Can eating broccoli cure autism (or fragile X)?

A fascinating new study is getting a lot of attention lately, so I decided to give it a closer read. A group of MGH, Harvard, and UMass autism researchers tested a broccoli sprout extract containing the antioxidant and apparent active ingredient sulforaphane in a double-blind, placebo controlled trial (read the full text here).

The first caveat regarding this study is that the number of subjects is quite small; indeed, there were only 14 completers in the placebo group (using a 2:1 drug:placebo ratio), which could give rise to all kinds of misleading things. One thing I noticed was that there was no appreciable placebo response. An excessive placebo response can doom a trial, but when you see no placebo response at all, a red flag should go up in your mind. In small studies like this, the superiority of drug over placebo can result from an unusually small placebo response---a statistical fluke in the randomization process, really. If we then compare the placebo group showing no response to the drug group showing a typical placebo effect, but no actual treatment effect, it can appear that there is a big difference. This is essentially what happened in the first Novartis trial; the seven fully methylated subjects just happened to show no placebo response at all; in the larger study of the drug, this effect disappeared (in fact, the fully methylated group had an extra-large placebo effect.)

It was somewhat reassuring to see that ABC scores in the treatment group increased significantly 4 weeks after discontinuing, while still blind to treatment status, though there were numerous dropouts at this point, complicating interpretation. This kind of on/off effect is what you like to see, and is generally indicative of a true drug effect (of course it also means there is little carry-over effect, but most drugs do stop working when you stop taking them!) The effect of the broccoli sprout extract was significant, resulting in a 20+ point drop in ABC scores (note that the outcome measures were all the same as those used in recent fragile X trials.) However, the placebo response in the Novartis Phase IIb/III fragile X trials was actually quite similar in magnitude, so this could all be a statistical quirk.

Nonetheless, this is an intriguing result, especially since broccoli sprout extracts are widely available as nutritional supplements. But what about dosage? Can you actually get the same stuff used in this trial, and how much would you need? The study drug was a custom preparation which is not available anywhere, and the amount of sulforaphane given each day in the trial would be the equivalent of 20 or more capsules of the commercially available broccoli sprout extract, by my calculations. In addition, the potency of the extract was carefully monitored and maintained, implying that the compound is not entirely stable, and the pills you get at GNC might not even be as potent as they say (always an issue with unregulated nutritional supplements.) Still, 20 pills a day is possible as a treatment strategy. At this point, the evidence seems a bit weak, so I'd recommend waiting before trying this, but keep an eye on the broccoli story!