenotypes, genotypes, confounding (population stratification) and errors or biases at the meta-analysis level (publication and other selection biases). A grade A implies that there is probably no bias that can affect the presence of the association, grade B that there is no demonstrable bias but important information is missing for its appraisal, and grade C that there is evidence for potential or clear bias that can invalidate the association. Errors and biases are also considered in the framework of the observed summary OR. Whenever the summary OR deviates less than 1.15-fold from the null in meta-analyses based on published data, we acknowledge that occult publication and selective reporting biases alone may invalidate the association, regardless of the presence or absence of other biases, and therefore assign a grade of C. When the summary OR deviates more than 1.15-fold from the null, we assign a grade of C when the modified regression test (Hardbord et al, 2006) or excess test suggest the possibility of publication-bias or significance-chasing bias or when the association is no longer nominally statistically significant upon exclusion of the initial study or studies violating HWE.

4. Mendelian PD Genes

Note that the PDGene database will not sample and catalogue reports of rare causal mutations leading to PD or parkinsonism of classic Mendelian inheritance (such as disease-causing mutations in a-synuclein or parkin), but will only focus on genetic association studies using polymorphisms with a minor-allele frequency of 1% or greater in at least one of the studied control populations. For a collection of Mendelian PD genes, please consult the Parkinson Disease Mutation Database curated by the Flanders Institute for Biotechnology and University of Antwerp, the "Parkinson's disease Mutation Database curated by the Parkinson's Institute from Leiden University Medical Center, or the Mutation Database for Parkinson's Disease curated by the Institute for Infocomm Research in Singapore. Note that despite the exclusion of disease-causing mutations, data from bona fide association studies using common polymorphisms in any of the Mendelian PD genes will be included and meta-analyzed for PDGene (e.g. see the entries for SNCA or LRRK2).

References

Bertram L, McQueen MB, Mullin K, Blacker D, Tanzi RE. (2007) "Systematic meta-analyses of Alzheimer disease genetic association studies: the AlzGene database." Nat Genet 39(1): 17-23. Abstract

de la Fuente-Fernandez, R. (2003). "A note of caution on correlation between sibling pairs." Neurology 60(9): 1561; author reply 1561.

DerSimonian R, Laird N. "Meta-analysis in clinical trials. Control Clin Trials." 1986 Sep;7(3):177-88.

Egger M, Davey Smith G, Schneider M, Minder C. "Bias in meta-analysis detected by a simple, graphical test." BMJ. 1997 Sep 13;315(7109):629-34.

Evangelou E, Maraganore DM, Ioannidis JP. "Meta-analysis in genome-wide association datasets: strategies and application in Parkinson disease." PLoS ONE. 2007 Feb 7;2:e196.

Farrer, M.J. (2006). "Genetics of Parkinson disease: paradigm shifts and future prospects." Nat Rev Genet 7(4):306-18.

Forman MS, Lee VM, Trojanowski JQ. (2005). "Nosology of Parkinson's disease: looking for the way out of a quagmire." Neuron 47(4):479-82.

Fung HC, Scholz S, Matarin M, Simon-Sanchez J, Hernandez D, et al. "Genome-wide genotyping in Parkinson's disease and neurologically normal controls: first stage analysis and public release of data." Lancet Neurol. 2006 Nov;5(11):911-6.

Gasser T " Molecular pathogenesis of Parkinson disease: insights from genetic studies." Expert Rev Mol Med. 2009 Jul 27;11:e22.

Homer N, Szelinger S, Redman M, Duggan D, Tembe W, et al. (2008) "Resolving individuals contributing trace amounts of DNA to highly complex mixtures using high-density SNP genotyping microarrays." PLoS Genet. 2008 Aug 29;4(8):e1000167.

Ioannidis JP, Boffetta P, Little J, O'Brien TR, Uitterlinden AG, et al. (2008) "Assessment of cumulative evidence on genetic associations: interim guidelines." Int J Epidemiol 37(1):120-32. Abstract

Khoury MJ, Bertram L, Boffetta P, Butterworth AS, Chanock SJ, et al. (2009) "Genome-wide association studies, field synopses, and the development of the knowledge base on genetic variation and human diseases." Am J Epidemiol 170(3):269-79. Abstract

Lin MT Simon D.K. (2005). Comment on "No evidence for heritability of Parkinson disease in Swedish twins." Neurology.64(5):932

Maher, N. E., L. I. Golbe, et al. (2002). "Epidemiologic study of 203 sibling pairs with Parkinson's disease: the GenePD study." Neurology 58(1): 79-84.

Maraganore DM, de Andrade M, Lesnick TG, Strain KJ, Farrer MJ, Rocca WA, Pant PV, Frazer KA, Cox DR, Ballinger DG. "High-resolution whole-genome association study of Parkinson disease." Am J Hum Genet. 2005 Nov;77(5):685-93.

Pardo, L.M., van Duijn, C.M. (2005). "In search of genes involved in neurodegenerative disorders." Mutat Res 30;592(1-2):89-101.

Tanner, C. M., R. Ottman, et al. (1999). "Parkinson disease in twins: an etiologic study." Jama 281: 341-6.

Sveinbjornsdottir S, Hicks AA, et al (2000). "Familial aggregation of Parkinson's disease in Iceland." N Engl J Med 343(24):1765-70.

Wirdefeldt K, Gatz M, et al. (2004). "No evidence for heritability of Parkinson disease in Swedish twins." Neurology 63(2):305-11.