Emily Willingham, Ph.D
Two autism studies made recent headlines: How a Gene Mutation Leads to Autism and Proximity to Freeway Increases Autism Risk. The first is a close-up on a particular gene using mice as a model and the other is an epidemiological study. The mouse study examined a gene/protein known as SHANK3, which has been implicated previously in autism. The epidemiological study looked at correlations between living near highways and autism rates.
Let’s take a look at the mouse study [PDF] first. SHANK3 has emerged as a protein of interest because of its roles in nerve connectivity. Researchers also have found that losing one copy of the SHANK3 gene — normally, people have two functional copies — can be associated with behaviors linked to autism. The authors of this new mouse study, which is being published in Molecular Autism, decided to knock out one Shank3 copy in mice to monitor the results. They looked specifically at connections among nerves and at social behaviors in mice missing a copy of the gene and compared them to genetically intact animals.
It’s always important to remember that people are not giant, tailless rodents. Mice are good models for whole-organism investigations of a gene’s effects, but they serve only as a starting point. The researchers found that mice missing a copy of Shank3 had deficits in nerve connectivity and in flexibility of forming nerve pathways. Animals lacking the gene also differed in their male-female social interactions.
From what we know about autism, social deficits aren’t limited to communications between males and females but cover a range of interactions. It’s unclear why the authors selected this male-female interaction as one of their behavioral tests, as such specificity is not a key manifestation of autism spectrum disorders. They do not describe their rationale in the paper.
One thing is clear: they used a behavioral test developed specifically for autism models, and the mice lacking a copy of Shank3 did fine on it. In other words, on an autism-specific behavioral test, the Shank3-less animals were no different than their genetically intact counterparts. Indeed, the authors do not even mention this outcome in their results but describe it briefly in the discussion. This finding seems to imply that for the specifically autism-related outcome, a missing copy of Shank3 had no effect at all.
This study looked at a missing copy of Shank3 . But too much SHANK3 also has been associated with ASDs, as has mutation in SHANK2, a related gene. Sure, this stuff is intriguing, but it’s not the promised How a Gene Mutation Leads to Autism as at least one headline put it. A better headline is “Gene changes linked to differential brain structure and behavior in mice,” but the latter, while accurate, doesn’t reach out and grab readers as the former does.
This kind of overpromising doesn’t help autistic people or autism families. This study has not clarified a role for SHANK3 specific to autism, yet in a news release associated with the work, one of the authors is quoted as saying, “These results have helped us determine a pathological mechanism behind neurodevelopmental disorders like autism.” The release goes on to say, “Armed with this breakthrough, we can begin testing drug compounds that treat the disease at its root cause, improving nerve cell communication.”
This is overpromising, big time. This paper has not defined a “pathological mechanism” for autism. Their animals didn’t even show differences on an autism-related test. At most, they’ve identified a link between SHANK3 and intersexual behaviors, differential brain structure, and nerve communication.
Furthermore, researchers are nowhere near testing drug compounds to target SHANK3 for autism. Evidence for a connection between SHANK3 and autism is sketchy, at best. While this gene shows some association with autism in a few studies, it certainly is not the autism gene. Indeed, there are likely many, many genes associated with the spectrum of behaviors and other signs of autism. SHANK3, if it is related to autism, is simply one of many, many possible candidates. This study has not changed that.
The freeway proximity epidemiological study is on even shakier ground. Such studies simply identify associations, using data to identify links between different types of information. This particular study, published in the open-access journal Environmental Health Perspectives, examined associations between autism rates and proximity to highways. The authors report that proximity to freeways during the third trimester of development was associated with twice the risk of autism, while proximity to freeways at birth was associated with just under twice the risk. The authors suggest an examination of associations with air pollutants.
And therein lies the problem with epidemiology — the gap between a numbers game of correlation and real cause-effect. Why didn’t the authors access some of the available data on air pollution in their areas of study? Having done so would have provided a much more solid rationale for even suggesting it.
Oddly enough, the authors found no correlation between autism and living near other busy highways, or “major roads” as they refer to them in the study. Only freeways. More confusing, they define freeway “as a state highway or interstate highway,” and a major road “as a state highway, interstate highway, or major arterial.”
Even for people living very close to a “major road” — as close as 42 meters — the autism rates weren’t greater. Only for people living 309 meters or closer to a “freeway” showed this association. That’s a bit of a problem, as it undermines any conclusion that proximity to a busy highway — you say Interstate 5, I say US 99 — might be associated with autism rates. If there is something busy-highway-specific about autism … wouldn’t one expect that any busy highway would show the association? And wouldn’t being closer to a busy highway kick up the connection?
They report no data on air pollution, even though suggestion that air pollution might be a factor could be fairly easily checked. If air pollution were causative in autism, then an association between bad air and autism rates should emerge in the world’s worst-polluted areas. The authors don’t comment on this and don’t provide any data.
In other words, this study seems dodgy even for epidemiological studies. If you live near a freeway (but no worries about major roads!) and are pregnant, there’s likely little need for you to start to make relocation plans.
For a more-detailed look at this paper and a closer analysis of the findings, I highly recommend David Gorski’s take over at Science-Based Medicine. He notes another glaring absence from the study: A consideration of population density and urbanicity as confounding factors. In other words, the authors failed to include these in their analyses. Either of these factors might be the true link to autism rates in this study, with freeways merely serving as, well, the road that connects them.
As always, we encourage readers to look behind the headlines when autism is involved. -Eds