A new study casts doubt on the utility of combining neuroimaging and genetics in hopes of understanding the genetic underpinnings of cognition. The researchers report today (September 30) in the Journal of Neuroscience that they were unable to reproduce the results of an influential 2012 study in the same journal that showed an association between genetic variants and brain function as shown on an fMRI during a reading test. In analyzing and replicating the experiment, the team noted several characteristics common to neuroimaging and genetics studies that they say could be responsible for the false-positive result.
“There was a belief that we could find out why different brains work in different ways by running basically the same MRI studies we always did, but now also taking a saliva sample to examine the genetic variation in the people being studied,” Julia Uddén, a psychologist at Stockholm University and the lead author of the latest study, tells The Scientist in an email. “But the effect sizes being reported seemed surprisingly large for studies of healthy people, given what we know about the complexity of genetic influences on language functions, so we were skeptical about the findings and wanted to see whether we could provide an independent replication.”
In the 2012 study, Philippe Pinel, a cognitive neuroscientist at the École des Neurosciences de Paris Île-de-France, and his team took DNA samples from 94 patients and then had them complete a reading test while an fMRI scanner measured their brain activity. They found that certain genetic variants were correlated with differences in brain activity during the task. The work, which Pinel says was itself a replication experiment of several independent linkage studies, is regularly cited as evidence for the idea that neuroimaging can be an effective way to associate genes with cognitive function (according to Google Scholar, it’s been cited 181 times).
“This 2012 study is often held up to support the idea that functional neuroimaging can give stronger connections to genetic data than other ways of measuring behavior or cognition,” says Simon Fisher, a geneticist at the Max Planck Institute for Psycholinguistics and the senior author of the current study. “Our study attempted to replicate the findings from that prior study in independent cohorts that, combined, are four times the size of the original investigation.”
For the new study, the researchers recruited 427 participants in two different labs to perform a similar reading task, fMRI, and genetic screen. It found no association between the genes examined in the 2012 study and brain function. Uddén and her team pointed out that in 2012, more data and rigorous statistical analysis would likely have negated the original findings.
“The field in general is supporting this idea that smaller sample sizes have yielded false positives, and moving forward we can begin to design and implement studies with larger populations,” says Genevieve Konopka, a neuroscientist at UT Southwestern Medical Center who was not involved in either study.
Pinel suggests one possible reason that the new study did not link particular genetic sequences to brain activity could be the variability of the sample. The 2012 cohort was highly homogeneous, made up of individuals with similar levels of education who lived in the same area. The new study used many more participants from different areas and backgrounds. Still, Pinel says, “the null results reported by Fisher et al. challenge our results and [demonstrate the] need to stimulate a better understanding of the possible sources of inconsistency.”
J. Uddén et al., “Towards robust functional neuroimaging genetics of cognition,” Journal of Neuroscience, doi:10.1523/JNEUROSCI.0888-19, 2019.
Emma Yasinski is a Florida-based freelance reporter. Follow her on Twitter @EmmaYas24.
Correction (October 1): The original image showed a CT scanner, rather than an MRI machine. The Scientist regrets the error.