Last month, a study published in the American Journal of Psychiatry made international headlines by claiming that schizophrenia is not a single disease, but rather a group of eight distinct genetic disorders. Now, ten leading geneticists say this extraordinary claim isn't justified by the data.
In the AJP paper, C. Robert Cloninger, Igor Zwir, and colleagues analyzed genetic influences on more than 4,000 people with schizophrenia. This helped them identify distinct gene clusters that appeared to contribute to eight different classes of the disorder. The researchers concluded that, "[S]chizophrenia is a group of heritable disorders caused by a moderate number of separate genotypic networks associated with several distinct clinical syndromes." It was a stunning declaration — one with profound implications to our understanding of the disorder and how it might be treated.
Writing in Genomes Unzipped: Public Personal Genomics, a group of geneticists led by Patrick Sullivan of Sweden's Karolinska Institute are now claiming that the authors of the study did not provide the necessary evidence. They write:
Their claims are based upon complex (and we believe flawed) analyses that are said to reveal links between clusters of clinical data points and patterns of data generated by looking at millions of genetic data points. Instead of the complexities favored by Arnedo et al., there are far simpler alternative explanations for the patterns they observed. We believe that the authors have not excluded important alternative explanations – if we are correct, then the major conclusions of this paper are invalidated.
The geneticists identified five specific areas of concern that they feel were not adequately addressed in the study, "each of which calls into question the conclusions."
- Ancestry/population stratification: The geneticists are not happy with the samples used, saying they introduce too many confounders that could give false associations.
- The x chromosome: The researchers failed to account for sex in their matrix factorization; in statistical genetics, sex chromosomes require special handling — an issue not addressed by the authors.
- Linkage disequilibrium: The geneticists question the authors' interpretation of SNP clusters (single nucleotide polymorphism).
- SNP selection: The reported P-values are incorrect, say the critics, the result of a selection effect.
- Replication: "[I]t is unclear precisely what [was] replicated, exactly how this was done, and whether the degree of 'replication' deviated from that expected by chance," write the geneticists.
Sullivan and his colleagues say it's "highly likely" that the results are not relevant for schizophrenia, and urge "great caution" in the interpretation of the results of the study.
We contacted the study's lead authors for a response.
"My colleagues and I will prepare a reply to the comments in PubMed Commons next Monday," said the Washington University School of Medicine's C. Robert Cloninger via email. He continued:
Patrick Sullivan and his colleagues organized PGC on the rationale that increasing sample size could correct the problem of weak and inconsistent results in GWAS [genome-wide association studies]. Unfortunately their rationale is questionable as shown by the facts that their subjects are clinically heterogeneous and their associations remain weak.
Cloninger said it's unfortunate they have taken the unconventional route of publishing criticisms without peer review, which ultimately is not in the best interests of objective science.
"In contrast, our paper was thoroughly peer-reviewed," he told io9. "Their criticisms are seriously flawed and their rush to specious fault-finding does not serve science or anyone well."
To which his colleague Igor Zwir added:
This is not the way real science works — science is peer review and that's the reason why we're delaying. We have a response, but we're going to go about it through the appropriate channels. Otherwise we'd be guilty of doing what they've done.
In response to these criticisms and concerns, Gerome Breen of King's College London sent io9 this email:
Prof Cloninger is not addressing the issue, which is about the paper he and his collaborators published. Science moves quickly these days and we wanted to use modern means of communication to make the point that there are apparent serious flaws with the study. Letters to editors are essentially old technology, science blogs and PUBMED commons are the way scientists communicate now.
It would also be good if all the parties involved declared any potential conflicts of interest. We wrote our piece, not as part of the Psychiatric Genomics Consortium, but rather as a set of experts in genetics of this disorder and in the type of analyses they use, in the desire to stop false memes entering the field and to avoid giving the public false conceptions of what genetics and cannot do for this really serious disorder.
When asked to clarify what he meant by a potential conflict of interest, Breen responded:
Many previous papers like this [have] been accompanied by patenting or the launch of a company or test. We don't know if the authors are doing that but we would prefer if they made it clear one way or the other.
So he's got no evidence for conflict of interest. That said, the paper also does not contain a COI statement, which is pretty standard for papers like this.
*Update: October 6, 2014: Cloninger and colleagues have published their formal response to these criticisms at PubMed Central. Read it here.
Read the entire Genomes Unzipped article here.
Top image: a genotypic network in which nodes indicate SNP sets linked by shared SNPs (blue lines) and/or subjects (red lines). Arnedo et al.