Why a placebo-controlled trial? I don't care how a new medication for schizophrenia compares vs placebo. I care how it compares vs standard of care - antidopaminergics. Placebo controlled trials are fair for conditions that don't yet have a standard of care. I want to see a trial where efficacy and tolerability are compared vs. a popular antidopaminergic (vs. combo?).
We have a word for trials in other areas of medicine that are placebo controlled instead of compared vs. SOC - unethical. Consider cancer. It would be egregious to compare a new cancer therapy vs placebo if the SOC is some other treatment. When we accept trials like this in schizophrenia, we affirm that schizophrenia isn’t as bad and we’re hungry for sub-par evidence.
I’m not a researcher but there are ways to make good trials to surmount with the practical concerns you raise.
We have drugs to treat schizophrenia. I don’t think we should waste money, time, and the quality of research participants’ lives to ask if new drugs are merely effective on their own. I want new drugs that are better (either more effective or better tolerated).
“Better tolerated” seems extremely clear from this trial, does it not?
Of course you do have a case that ex ante there should have been another arm. But ex post, it’s pretty clear none is needed because the typical antipsychotic side effects are so striking and didn’t show up at all.
We can infer that, but we can't say that definitively without a head-to-head trial. It's not easy to compare placebo-controlled trial A with placebo-controlled trial B.
Let's say they ran another trial and compared this drug vs a popular antidopaminergic. Such a trial might show that this drug is better tolerated (fewer adverse effects, fewer discontinuations), but doesn't work as well (or at all) vs. SOC. Depending on the effect size, that might mean this new drug has no place in our current array of therapies for schizophrenia. We'd need such a trial to draw that conclusion though.
I don't think you need to do sophisticated statistics to compare multiple trials here. The trial as-is shows a very obvious lack of the big antipsychotic side effects. Too big to demand any statistical sophistication.
That, plus having a decent effect vs. placebo, plus being in itself reasonably well-tolerated, would seem to be enough to qualify this as a useful schizophrenia treatment.
Of course, as you say, it would have been a better choice to have run a better trial and get more specific information. But one can reasonably draw conclusions from the trial as-is.
1. From a cynical point of view, you don't want the study to show inferiority to the comparator and have the debut headline be "FDA Approves Cobenfy, But Phase 3 Trial Shows Inferiority to Olanzapine."
2. From a practical point of view, it seems like it would make study design and enrollment a lot harder. What drug are you going to pick and why? Are you going to allow in people who have previously responded to that drug, or not? It seems much cleaner to use placebo to show preliminary efficacy
3. You might end up in a situation where your results are hard/impossible to interpret if there's no placebo. Suppose that your study turns up a null result for both xanomeline and your comparator. Obviously there's something very wrong with the population you selected for your study, but it doesn't give you any clear information about whether or not xanomeline is effective or not. Now you have to spend $20-30 million dollars running yet another Phase 3 trial and waiting another 2-4 years to collect the results, etc. etc.
That said, I think this study would be improved with three arms, placebo, active comparator, and xanomeline. I suspect that this isn't done for cynical reasons.
Good stuff. I haven't followed the KarXT stuff much so this is a good update. When I got to the end I was thinking what you were thinking - what about combination treatment? I'd be interested to see what xanomeline/trospium + really low dose haloperidol (or other high-potency 1st gen) does, as well as xanomeline/trospium + various SGAs.
Why a placebo-controlled trial? I don't care how a new medication for schizophrenia compares vs placebo. I care how it compares vs standard of care - antidopaminergics. Placebo controlled trials are fair for conditions that don't yet have a standard of care. I want to see a trial where efficacy and tolerability are compared vs. a popular antidopaminergic (vs. combo?).
The cynical view is the correct one here.
We have a word for trials in other areas of medicine that are placebo controlled instead of compared vs. SOC - unethical. Consider cancer. It would be egregious to compare a new cancer therapy vs placebo if the SOC is some other treatment. When we accept trials like this in schizophrenia, we affirm that schizophrenia isn’t as bad and we’re hungry for sub-par evidence.
I’m not a researcher but there are ways to make good trials to surmount with the practical concerns you raise.
We have drugs to treat schizophrenia. I don’t think we should waste money, time, and the quality of research participants’ lives to ask if new drugs are merely effective on their own. I want new drugs that are better (either more effective or better tolerated).
Would you change the way you treat patients based on the study design and results? This study design is so flawed the answer would have to be "no".
What makes you say that this study design is so flawed?
“Better tolerated” seems extremely clear from this trial, does it not?
Of course you do have a case that ex ante there should have been another arm. But ex post, it’s pretty clear none is needed because the typical antipsychotic side effects are so striking and didn’t show up at all.
We can infer that, but we can't say that definitively without a head-to-head trial. It's not easy to compare placebo-controlled trial A with placebo-controlled trial B.
Let's say they ran another trial and compared this drug vs a popular antidopaminergic. Such a trial might show that this drug is better tolerated (fewer adverse effects, fewer discontinuations), but doesn't work as well (or at all) vs. SOC. Depending on the effect size, that might mean this new drug has no place in our current array of therapies for schizophrenia. We'd need such a trial to draw that conclusion though.
I don't think you need to do sophisticated statistics to compare multiple trials here. The trial as-is shows a very obvious lack of the big antipsychotic side effects. Too big to demand any statistical sophistication.
That, plus having a decent effect vs. placebo, plus being in itself reasonably well-tolerated, would seem to be enough to qualify this as a useful schizophrenia treatment.
Of course, as you say, it would have been a better choice to have run a better trial and get more specific information. But one can reasonably draw conclusions from the trial as-is.
Lots of answers that come to mind
1. From a cynical point of view, you don't want the study to show inferiority to the comparator and have the debut headline be "FDA Approves Cobenfy, But Phase 3 Trial Shows Inferiority to Olanzapine."
2. From a practical point of view, it seems like it would make study design and enrollment a lot harder. What drug are you going to pick and why? Are you going to allow in people who have previously responded to that drug, or not? It seems much cleaner to use placebo to show preliminary efficacy
3. You might end up in a situation where your results are hard/impossible to interpret if there's no placebo. Suppose that your study turns up a null result for both xanomeline and your comparator. Obviously there's something very wrong with the population you selected for your study, but it doesn't give you any clear information about whether or not xanomeline is effective or not. Now you have to spend $20-30 million dollars running yet another Phase 3 trial and waiting another 2-4 years to collect the results, etc. etc.
That said, I think this study would be improved with three arms, placebo, active comparator, and xanomeline. I suspect that this isn't done for cynical reasons.
Good stuff. I haven't followed the KarXT stuff much so this is a good update. When I got to the end I was thinking what you were thinking - what about combination treatment? I'd be interested to see what xanomeline/trospium + really low dose haloperidol (or other high-potency 1st gen) does, as well as xanomeline/trospium + various SGAs.