Quote:
Originally Posted by Titty Meat
How many people in a study have to show success in order for the drug to move forward? I've read Favilavir has shown to be successful with about 70 in a clinical study in China. Also read that APN01 was developed in the early 2000s during the SARS outbreak and helped prevent damage to lungs. Could a cocktail of these drugs work the same way HIV patients take a number of medications in combination to combat the virus?
|
The short answer is that it's complicated. The long answer will follow:
1) The evidence is only as good as the trial design. In a situation like this, you aren't going to have good trial design, because people deteriorate quickly, there are no known beneficial treatments, and the need is so great that you aren't going to be able to blind patients and/or investigators as there is no standard of care to compare against. Ideally, you would want a large, multicenter trial with thousands of patients from a variety of demographics. They would need a random distribution between the intervention and the control group that is also consistent within subgroups. The trial should also be forward-looking (prospective) rather than collecting data from old cases and compiling it together (retrospective). The more patients you have, the greater your predictive power of the trial can be. Blinding helps prevent the placebo effect and reporter bias, but that really can't be done at this stage either.
2) In order to pass muster, most operators have agreed that a p-value of greater than two standard deviations is necessary to reduce the chances of it being a statistical anomaly. What that means is that there has to be less than a five percent chance the effects are random for it to be statistically significant. However, this is also complicated--what is the difference between clinical and statistical significance? Let's say I have a blood pressure drug I'm bringing to market and it lowers BP by an average of 2mmHg compared to a popular ACE-inhibitor, like lisinopril. Is that statistically significant? Yes. Is it clinically significant? Not likely. So the other thing I have to do is consider how much of a clinical impact this drug may have. That's why Tamiflu is overused in my opinion. Unless you are very young or old/immunocompromised, Tamiflu only shortens the duration of flu symptoms by about 8 hours, and that's only if you begin treatment within the first 40 hours of symptom onset.
3) Trials normally progress through three distinct phases (ignoring the preclinical trials with animal and computer models). The first is a dose escalation study--what can the body handle. Usually this is done in healthy volunteers depending upon the drug, but in more emergent cases, it is applied to individuals with the illness. The second phase is generally a placebo comparison--is this better than nothing at all? The third phase is comparison against an active comparator. Is this better than the current standard of care? As you can see, a trial for a COVID-19 agent isn't going to be able to meet these standards very well, either.
So, let's apply this to COVID-19:
What am I looking for first: a mortality benefit. What am I looking for next: morbidity benefits: decreased hospitalization length, reduced clinical sequelae, etc.
So, what are these studies like: the favipravir study you mention is just a dose escalation study of about 60 people. You can get enough data with a sample size that small to meet statistical significance, but the confidence intervals will be broad and uncertainty high due to the small sample size.
Also, what are the impacts of polypharmacy? Generally, multiple agents are needed in the case of antivirals, whether it's Hep C or HIV. However, one must be careful in choosing those agents due to drug interactions.
Protease inhibitors are a great example. Their use revolutionized treatment for HIV, but protease inhibitors are among the strongest inhibitors of an enzyme in your body called cytochrome p450 3A4, which metabolizes more drugs than any other drug type. Well, what does this mean? Let's say that I'm taking carbamazepine for seizures and I start a protease inhibitor. What happens? The inhibition of 3A4 causes carbamazepine levels in my system to build up to toxic levels because I'm not clearing the drug quickly enough.
So, what happens if one agent is a substrate and the other is an enzyme inducer? The second drug wipes out the efficacy of the first drug. In other cases, the inhibition (carbamazepine example) can actually be a benefit. Ritonavir is a potent 3A4 inhibitor, and is used to actually boost the level of other protease inhibitors (as they are also 3A4 substrates). Thus, when weighing such therapy, I must consider those effects as well.
Also, I have to think about the side effects of each drug. For years, patients coming into the hospital who may have had an infection were given broad spectrum antibiotic coverage often with vancomycin and piperacillin-tazobactam (Zosyn). This would cover gram-positive bugs, gram-negative bugs and anaerobes. However, both agents are harmful to the kidneys, and the rate of acute kidney injury from their concurrent use is high, especially in an elderly population. This is also why you should avoid the use of fluoroquinolone antibiotics in the elderly. They increase the QT interval in the heart (old people are already prone to arrhythmias) and increase the risk of tendon rupture. So, who am I giving my new therapy to? What are its side effects? How do those side effects affect that person? What are their comorbidities?
However, there's a reason why HAART works. In the case of HIV, which is prone to rapid mutation, suppression of one target (reverse transcriptase was the first and only target for almost a decade) leads to rapid development of resistance. Suppression of multiple targets (reverse transcriptase *2 plus viral protease) lead to a revolution in therapy, although not an elimination of resistance. We are still learning about SarsCoV2. Sites of attachment, entry, etc. More time will provide greater clarity on the viral life cycle, which will better elucidate treatments. Time, unfortunately, is not on our side in this case.
Key takeaways:
There is no magic number of patients, but more (hundreds to thousands) is always better and far, far more predictive than dozens.
Polypharmacy may work if the mechanisms of action are synergistic or additive, but the risk of drug interactions and drug-disease interactions also increases.