A Research Blog

Enjoying some time away from the bench. (L to R: Dr Farhan Mohammad, Asst Prof Adam Claridge-Chang, Dr Joses Lim)

Adam Claridge-Chang is Assistant Professor with the Neuroscience & Behavioural Disorders Programme at Duke-NUS Medical School. We talked to him about his most recent publication in Neuroscience & Biobehavioural Reviews: Concordance and incongruence in preclinical anxiety models: systematic review and meta-analyses.

Q: How did this paper come about?

Anxiety disorders are the most prevalent mental illnesses, but pharmaceutical companies have not been successful in finding effective treatments due to poor understanding of the underlying causes of mental illness. We were interested in finding new ways to understand anxiety by utilising the powerful genetic tools available in fruit fly (Drosophila melanogaster) research. Specifically,

Dr Farhan Mohammad, a talented postdoc in my lab, was working to set up a fruit fly model for anxiety. When he turned to the rodent anxiety literature to guide the development of the fly model, he was surprised to find a lack of consensus about which genes regulated anxiety. This posed an obstacle, since our strategy to validate the fly model relied on a direct comparison with preclinical data. At this point, conducting a meta-analysis seemed the best way to make sense of what the rodent data on anxiety was really telling us.

Farhan and I had to retrain in biostatistics and learn how to do meta-analyses. Luckily, there are good textbooks in this area, and we were able to collaborate with biostatisticians in Duke-NUS’s Centre for Quantitative Medicine on other projects, which really helped us prepare for this project. Along with Dr Joses Ho, another talented postdoc in my lab, we recruited help from local final year students at Temasek Polytechnic, NUS and NTU to analyse the thousands of papers included in this meta-analyses.


Q: What are the main findings of your paper, and which to you is the most significant?

The thing that struck us was the prevalence of publication bias in the rodent anxiety literature. For example, we saw strong evidence for publication bias in the diazepam literature, even though nobody doubts that this drug effectively reduces anxiety.

How serotonin affects anxiety remains unclear. The selective serotonin reuptake inhibitors (SSRIs) are an important drug category that are thought to reduce anxiety by inhibiting the serotonin transporter.  Currently, drug companies sell about $5 billion of SSRIs annually. However, our meta-analysis shows that knocking out the serotonin transporter gene actually increases anxiety—this is the exact opposite of what you would expect for a protein targeted by an anxiety-reducing drug. It will be important to better understand the role that SSRIs and the serotonin transporter play in anxiety.


Q: Are these discrepancies related to the meta-analysis process itself, or is this a sign the anxiety field has some issues?

The meta-analysis process reveals discordance already present in the literature. By comparison, we did another meta-analysis on fly genetics. Our results indicated a high level of data integrity, and differences in conclusions between studies tended to be due to how the statistics was carried out rather than publication bias.


Q: What is your theory behind the conflicting data?

I have no theory that would explain the conflicting data, but it is clear that there appears to be two paradoxes at play. Firstly, since the SSRIs are used to reduce anxiety, then knocking out the serotonin transporter gene should also reduce anxiety, which does not happen (anxiety increases). Secondly, if knocking out the serotonin transporter gene causes anxiety in mice, then a similar effect should also be observed with the use of SSRIs, but that is not observed.


Q: What is the one thing you hope readers take away from this paper?

A: If you find your field confusing and conflicted, consider doing a meta-analysis. After you’ve completed the meta-analysis, you might still be confused, but at least you’ll know specifically what you’re confused about.


Q: In the grand scheme of things, how do you think preclinical studies should be conducted such that it gives reliable precursor data to support moving into the clinic? What do we need to do?

A: Things might improve if preclinical research was conducted more like clinical research. Researchers could register their studies, and report their findings regardless of the outcome, such that both positive and negative data were accessible to all. Some of the difficulties often faced with running clinical trials, such as subject recruitment, ensuring comparable clinical cohorts and patient confidentiality, are not issues in preclinical studies, so preclinical transparency is easier in principle. Marcus Munafo at Bristol University and others have also encouraged the formation of consortia amongst research groups to share data and work together. This might mean having more authors on the paper, but when published, the data are reliable and meaningful.

More emphasis can be placed on conducting meta-analyses and replication studies in preclinical work. Both are extremely important to gain consensus and confirm findings in any research field. Such studies would detect publication bias and provide confidence that the field is heading in the right direction.


You can read the Claridge-Chang Lab’s study here.


The authors were supported by Biomedical Research Council block grants to the Neuroscience Research Partnership and the Institute of Molecular and Cell Biology. Farhan Mohamed and Adam Claridge-Chang also received support from Duke-NUS Medical School. Joses Ho received support from the A*STAR Graduate Academy. Adam Claridge-Chang received additional support from a Nuffield Department of Medicine Fellowship, a Wellcome Trust block grant to the University of Oxford, A*STAR Joint Council Office grant 1431AFG120 and NARSAD Young Investigator Award 17741. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.



Search form