2 July 2008. We at the Schizophrenia Research Forum won’t pretend to be impartial about a particular paper in the July issue of Nature Genetics. The excitement that permeates our virtual walls comes from seeing Lars Bertram and colleagues at Massachusetts General Hospital report findings based on their SchizophreniaGene, which resides at SRF. Their meta-analyses point to associations between schizophrenia and 24 genetic variants in 16 different genes, but only four of those associations—namely, those involving DRD1 rs4532, DTNBP1 rs1011313, MTHFR rs1801131, and TPH1 rs1800532—earned high marks for epidemiologic credibility by new criteria proposed by the Human Genome Epidemiology Network, or HuGENet. These results, and the methods used to find them, may speed the search for schizophrenia genes.
The deluge of papers claiming to either incriminate or absolve specific genes in schizophrenia has taken interested parties on a wild ride. “For health care providers, researchers, and the general public, the accumulating information is increasingly difficult to follow, evaluate, and interpret,” the SZGene researchers write. Furthermore, they note, “No single gene or genetic variant has been established as a bona fide schizophrenia susceptibility gene, at least not with the confidence accorded to other genes associated with susceptibility to complex disease.”
Forging a new path
To contain the chaos, first author Nicole C. Allen, now at Columbia University, and associates created SZGene, a public database of genetic association studies in schizophrenia. They update it regularly in an effort to include every relevant English-language study published in a peer-reviewed journal. SZGene grew out of the AlzGene database, which Bertram and colleagues developed with help from our friends at the Alzheimer Research Forum (see Bertram et al., 2007).
Allen and colleagues performed meta-analyses of studies indexed in SZGene. They consider their work a step forward not only due to its scope, but also because “it systematically examines sources of biases and assigns a score for the epidemiologic credibility of the findings.” In addition, it scrutinized whole-genome as well as candidate gene studies.
The researchers began with the studies indexed in SZGene as of April 30, 2007, when it listed 1,179 papers that presented data on 3,608 genetic variants in 516 different genes. Only variants for which they had data from at least four independent case-control samples received further scrutiny, narrowing the focus of analysis to 118 variants in 52 genes. “This nearly doubles the number of meta-analyses thus far published in the field,” the authors write.
The new meta-analyses identified 24 variants in 16 genes as potential schizophrenia risk factors, with polymorphisms in APOE, COMT, DAO, DRD1, DRD2, DRD4, DTNBP1, GABRB2, GRIN2B, HP, IL1B, MTHFR, PLXNA2, SLC6A4, TP53, and TPH1 distinguishing cases from controls. For variants fingered as risk-raisers, odds ratios averaged 1.23; for potentially protective ones, they averaged 0.82.
“Notably, we identified significant risk-modifying effects in seven genes (DAO, DRD1, DTNBP1, GABRB2, HP, PLXNA2, and TP53) for which, to the best of our knowledge, no previous meta-analyses had been published,” the authors write. Contrary to past meta-analyses, the new one found no connection between schizophrenia and certain variants in BDNF, DRD3, NRG1, DAOA, or COMT.
Only two genomewide association studies (see SRF related news story) had been published before the cutoff date. One found an association between schizophrenia and variants in plexin A2 (PLXNA2) in individuals of European descent that a study of a Japanese sample could not replicate. The odds ratio found by Allen and colleagues, 0.82, hinted at a protective role for the gene. The other whole-genome study produced leads involving variants in CSF2RA and IL3RA, but too few samples existed for meta-analysis.
Making the grade
To evaluate the “epidemiological credibility” of the significant meta-analysis results, the researchers applied criteria proposed by the Human Genome Epidemiology Network (HuGENet). The benchmarks, drafted at a meeting in Venice, Italy, led by study coauthor John Ioannidis of the University of Ioannina (Ioannidis et al., 2008; Ioannidis et al., 2006), facilitate the grading of associations on the amount of relevant evidence, their consistency of replication, and their protection from bias.
Each association received an A, B, or C grade on each of the three dimensions assessing the evidence in support of an association. The grade for the amount of evidence reflected the number of minor alleles in the sample subjected to meta-analysis (e.g., an “A” grade for >1,000 minor alleles in cases and controls combined). The replication grade considered the degree of heterogeneity of odds ratios across all studies included in the meta-analysis. The protection from bias grade took into account potential biases due to genotyping, confounding, and selective reporting. Associations were deemed “strong” if they received three A grades, “moderate” if at least one B and no C grade, and “weak” if one or more C grades.
Only four associations, involving variants on DRD1, DTNBP1, MTHFR, and TPH1, earned straight A’s. “On the basis of the current data, these genes seem to be the best contenders to contain genuine susceptibility alleles modifying disease risk within the whole domain of genetic epidemiology in schizophrenia,” the researchers write.
DRD1 encodes the most plentiful dopamine receptor in the central nervous system. (For information about the dopamine hypothesis in schizophrenia, see SRF Current Hypotheses). Allen and colleagues write, “This receptor is thought to have a role in regulation of cognitive functions in the prefrontal cortex, possibly through interactions with NMDA-mediated neurotransmission, and to be involved in the action of clozapine.” DTNBP1, a top schizophrenia gene candidate, encodes dysbindin, a protein that may be expressed abnormally in schizophrenia and which may play a role in modulating neurotransmission via dopamine D2 receptors (see SRF related news story). MTHFR produces methylenetetrahydrofolate reductase, indirectly affecting homocysteine metabolism and within-cell methylation processes that may be related to schizophrenia; some evidence suggests that it plays a role in schizophrenia symptoms. TPH1 makes tryptophan hydroxylase 1, the rate-limiting enzyme in serotonin production. Many atypical antipsychotic drugs act on serotonin receptors (see SRF related news story).
As for the remaining genes, four showed modest epidemiologic credibility (i.e., they received at least one B grade, but no C’s) and 16, weak credibility (at least one C grade). Their lackluster grades do not rule them out as schizophrenia genes, any more than an A grade designates a surefire one. They simply say whether, based on the assumptions made in the guidelines, enough strong evidence exists at this time to posit a connection.
According to Allen and colleagues, “Our project represents the first comprehensive online resource for systematically synthesized and graded evidence of genetic association studies in schizophrenia.” However, they urge caution in interpreting its findings pending confirmation by future studies and clarification of the molecular pathways by which potential schizophrenia genes might affect risk. Even if updates to SZGene change the list of hot gene prospects, the methods used in this study could raise the bar for meta-analyses of schizophrenia genes. Furthermore, the researchers write, “The approach presented here can be easily adapted to genetic association studies of other common diseases of public health significance.”—Victoria L. Wilcox.
Allen NC, Bagade S, McQueen MB, Ioannidis JPA, Kavvoura FK, Khoury MJ, Tanzi RE, Bertram L. Systematic meta-analyses and field synopsis of genetic association studies in schizophrenia: The SzGene database. Nature Genetics. 2008 July;40(7):499-506. Abstract