Saturday, July 19, 2014

What to make of the clinical trial failures in fragile X

To summarize: the official results from both the adult and adolescent trials of AFQ056 for fragile X were decidedly negative. Not only was there no improvement on pre-designated outcome measures, but subjects did a bit worse on the drug than on placebo (and placebo effects were robust, but in line with other trials in developmental disorders.) However, it is important to note that trial subjects got better on the drug, just not as much better as on placebo. Extensive analysis of the trial results after the fact ("post-hoc analysis") revealed no obvious differences in clinical responses between the different drug dosage groups and placebo. Subjects with a fully methylated fragile X mutation didn't respond any better than those with a partially methylated mutation, more severely affected subjects didn't respond any better than less severely affected (or vice versa.) So, the drug simply didn't work in any way that anyone can see or measure objectively, though many families reported excellent results, and most of the subjects in the open label extension continued on the drug, apparently satisfied with the drug's effect.

Our personal experience, which may have been a bit aypical because of co-administration of minocycline, was much more positive than the official results. Even so, the great results we saw at the beginning of the trial, and then during the first few months of the open label extension, just didn't last. We think there is still some beneficial effect at this point, but we were certainly hoping for continuous improvement.

The most important point is that this isn't just a problem with outcome measures. It would be great if we could attribute this negative outcome entirely to inadequate outcome measures. It's still possible, though increasingly implausible, to claim that the drug is actually improving cognition and other important developmental outcomes which aren't captured by the outcome measures, while at the same time having little beneficial effect (and even some adverse effects) on behavior. However, that isn't what we saw at all. We saw clear improvements in mood and behavior (not to mention reflux), and our son's ABC scores plummeted (before creeping back up later.) We have not seen any major leaps forward in development, or dramatic improvements in cognitive function. Now, improved behavior and mood can make everything seem better, at least for a while, and this may be what some families were observing. When people feel better, they generally do more and expand their range of activities. But this effect won't alter developmental trajectories. In other words, it won't be truly disease-modifying unless there is a sustained effect on all major fragile X symptom domains.

Where does this leave us, then? Well, first of all, mGluR5 antagonists probably won't be a viable treatment for anything if tolerance develops to this extent, and we've seen tolerance develop to the anti-reflux effects as well as the CNS effects. It's no coincidence that so many of the mGluR5 programs around the world have failed and been discontinued. This implies that this target many not be viable for fragile X therapeutics, although it is possible that trials of higher doses of mGluR5 antagonists, perhaps in combination with other drugs, could still be effective. However, this greatly complicates the "re-purposing" strategy, since most drugs are developed with a particular dose range in mind, and most drug companies would need to go back and repeat this process specifically for fragile X. It is not likely that any company will be enthusiastic about incurring this expense in light of recent experience.

We have many more thoughts about where the fragile X research field needs to go, and we'll discuss those ideas in future posts.

Monday, June 30, 2014

Further thoughts on our AFQ experience

So, the Novartis trials ended with a rather negative result. But we did see a lot of positives in our son along the way. Could it be that there was something about our case that made a response more likely? Well, I do think that the combination of minocycline and AFQ056 was probably much more effective than just AFQ alone. The preclinical evidence for minocycline as a disease-modifying agent in fragile X is very strong, and the clinical trials done to date, though probably not rigorous enough to be considered definitive proof of efficacy, have been positive (unlike some others!) It stands to reason that a combination of two drugs would work better, given that fragile X is a disorder that affects the brain globally. In other words, one drug might work well in one area of the brain, but not fix everything; another drug might work well in other areas or other pathways, so that the combo gives greater "coverage" and a better clinical response, especially with the rather crude outcome measures we're using. It would be interesting to see if trial subjects on minocycline had any better pattern of response, but I doubt there were enough subjects on the combination to analyze properly (especially since many of the subjects were outside the US, where minocycline use in fragile X is much less common.)

One question I've been asked a few times is, if there is tolerance developing to mGluR5 antagonists, why don't the rating scales from the trials show that? Why don't we see the scores go down, then back up? Of course, all of the above has an effect---most subjects aren't getting enough drug effect to notice in the first place. But this is also where the shortcomings of the outcome measures may come into play. The resolution of these rating scales, especially in the context of moderately large placebo responses, is simply nowhere near good enough to quantify tolerance (although, I should note that our scores did go way down, then back up a bit toward the end of the study; I believe our son's clinician ratings also showed this.) It's also quite possible that there are sub-groups of fragile X subjects who are particularly responsive to one drug or another. We've had friends who participated in both the arbaclofen trial and one of the mGluR5 trials, and some have had terrible experiences with one, but great effects from the other. Ideally, you'd like to be able to predict who would do better, either with a biomarker or via phenotypic features, but there's nothing available right now to do that. Indeed, we are caught in a bit of a Catch-22 scenario: we need to have a clinical trial success to demonstrate which outcome measures, biomarkers, and endophenotypes are most useful, but we need to have some of those things first to have successful clinical trials!

Perhaps one solution is to have pilot trials with combination treatments, as a way of getting more robust and reliable effects from the drugs. This would allow for better study of outcome measures and biomarkers, and it would allow us to figure out which ones are actually drug-responsive. It may be unrealistic to expect that any one drug can move the needle far enough to be measurable, but a two drug combo may have a better chance of success.

Wednesday, June 25, 2014

Novartis trial results are in, and they're not pretty

This year's Gordon Conference just finished,and Novartis presented their results for the first time (though advisors and advocates had been given a private peak months ago.) To say that the trial results for AFQ056 were disappointing would be the understatement of the century! While the company has already announced that the adult and adolescent trials failed to meet their pre-designated endpoints, the numbers looked really bad. This wasn't a case of the drug working, but placebo effects leading to an outcome that wasn't statistically significant; in this case, the effect of the drug was statistically significant, but in the wrong direction!

So, what went wrong? The evidence for using mGluR5 antagonists in fragile X was really strong going into these trials---in fact, about as good as it ever gets. The drug itself was an advanced compound that had been studied extensively. Of course, outcome measures have been a problem, and they're definitely not good enough. However, Novartis did an exhaustive analysis of all the data, plus they even went back and looked again at subjects who were reported by clinicians and caregivers as exceptionally good responders. None of this "post hoc" analysis revealed any positive effect of the drug compared to placebo---not even a glimmer of hope in adults or adolescents (and both trials came up with very similar results, providing some degree of cross-validation). The ABC and other outcome measures used might leave much to be desired, but they weren't the reason for these trial failures. Indeed, the trials were not "failures" in any sense, except that the results were decidedly negative. Novartis conducted an excellent pair of clinical trials which showed convincingly that AFQ056 doesn't work for fragile X, as least not as administered in these trials.

Perhaps the Roche compound will show much better efficacy; it is a different drug, and somewhat different methods are being used in the Roche trials. But our personal experience may offer some insights into what is going on, and this also leads me to be a bit pessimistic about the chances for mGluR5 antagonists in general.

Our 25 year-old son was one of the last adults enrolled in the AFQ trial; I should note that he was also on minocycline during this entire time, as he has been for many years. This is thought to have potentially disease-modifying properties for fragile X, and so this may have boosted the effect of the AFQ. His other meds are valproate, topiramate, and sertraline. He also has a lot of severe gastro-esophageal reflux, which means he spits up his food a lot if he eats too fast (it seems to be reflux, which is a physical effect based in the stomach and esophagus, rather than nausea, which is a centrally mediated process controlled by the brain.) In any case, when he started the blinded phase of the trial, we saw a rapid improvement in his mood, language, social function, and basically everything else. But we only saw one "bump" up during the dose titration, which led us to think he was on the lowest dose of the drug (this was later confirmed.)

Unexpectedly, his reflux went away entirely along with all this other improvement; I should also note that this was all after the 4 week placebo run-in (we'd been tipped off about that long ago!) But this is a guy who had never in his life gone for a whole week without barfing; this was a regular, daily event for us. Our son had not one single episode of reflux for more than one month after starting on that tiny dose of AFQ! Call it a side effect, or a bonus, but it was something very objective and hard to ascribe to placebo. Indeed, mGluR5 antagonists are known to increase tone in the lower esophageal sphincter, and have been studied as treatments for reflux. The first 6 weeks or so were great---we started doing all kinds of new things, like sailing and traveling, dining out and going to lots of new places. This, I thought, was what we had been waiting for! But then the effect started to wear off, little by little. The reflux came back, though only a bit here and there.

We figured this was no big deal, that in the extension phase we'd get to a higher dose and that would work much better. Which turned out to be true! When we started the open-label extension, our son got noticeably better with each dose increase, and tolerated the full 100 mg twice a day with no side effects to speak of, except the good one (no reflux at all!) This kept up for several months, and we were very happy. We assumed that things would just keep getting better and better as long as we kept up the AFQ. However, after a while (hard to say exactly when, because I think it was very gradual), many of the old behaviors started creeping back in, as did the reflux (in fact, they seemed bound together quite tightly.) As I write this, about 9 months after starting the full dose of AFQ, I'd estimate we've lost about 70-80% of the effect of the drug. We're probably not all the way back to baseline, but we're not finding that things are getting better and better all the time.

Now, even I realize that the placebo effect is very powerful, and that we are hardly objective observers in this case. I could easily be imagining improvements in mood or behavior, or we may have treated our son differently because of our expectations, leading to better behavior. But I can't explain the reflux as anything other than a real, physical effect. He ate the same food, the same way as ever. The mGluR5 antagonist simply helped him keep it down---it's known to do that. But even that effect wore off! Tolerance developed to that physical effect at approximately the same rate as the "central" effects that were the primary objective in this case.

So, in a nutshell, the problem we've seen, and the likely reason for the trials to fail, has been the development of extensive tolerance. This is quite unexpected, it's probably something intrinsic to mGluR5 in fragile X, and something I'll discuss more in upcoming posts.

Wednesday, March 20, 2013

It’s not just mGluR5…there’s more than one way to fix fragile X

In case you somehow got the impression that FRAXA is only interested in mGluR5 research, or that we’ve quit funding basic research altogether, the most recent edition of the Proceedings of the National Academy of Sciences (PNAS) illustrates just why that’s not true.

Long after the seminal finding of increased mGluR-LTD in fragile X by Kim Huber and Mark Bear, FRAXA funded studies by Nobel-laureate Susumu Tonegawa investigating the role of the enzyme PAK in regulating the shape and function of dendritic spines. The results of this FRAXA-funded project demonstrated that genetic reduction of PAK could rescue most fragile X knockout “phenotypes” (the animal model equivalent of human symptoms.) This was essential proof of concept work which established PAK inhibitors as potential therapeutics for fragile X, and was the basis for a new company called Afraxis. The only problem was that no specific PAK inhibitors actually existed at that time, but Afraxis has been busy trying to find useful drugs that inhibit PAK (actually a family of enzymes, not one single enzyme.) The paper just published by Tonegawa and his collaborators from MIT, Afraxis, and universities in India and Korea, shows that a new compound which Afraxis calls FRAX486, inhibits all the PAKs in the brain at relatively low doses, and displays all the properties you’d expect of a useful drug. Most importantly, it appears to reverse most of the important animal model phenotypes with a single dose (though, presumably, daily dosing would work better as a human therapy.)

Read the whole thing:

At this point, FRAXA has developed a large pipeline of drug candidates for the disease-modifying treatment of fragile X. We continue to fund basic research looking for more promising leads, but our main emphasis is on developing treatments based on the work we’ve already funded, which has given us plenty of candidates for therapeutics that can alter the course of fragile X. We’re excited about our partnerships with companies big and small, like Afraxis, because these partnerships are essential to bring new treatments to fragile X families in need.

Sunday, December 2, 2012

An internet friend asked for my feedback on the following article:

Primum Non Nocere: An Evolutionary Analysis of Whether Antidepressants Do More Harm than Good


Antidepressant medications are the first-line treatment for people meeting current diagnostic criteria for major depressive disorder. Most antidepressants are designed to perturb the mechanisms that regulate the neurotransmitter serotonin – an evolutionarily ancient biochemical found in plants, animals, and fungi. Many adaptive processes evolved to be regulated by serotonin, including emotion, development, neuronal growth and death, platelet activation and the clotting process, attention, electrolyte balance, and reproduction. It is a principle of evolutionary medicine that the disruption of evolved adaptations will degrade biological functioning. Because serotonin regulates many adaptive processes, antidepressants could have many adverse health effects. For instance, while antidepressants are modestly effective in reducing depressive symptoms, they increase the brain’s susceptibility to future episodes after they have been discontinued. Contrary to a widely held belief in psychiatry, studies that purport to show that antidepressants promote neurogenesis are flawed because they all use a method that cannot, by itself, distinguish between neurogenesis and neuronal death. In fact, antidepressants cause neuronal damage and mature neurons to revert to an immature state, both of which may explain why antidepressants also cause neurons to undergo apoptosis (programmed death). Antidepressants can also cause developmental problems, they have adverse effects on sexual and romantic life, and they increase the risk of hyponatremia (low sodium in the blood plasma), bleeding, stroke, and death in the elderly. Our review supports the conclusion that antidepressants generally do more harm than good by disrupting a number of adaptive processes regulated by serotonin. However, there may be specific conditions for which their use is warranted (e.g., cancer, recovery from stroke). We conclude that altered informed consent practices and greater caution in the prescription of antidepressants are warranted.

Read the whole thing at:

It's always interesting to read and critique articles like this, because it gives me a chance to think about the Big Picture---in fact, I think I'll expand on this and make it into an extra-long blog post. I guess this article was thought-provoking; it certainly got my juices flowing, mainly because a lot of people ask me questions related to the points made in this article, and I think it’s important to answer these questions, because it’s easy to scare people away from the medical treatment they need. Some might say I'm biased, because I prescribe drugs; however, I've also done many thousands of hours of psychotherapy, too.

As an overview, this article is typical of its genre, and can be seen as part of the ongoing psychology/psychiatry turf war. It's not really just science, but also an opinion piece from people with an ax to grind, and probably some vested interest. The article is of poor quality, and the authors clearly do not understand many of the basic neuroscience topics they are purporting to review; many of the references are review articles, not original studies, and rely heavily on their own previous review articles and on those of co-authors who clearly run in the same circles. This is OK when you're in a small field, but this is not a small field. They are using a tiny fraction of the literature to support an opinion (that antidepressants do more harm than good) which is absolutely not supported by the weight of the available evidence. They also have an annoying habit of stating something non-controversial (i.e. serotonin is in platelets) and referencing that extensively, then asserting something outrageous without any reference at all---the journal reviewers shouldn’t let them get away with this, so this speaks to the quality of the journal.

It is always true in the life sciences that there is some contradictory evidence, no matter the subject. We don't know everything, and our concept of how the brain works will be evolving for a long time to come---certainly for the rest of our lives, and then some. We cannot be paralyzed by what we don't know when there is real human suffering. Depression, anxiety disorders, and other associated conditions which are commonly (and very effectively) treated with these medications exact an enormous human toll, and if we're going to wait until we understand everything about them before we start treating, we're going to be waiting for a very long time. We do the best we can with basic neuroscience research, translational research on disease mechanisms, preclinical research in animal models, and then carefully controlled clinical trials. The science is never settled, and our understanding is always subject to review (remember that when you hear other scientific issues discussed in the popular press.)

One of their main points is that there is a carefully regulated and highly evolved homeostasis in all these systems (i.e. serotonin), and that anything that perturbs it must necessarily cause dysfunction. You could say this about any medical treatment for any condition, so I guess all of modern medicine is suspect from this point of view (with the possible exception of antibiotics, but they'd probably argue that perturbing normal flora is dangerous, too.) The point they're obviously missing is that the system is already perturbed, and that's why we have a disease state. Some of the authors have previously written that depression (and by extension, just about any mental illness) is just a normal evolutionary response; that may or may not be true, but we haven't evolved much at all (genetically speaking) as a species in the last 10,000 years, while our lifestyle has changed radically. The same is true for the extensive discussion of withdrawal and discontinuation effects in the article; while it's certainly true that many patients relapse when antidepressants are discontinued, that’s also true for virtually all medical treatments of chronic (i.e. non-infectious) diseases. When patients stop taking antihypertensives, blood pressure rebounds---often higher than ever. If an epileptic patient stops taking seizure meds, more seizures are likely. If a diabetic stops taking insulin, they’re still diabetic, and their blood sugars go up. All of this is expected, yet the authors seem to proclaim that they have discredited antidepressants with this revelation.

The surprising thing is how few patients actually relapse---if it’s done right, most people are able to discontinue their antidepressants and have something like a lasting “cure”. Of course, the devil’s in the details, and it’s often not done right. It would be nice to have more research in this area, and much is underway; in the meantime, good clinicians have figured out the best methods for successful discontinuation. My clinical pearls: treat for at least a year at the optimal dose (longer if the depression is more chronic), then decrease the dose very gradually, and do it at a time of year that’s good for the individual patient (often, Spring is the best time of year.)

They really get bogged down when talking about neurogenesis; they mistakenly assert that neurogenesis is widely considered to be the mechanism of action of antidepressants; this is one theory, but not the consensus or established view by any means, so I'm not sure why they spend so much time and effort trying to poke holes in one theory of antidepressant mechanism of action. Anyone who’s ever spent any time thinking about this has come to the conclusion that the antidepressant response is essentially an alteration of gene expression (one manifestation of which might be increased neurogenesis, but it’s hard to imagine that’s the full extent of it.) All the different monoamine manipulations caused by available antidepressants are likely just means to an end, different ways of altering gene expression in the brain. It is no accident that hormonal alterations like increased cortisol or decreased thyroid hormone, which are thought to cause some cases of depression, are also known to cause widespread alterations in gene expression. There are probably many ways to get depressed or anxious, and the available treatments are just ways that we’ve found to (partially) reverse those processes.

The authors argue that the evidence for enhanced neurogenesis with SSRIs and other antidepressants is equivocal. Since there are hundreds, even thousands, of studies demonstrating this phenomenon in many different ways (not just the one they critique) and a relative handful that show a negative result, you could call that "equivocal", but I would call it a preponderance of evidence. Remember, there is always conflicting evidence in science! They speculate that if neurogenesis is increased with antidepressants, you should see more brain tumors. This is incredibly silly, and something they just made up (no reference at all, because it just isn’t true!) Neurogenesis does not lead to neoplasia (cancer) at all---those are two entirely distinct phenomena, and anyone with any medical training would know that. If neurogenesis is increased, you would expect to see more seizures with antidepressants---and you do! All known antidepressants result in a small but significant increase in seizures in trial populations; the risk isn’t great, but it’s there, and it’s probably a direct manifestation of increased neurogenesis. The interesting thing is that antidepressants vary widely in their seizure risk, but this does not correlate with antidepressant efficacy. SSRIs are low-risk, while other drugs like maprotiline (Ludiomil) and bupropion (Wellbutrin) are clearly much higher-risk. This suggests to me that neurogenesis is a sideshow, and not the primary antidepressant mechanism, but that’s just my opinion.

The article also spends a lot of time emphasizing the point that antidepressants just don’t work as well in clinical trials as most folks might imagine, and that there have been many trials (often unpublished) where the antidepressant fails to separate from placebo, or has a very small effect size. To some extent this is true, especially for those who don’t really understand drug trials, and just how difficult they are to conduct. What happens when a clinical trial isn’t properly designed and conducted? The result is that improvement on drug is not statistically superior to placebo; note that this does not mean that the trial drug doesn’t work. It could mean that, or it could mean that the trial simply wasn’t well done, for any number of reasons. However, it is exceedingly unlikely the drug would show superior efficacy compared to placebo by chance, or because of poor methodology. So, it’s a kind of one-way street: virtually all problems (other than outright fraud) lead to a result not significantly different from placebo. Positive trial results are important, and they are accorded great significance for just this reason---they are very hard to come by, because the system is heavily biased against a positive outcome in any double-blind trial. Of course, there are limits to this. If a drug goes through 20 trials, and only one shows significant efficacy compared to placebo, then it’s probably not a very effective drug. But if a drug is in 20 clinical trials, and only 12 are positive, that’s not bad (and fairly typical in psychiatry.) That is an effective drug, and the FDA understands that. It’s also important to understand that most subjects in clinical trials are chronic, difficult, treatment resistant patients; it is rare to find drug-na├»ve subjects in any psychiatric clinical trial. So, critics of antidepressants are always fond of citing big meta-analyses, where results are pooled from all the available studies, good and bad. This invariably diminishes the effect size and increases the placebo effect compared to the more successful, well-done studies. Psychiatric research is just like any other human endeavor, some of the people doing it are incompetent, and their efforts muddy the waters for everyone else.

In the real world, we tend to see much higher response rates among the general population. For example, in my practice of young and middle-age adults in a private outpatient setting, I generally found a response rate to SSRIs for depression and anxiety disorders of about 90% in patients on their first med trial---well in excess of the usual 70% response rate in most clinical trials. These were real, life-changing responses to treatment, with no “poop-out” after a while. Placebo responses never last, so I’m confident that most of these were real therapeutic effects. My success rate in discontinuing the antidepressants in this category of patients was about 60%, using my conservative methods; the unsuccessful 40% went back on the same medication, at the same dose, and virtually always did fine until the next time they wanted to attempt drug discontinuation.

This paper also includes a rather histrionic recitation of side effects of meds, focusing on SSRIs. Of course, all medications have side effects; it’s really surprising how many side effects are reported even in placebo-treated subjects. But most people on SSRIs don’t have major side effects, and many never notice anything at all; the body does adjust, so that the reported side effects usually dissipate as treatment goes on. Here again, this is not really different from what we see with any medical treatment. The authors go into gory detail about the risks of very rare side effects like hyponatremia (I’ve never seen a single case with any antidepressant---I have seen a bunch with the anticonvulsant Tegretol), and try to demonstrate that SSRIs cause cognitive impairment (there’s no evidence of it, and it has been tested directly.) Throughout, they make the most common mistake people make in this situation: they compare the rates of various problems in depressed people treated with antidepressants to the rates seen in normal untreated people. The problem is that depression causes many of these problems, like cognitive impairment, suicide, and excessive cardiac deaths. So, it’s hard to separate the symptoms of the disease from the side effects of the treatment.

I would also note that psychotherapy is not without side effects, especially since the authors seem to have the agenda of promoting psychotherapy. Now, I trained extensively in long-term, psychodynamic psychotherapy; I’ve done lots of it, and I’m very much in favor of it, as well as newer focused psychotherapy approaches. But I’ve also seen many patients (and some personal friends) devastated by poorly done or inappropriate psychotherapy. Drug side effects are usually easy to fix---just stop the drug. The kind of damage inflicted on these folks by incompetent psychotherapists can never be undone, not to mention the financial ruin and exploitation that went along with it. Most people assume psychotherapy has no side effects because they also assume it doesn’t do anything at all. Neither assumption is true.

As a final note, there have been several long-term studies of brain volume in people with depression (and similar studies in OCD). Having depression shrinks your brain on CT or MRI, and treatment with antidepressants (doesn’t seem to matter which ones, as long as they’re clinically effective) prevents this. This is a very powerful argument against most of the pseudo-biological speculation offered in this article, yet the authors gloss over these inconvenient facts entirely. Depression is very bad for your brain; antidepressants clearly help. Are they perfect? No, but their benefits clearly greatly outweigh their risks in the vast majority of people with serious Mood Disorders and Anxiety Disorders.

Thursday, September 20, 2012

2 New Arbaclofen Studies Getting Lots of Press!

Two complementary studies were just published in Science Translational Medicine, and they’ve been getting a ton of attention in the media. Since lots of people are asking about these articles, and lots of reporters are calling to get my comments (and subsequently misquoting virtually everything I tell them,) I thought I’d give my official commentary here. .................................................................................................... The first article (Henderson et al.) comes from researchers at Seaside Therapeutics, along with multiple academic collaborators around the world (including some of FRAXA’s favorite past grantees!) The authors present a series of experiments demonstrating preclinical validation of arbaclofen as a potentially disease modifying therapeutic in fragile X. This is significant, because arbaclofen entered clinical trials based on clinical observation of beneficial responses to (regular) baclofen in patients with fragile X, but without much preclinical testing in animal models of fragile X. Using a number of protocols previously established and accepted by fragile X researchers (and virtually all funded by FRAXA), this team showed that arbaclofen can rescue abnormal protein synthesis, AMPA receptor trafficking, audiogenic seizures, and dendritic spine abnormalities in the KO mouse. Thus, arbaclofen has passed a lot of the same tests that mGluR5 antagonists have passed to demonstrate potential for disease modification. While there is no single lab test which can show that a drug can cure fragile X, we think that the ability to correct a broad array of abnormalities, as shown in this study, strongly suggests that a drug strategy may be able to alter the course of fragile X (and therefore be “disease modifying”.) So, this looks really good for arbaclofen. As always, the usual caveats apply: mice aren’t humans, and the most critical aspect of translating this kind of preclinical finding into disease modifying clinical treatment is dosage. The authors of the study were careful to utilize realistic and practical doses of arbaclofen in their experiments, but it is always possible that the doses which can be tolerated by fragile X patients will not be adequate for truly disease-modifying effects (though perhaps still adequate for useful psychotropic effects.) ...................................................................................................... The second paper (Berry-Kravis et al.) presented the data from Seaside Therapeutics’ Phase II trial of arbaclofen in 63 subjects with fragile X; this seems to be the article that is getting most of the attention from the science press. It is essential to understand that Phase II clinical trials are primarily designed to demonstrate the safety and tolerability of the test drug in the target population (i.e. fragile X.) Efficacy is a secondary consideration, but criteria are specified in advance which can provide an initial demonstration of efficacy, and a Phase II trial can be quite useful for defining qualitative aspects of the drug response. In this article, it was clear that arbaclofen was quite safe and well tolerated; indeed, the low rate of side effects compared to placebo might suggest that the dose range chosen for the study was a bit too low. While this trial failed to demonstrate any advantage (statistically, or even numerically) of arbaclofen over placebo in the chosen primary outcome measure (the Irritability subscale of the Aberrant Behavior Checklist, or ABC-I) there was a definite sense from parents and the clinicians conducting the study that those on the active drug did better than those on placebo (“blinded treatment preference”.) In other words, the clinicians and the people closest to the trial subjects could tell who was on the active drug, with a moderate degree of certainty, and those subjects seemed to do better. Unfortunately, this is not the kind of outcome measure that the FDA would allow for approval of a new drug, but it is an indication that arbaclofen is doing something useful. .................................................................................................... As outlined in the article, things got much more interesting when the data were examined after the fact (“post hoc analysis”.) While the Irritability subscale of the larger ABC (the designated primary outcome measure) did not change with arbaclofen treatment, the Lethargy and Social Withdrawal (L/SW) subscale did improve slightly. Moreover, a recent re-factoring of the ABC to make it more “fragile X friendly” showed that the L/SW scale could be replaced with a slightly modified Social Avoidance subscale which was more relevant to fragile X. If the trial results were analyzed in light of this fragile X-specific re-factoring of the ABC, and if only the subjects with elevated Social Avoidance scores were included in the analysis (roughly half of the study population,) then arbaclofen appeared to lower Social Avoidance scores significantly. Additionally, subjects entering the study with elevated Social Avoidance subscale scores showed statistically significant improvement on the Clinical Global Impression Improvement scale (CGI-I), a standard assessment which is considered an acceptable outcome measure by the FDA. Unfortunately, post hoc analysis is not acceptable for FDA approval. But this kind of analysis can inform subsequent studies, and Seaside is now conducting Phase III trials of arbaclofen, using the outcome measures found in this study to be most reliable in demonstrating the therapeutic effects of arbaclofen. So, in the end, perhaps the greatest value of this clinical trial is that it will make future trials of arbaclofen (and perhaps other fragile X treatments) better, and more likely to succeed.

Friday, September 7, 2012

What works, and what doesn’t

At the start, it’s always hard to know what methods will work best for something as complex as the development of disease-modifying treatments for fragile X. But, we’ve always tried to let the science lead us down the right path. At this point, the results are unequivocal, and they have shaped how we are looking for the Next Great Thing in fragile X treatments. As a bit of background, it’s worth noting that there are two basic ways of approaching treatment research for any disease: rational drug discovery vs. high-throughput screening. Rational drug discovery means exploring the basic mechanism of disease and identifying specific “treatment targets” that might be expected to correct the underlying problem. Usually, the target is an enzyme (a protein which facilitates biochemical reactions in the cell) or a receptor (a protein, usually on the cell surface, which detects small amounts of a chemical messenger, such as a neurotransmitter, and reacts in various ways.) Once a potential therapeutic target is identified, small molecules (i.e. drugs) which affect the target in the desired way can be tested in animal models. It’s usually best to look for targets which need to be inhibited for a therapeutic effect, since it’s usually easier to find enzyme inhibitors or receptor antagonists (there are lots of ways to interrupt biological functions with small molecules, but fewer ways to enhance them.) High-throughput screening (HTS) means finding a simple assay, usually based on single cells or even ground up cells, but placing thousands of them in small wells in test plates, then randomly adding thousands (sometimes hundreds of thousands!) of different drugs and looking for a response (via automated equipment), without worrying about the specific mechanism of action. The hope is that many pleasant surprises could emerge from this method, that existing drugs could show unexpected therapeutic effects. The system needs to be set up properly, with some thought given to the “readout” being examined (what specific reaction you’re looking for in those miniature test tubes, like activation of one specific gene), and it is understood that many false positives (as well as false negatives) will be generated, so any “hits” from the system need to be validated later in other model systems. This is one of the standard methods that drug companies have used to look for new treatments, and it is being used increasingly in academic labs as well. A related technique, high-content screening (HCS), uses more sophisticated robotics and computers to look at more complex readouts, like dendrite shape (for an example relevant to fragile X.) Over the years, we’ve funded a lot of both kinds of research. Certainly, basic research into the mechanisms of disease is necessary early on for both approaches, and we started out by focusing on that. At a certain point, however, we were able to sponsor HTS projects, as well as projects looking at preclinical (animal model) efficacy of potential therapeutic compounds identified by the rational approach. Looking back on the results from the past 15 years, it is clear that the rational drug discovery approach has been far more productive, and that HTS has been a big disappointment. It turns out that the fragile X experience has mirrored the general experience with HTS in pharma and academia, where the overall results have simply not justified the investment. In our case, the rational approach has produced many notable successes like lithium (and other GSK3 inhibitors), ampakines, mGluR5 antagonists, minocycline, GABA (A&B) agonists, MEK and ERK inhibitors, BK channel openers, and many others---some you’ve heard a lot about, others which we’ll be talking about much more in the near future. There is now a large pipeline of new disease-modifying treatments in development for fragile X, and all of them have come from the rational drug discovery approach. Many of these therapeutic strategies have important cross-application in the treatment of other disorder (especially autism, but the list of potential applications in other brain disorders is quite long, indeed.) Why no success with HTS? Well, it is becoming apparent (at least in retrospect) that the biggest problem results from the nature of fragile X itself. For example, when scientists have tried to find a simple cell-based readout, they find that fragile X cells (stem cells, in this case) display increase proliferation. Run that through a high-throughput screen, and you get lots of “hits” which rescue this particular phenotype. The problem is that they’re all cancer drugs which are far too toxic to use in fragile X. As it turns out, the same signaling pathways which are over-active in fragile X neurons are really just re-configured versions of the signaling pathways which regulate cell growth and division in other cells. Part of the differentiation process, which makes one cell a liver cell and another, a neuron, is just a minor tweaking of these same pathways to accomplish radically different things in various parts of the body. All cells have similar building blocks, and drugs can’t always distinguish between them. So, it is very important to choose the readout of the HTS system very carefully, and with some consideration of this basic nature of fragile X pathology. We are hopeful that high-content screening may allow us to find novel agents which can rescue very specific defects found in fragile X (like abnormally developed dendrites,) but this is still an emerging technology. Again, we must go where the science leads us, and learn from these valuable experiences. While fragile X is turning out to be a different sort of problem from what we might have originally expected, we are learning ways to fix the abnormalities caused by this disorder, and better ways to look for new treatments. In the end, fragile X research and rational drug discovery will ultimately teach us a great deal about how the brain works, and about what’s wrong in a host of other disorders like autism, Alzheimer’s Disease, and schizophrenia. Even though we’ve always known this fragile X research was important work, because we care so deeply about our children’s futures, it may turn out to be even more important than we could have imagined. (double-posted here and at the official FRAXA Blog---check it out at )