Clinical study addresses Fragile X, not autism. News release hints otherwise

Another news release mentioning autism for a study that doesn't address autism
I am writing this post before the embargo on an upcoming paper in Science Translational Medicine on a treatment for Fragile X lifts. That means that news outlets subscribed to receive science-related news releases from universities will have had the news release and related research papers in-hand for 2.5 days after I've written this. I expect to see the news releases themselves--and their headlines--linked as-is on FB, on autism community pages, and on Twitter. [ETA, 4:14 EST: If you doubted that prediction, see here; now updated here, 9.20.12. EST: 7:16--One jaw-dropping example is this USA Today piece, which uses the word 'autism' 17 times but the term 'Fragile X 11' only times in an article about a clinical trial for a drug for Fragile X. It describes Fragile X in the headline as a "rare disease" without mentioning it by name, while specifically mentioning autism. For a great example of what makes a good article about this paper, see this SFARI writeup by Karen Weintraub.]

There are three related news releases, two from universities and one from a pharmaceutical company. The two university releases, entitled respectively, "Medication effective in treating social withdrawal in Fragile X and potentially autism patients" and  "New targeted drug for treating fragile X syndrome, potentially autism, is effective," open with similar ledes (one reproduced below):
An investigational compound that targets the core symptoms of fragile X syndrome is effective for addressing the social withdrawal and challenging behaviors characteristic of the condition, making it the first such discovery for fragile X syndrome and, potentially, the first for autism spectrum disorder, a study by researchers at Rush University Medical Center and the University of California, Davis MIND Institute has found. 
The finding is the result of a clinical trial in adult and pediatric subjects with fragile X syndrome. It suggests, however, that the compound may have treatment implications for at least a portion of the growing population of individuals with autism spectrum disorder, as well as for those with other conditions defined by social deficits. 
A look at the paper reveals that the results are a tad more modest than these paragraphs imply and that the authors do not mention "and potentially autism." Given that this trial was in people with Fragile X, not autism, whoever wrote the headlines for these releases appears to have tacked on autism for the click value alone. Adding "and potentially autism patients"? Gratuitous. The authors don't mention this potential extrapolation in the abstract or in the discussion. In the end, though, that is odd. I'll tell you why in a mo'.

What they really found
The results of this trial (listed at here) are rather modest. Before anyone takes that as a criticism, I'll add that rather than reach for the unrealistic goal of "global cure for everyone!", it's much more realistic to target mitigation of gaps in a neurobiological condition with an understanding that even that mitigation won't work on everyone. This study might have identified a gap that in some cases, the drug mitigates. 

The authors report no improvement at all in the primary outcome measure, which was irritability. In fact, a look at their numbers shows that the placebo versus treatment measures for irritability were practically identical (end of treatment, 16.4 for the drug and 16.2 for placebo). 

They found improvement in what they call "social avoidance," a new subscale of the Aberrant Behavior Checklist (known as the ABC) that some members of the current author group developed based on a population of people with Fragile X and published this summer. In the abstract of the paper reporting this new subscale, the authors note, "Reformulated ABC-C scores based on this FXS (Fragile-X syndrome)-specific factor structure may provide added outcome measure specificity and sensitivity in FXS clinical trials." That seems to have been the case here, as this subscale is the only potentially objective clinical measure that showed any change related to treatment for the overall population. 

Regarding the social avoidance findings, the baseline scores for the new social avoidance subscale (a 7-point scale) in the treated group were higher than those in the placebo group (4.5 vs 3.9) while final values were almost the same (3.3 vs 3.6 in treated versus placebo). 

And finally, for the overall group, the investigators identified a very modestly significant (p=0.04) improvement in parent-reported "problem behaviors," as rated by parents using a visual scale and based on what parents identified as the top three (unspecified) most problematic behaviors. These behaviors, as "nominated" by parents, included aggression and outbursts, anxiety, self-injurious behaviors, and stereotyped behaviors. The placebo group showed some improvement (from baseline of 1.9 to 3.1 at the end of the study); the treatment group again started at a higher level than the placebo, at 2.2 and improved, per parental report, in unspecified behaviors to 4.2. I believe these visual scales usually run to 10.

In the same paper that the authors published about establishing a new "social avoidance" subscale for the Aberrant Behavior Checklist, they also found that scores of 8 or greater on the lethargy/social withdrawal subscale encompass the upper half of severity in males with Fragile X. Those who fall below this score, they say, have less social impairment. Reasoning in the paper that this relative lack of social impairment might dilute beneficial effects of arbaclofen on social impairment, they did one of many, many of the study's subanalyses on only those participants whose scores in this subscale were 8 or higher. In this group, treatment was associated with significantly reduced social avoidance scores and improved socialization scores.

Finally, the authors report a couple of other things worth noting. In these studies, the clinicians and parents can state which treatment period seemed to be having an effect, although they don't know whether the treatment was the drug or the placebo. Twice as many clinicians (26 of them) selected the drug-treatment periods as being linked to an effect, compared to 13 who chose the placebo periods as effective, but this difference was not significant. The split was about the same for parental perception of effectiveness, and again was not significantly different. 

What they did
The drug they're trying out is called arbaclofen. It's one of two possible molecular conformations of the drug baclofen, which serves to address a variety of medical problems, mostly related to relaxing muscles. Baclofen is a mix of both conformations, but arbaclofen is converted only to the conformation that has the desired effects in the body, giving it a better profile as a medication. Those effects are related to restoring a balance in the conversational give-and-take among nerve cells, which is hypothesized to be out of whack in people with Fragile X and possibly some people with autism. The Centers for Disease Control and Prevention says that about 46% of males with Fragile X also have an autism diagnosis, so what underlies Fragile X doesn't necessarily always also lead to autism. 

Mediating this nervous system conversation are two proteins, one called GABA-B that works to muffle the noise level and another called mGluR5 that works to turn up the volume. Fragile X syndrome arises when cells no longer make another protein, FMR1, that turns down the volume in this neural noise. FMR1 might do that by telling mGluR5 to shut up; without it, mGluR5 gains the noisy upper hand. Drugs like baclofen and arbaclofen kick up GABA-B's activity, bringing some calming balance back to the neural conversation. 

This clinical trial thus rested on the idea that arbaclofen could work through GABA-B and dampen neural noise, which would be presumably manifested in the Fragile X population as reduced irritability. The research group included members from Rush University Medical Center in Chicago, the MIND Institute at UC-Davis, and Seaside Therapeutics, a pharmaceutical development company focused on clinical trials for Fragile X and autism, specifically targeting GABA-B and mGluR5. 

They enrolled a total of 63 people (eight females), ages 6 to 40, with Fragile X syndrome. Most were taking some form of psychoactive medication. This trial had a randomized, double-blind, placebo-controlled, multisite, two-period crossover design. That means that participants were randomly assigned to receive either the drug or placebo first, followed by a break for both groups, and then a switch of the groups, so that the placebo group from the first round got the drug in the second round, and vice versa. That way, all trial participants ended up receiving placebo and drug or drug and placebo sequentially. Neither participants and families nor investigators knew who had received what, so the trial was double blind. It took place at 12 centers in the United States (hence the "multi) from December 2008 to March 2010.

As the study progressed, some participants dropped out so that overall analyses involved fewer than 63 people. Also as the study progressed, researchers increased the drug dose as tolerated every 3 or 4 days, from 1 mg twice a day to a maximum of 10 mg three times a day. Participants took their first round of drug or placebo for four weeks, tapered off of their intervention for a week or two (minimum of seven days), and then started their second treatment period.

Safety, tolerability, efficacy
Otherwise known as, "Did it hurt more than harm, bother more than benefit, and if beneficial, how much?" The description of this study is that it was a "safety, tolerability, and efficacy" study for arbaclofen, also known as STX209. Events that emerged during treatment in 5% or more of participants included headache, sedation, fatigue, irritability, diarrhea and vomiting, aggression, and upper respiratory tract infections. Note that just because these events happened during treatment doesn't mean that treatment caused these events. Baclofen (the mixed version of this drug) is known for causing some sedation and muscle weakness. 

Three people dropped out of the study because of "increased irritability" during the "washout" or treatment-break period. Worth noting, the authors say that use of antipsychotic medications in the study also may have blunted the effects of STX209 on irritability, the primary endpoint, because antipsychotics are known to diminish such symptoms in autism. 

As noted, the authors found no improvement in the primary endpoint, which was irritability. This finding contrasts with reports from Seaside Therapeutics for trials of arbaclofen in autistic populations that arbaclofen [in a news releasePDF] appeared to reduce irritability in autistic people in a phase 2 trial. That company also is a funder of this current trial for Fragile X. These conflicting outcomes in a Fragile X versus autism population simply underscore that just because autism can be comorbid with Fragile X, 'autism' and "Fragile X' are not interchangeable. 

The authors performed a lot of subanalyses in this paper, but one subanalysis that they did not report was examining the autistic subset in this population. They had one; 37 of the 63 participants met the criteria for comorbid autism. That's more people to analyze than there were in the subanalyses they did for scores 8 or greater, so I can't see why they didn't do those, too, particularly given the consistent conflation of autism and Fragile X, the interest in targeting similar pathways between the two conditions, and the studies evaluating the effect of STX209 in autism. If they'd done and reported such an analysis, at least the news releases would have been more accurate. [I emailed the media contact at UC-Davis with this question on Monday. She emailed me back pretty quickly, asking where this piece would appear. I replied immediately, and two days later have not gotten a response to my question.]

There are several ongoing trials for STX209 listed right now, including a safety trial for arbaclofen in autism, which hasn't been verified/updated in two years. The current outcome of interest appears to be social withdrawal, and I wonder if these findings with the new subscale of "social avoidance" won't result in some more post-hoc (after the plan) analyses for these studies.

Finally, for anyone's information, below is a summary of the author contributions and conflicts of interest for this work. If you find that I'm being too rational about or supportive of the study findings, please don't bother calling me a "pharma whore" in the comments. It's been done. For sensitive readers, a snark alert: I can't help one aside, and that is, "Wow! That's a lot of people overseeing the execution of a study!" and I love that the co-founders of Seaside Therapeutics provided the "motivation." [Sorry, I'm just imagining the struggle people who hate pharmaceutical companies might be having right now with information that something with any clinical efficacy originated with one. The horrerz]. 

Funding: This study was sponsored by Seaside Therapeutics Inc. Author contributions: E.M.B.-K., B.R., P.Z., M.C., J.G.-H., R.L.C., and R.J.H. designed the study. B.R., P.Z., M.C., K.W.-B., P.P.W., and R.L.C. oversaw the execution of the study. E.M.B.-K., D.H., and R.J.H. led the subject testing and data collection. K.W.-B., Y.M., and D.V.N. led the statistical analysis. R.L.C. and M.F.B. provided the original insights and motivation for the study. All authors contributed to writing and revision of the manuscript. Competing interests: P.Z., M.C., K.W.-B., P.P.W., and R.L.C. are full-time employees of Seaside Therapeutics. R.L.C. and M.F.B. are co-founders of Seaside Therapeutics. J.G.-H. has consulted for Seaside Therapeutics. E.M.B.-K., D.H., and R.J.H. were compensated for their data collection efforts in this study.