It's not their risk to take. Getting a drug to market is a complex process that can be derailed by any number of mistakes ranging from the mundane like improper blood draw during clinical trials to the tragic like accidental deaths that puts public pressure on the FDA to shut it down or spook away investors. Any hiccup in this process can kill it and once a patent expires, the chances of someone bringing that drug to market is almost zero, regardless of its efficacy.

The vast majority of drugs and therapies are developed via private investment and many companies go public before they make even a cent of revenue to fund the clinical trials. Public sentiment is the driving force behind their stock prices because they're literally forbidden from making money until the FDA approves their drug. Lots of them go bankrupt every year before finishing clinical trials or releasing a product and rushing the process is quick ticket there.

Patients don't get to put at risk the decades of good a drug can do because they're desperate and afraid of dying.

Michael Jackson, Joan Rivers and Lisa-Marie Presley for example would probably all still be alive if they didn't have the money to simply buy the healthcare they wanted, damn the consequences and cost.

You offer someone death, they get death.

This is about people desperate for cures and treatments that probably won't work, and may well give them the opposite outcome from what they hope for.

There is a difference in kind.

This applies just as much to approved treatments as unapproved, for the same reasons, and has the exact same solutions such as having a third party with no vested interests explain to the patient. In fact it has those problems less than approved treatments because the companies are being paid by the insurance company for the approved treatments but not the experimental ones.

What constitutes the acceptable tradeoff is the heart of the matter. On the one hand, even the current approval process does not always guarantee efficacy (Aduhelm to mention it by name). Medical trials work with considerable uncertainty, and the risk/benefit tradeoff is always a question of probabilities. Even with the current process, there is no certainty or guarantee.

It is possible to take those probabilities and to talk about the expected number needed to treat (NNT) for a medication, the expected side effect profile, how many people might live and how many might die. Things that may sound cold, but that are already routinely done today.

It is hard to know where the tradeoff should be, how much margin of error is the right amount before something is approved, knowing that there is uncertainty even in large trials, but it's not a choice between the totally unapproved and the status quo. (Better yet, we should not model the choice as a single parameter that goes from more choice & more snake oil to less choice & more safety. It's also often possible to improve one aspect of a bureaucracy without it being a zero-sum game.)

> It is possible to take those probabilities and to talk about the expected number needed to treat (NNT) for a medication, the expected side effect profile, how many people might live and how many might die.

NNT > NNH = efficacy. To calculate these the trials need to be done, which is not the case for experimental therapies as they have undetermined benefits and harms (hence why they’re in trial).

The choice here is between conventional options that have known benefits/harms, however crappy they may be, with something that has unknown values.

I think that situations like TFA make a compelling case for selectively broadening access, for some drugs and some pathologies. The characterization of a choice between something proven and something with unknown values is too coarse and not a good model.

Clinical trials are very expensive, drug development as a field tends to move forward with what it thinks will have a chance of making it through. While most new drugs indeed fail (the large majority of them), in diseases with very poor prognosis like OP's cancer, I think we would benefit from still looking at the tradeoff as a decision under uncertainty, instead of a complete unknown. Even with less data than we would otherwise want.

Looking at experimental therapies as completely undetermined leads to a binary choice. There is supporting evidence that allows those therapies to move forward with experimental trials in the first place. The counterfactual of not treating patients with poor prognosis being what it is, my overall point is that treating experimental drugs as total unknowns with benefits and harms that cannot be quantified until the full process is finished is not consistent with the knowledge that, even after all trials are completed, this is still a uncertain statistical result, only with slightly more data than before.

In other words, we are always working with partial unknowns so when the prognosis is bad a larger unknown may be good overall and save lives, even though it wouldn't be acceptable otherwise.

Which is literally what just happened with osimertinib for NSCLC. It was approved for clinical use and was implemented without any overall survival data.

Safe for clinical trial means the harms are considered safe, has little to no relevance for efficacy.

Once again you’re ignoring the fact that there are (almost) always alternative treatments that have been proven to prolong life.

No one is offering nothing to poor prognosis patients.

From the blog post:

> Monday I’m starting chemotherapy, but that’s almost certainly going to fail, because a CT scan shows four to six new gross tumors, four in my neck and two, possibly, in my lungs.

In someone with such an aggressive disease everything is almost certainly going to fail, experimental or not. Chemotherapy will buy some time. Who knows what an mRNA treatment will do but there’s certainly no (not little, no) evidence that it will be curative.

I think sticking with standard of care until new drugs have gone through the full, current process may be net negative under the current process, and considering the counterfactual especially for patients with poor prognosis.

Chemoterapy will buy some time with certainty, but in expectations broadening access even slightly means we can gain confidence in the experimental treatments that do work slightly sooner, and that improves the standard of care for everyone. Experimental drugs sometimes do work, and it would be bias to only look at the risks while sweeping aside the potential upside.

Consider what happens a few years from now with the current system and with a system that selectively broadens access in some cases, when the early data seems to support the risk/reward. Even if most new medications fail, the cost of delaying those few therapy that will happen to work is much larger than the risk being paid by the trial population. That's because only a few people participate in trials, but everyone benefits from the results. This is why we have trials at all.

In someone with an aggressive disease such as OP, that person may very likely benefit from having that choice. Today they may choose to buy some time with _relative_ certainty, knowing that chemo has very heavy side effects and that they may not enjoy very much of that remaining time. But they may not choose to try some experimental treatments, even when the risk/reward seems reasonable according to early data, even though that may result in a small chance of success, and certainly data that will help bring therapies quicker to more people.

The benefit of experimental treatment is not just to the individual participating in the experiment. Part of the benefit, that helps justifies the risk, is the idea that a fraction of the treatments will be worthwhile. Slowing down trials is safer in the sense that no one will be responsible for anyone losing their chance at chemo or dying, but it creates the exact kind of invisible deaths in expectation that TFA talks about.

Furthermore, Step 1 of the NCCN treatment guidelines for these kinds of cancers (i.e. advanced and/or poor responders to conventional chemo) is always that the patient is enrolled in a clinical trial.

In this authors case he does not meet inclusion criteria for the therapies he is listing, both trials require systemic therapy with novel PD-1 inhibitors (aka immune checkpoint inhibitors, immunotherapy) which when they work do so fantastically with little side effects.

Per his description he has not tried any systemic therapy yet.

I’m unclear whether he’s lumping chemo with immunotherapy (often the case by non-domain expert patients) but not even Moderna is interested in going from surgery directly to mRNA.

This isn’t the FDA getting in the way. He hasn’t exhausted good treatment options yet to be eligible for unapproved experimental ones. It would absolutely be detrimental to ignore ICIs in an eligible patient in favor of [insert non-approved experimental therapy with no evidence].

I’ve seen far more advanced metastatic H&N SCC survive for years on ICIs in my clinical practice. We don’t know all the specifics of this case but this isn’t nothing/days/weeks vs mRNA as the author is pitching.

I'll take your point, and I appreciate the time you took to write a good response, though I've read about enough examples other than the current article that I feel this may be rejecting a wider pattern on the details of the current submission (but maybe I'm wrong, and they all happened to be incorrect for similar reasons).

But I'd like to hear your thoughts on the broader argument of being more aggressive with trials in general. What sticks with me each time this argument comes up, regardless of the particular situation, is this idea that despite some mechanisms for accelerated approvals, we are still far slower than we could be, if we take into account the number of people saved by bringing therapies sooner.

There is a cost to letting people take dangerous medication. I hate the idea of peddling dubious cures to vulnerable people as much as I'm sure you do. It's the counterfactual that makes the argument, the people who do not receive treatment due to delay seems to vastly outnumber the people harmed by experimental treatments. And the difference seems large enough that it would remain true even if we went much faster. I think that's the strongest objection to the current system.

> But I'd like to hear your thoughts on the broader argument of being more aggressive with trials in general. What sticks with me each time this argument comes up, regardless of the particular situation, is this idea that despite some mechanisms for accelerated approvals, we are still far slower than we could be, if we take into account the number of people saved by bringing therapies sooner.

Everything in medicine is trade-off. On the extremes we can practice with no regulation and have immediate access to treatments without evidence accepting that many patients will be harmed as many new treatments are inferior or require phase IIIs on the other end and minimize harms but accept that delayed care will itself cause harm.

Current practice is somewhere in the middle, depending on various factors like efficacy, risks/harms, and alternatives many treatments enter practice after phase I/II or used off-label for close indications.

I like where we're currently at for oncology and don't think we need to change much in the process. The bigger hurdle is funneling those R&D $ into high yield research and perhaps making the cost of trials easier (e.g. research alliances, IT infrastructure to simplify multi-center studies and patient recruitment, ?active government involvement).

There's always fine tuning of that balance that can be done and some cases will prove things wrong on either side of the argument but generally speaking I think we're close to where we need to be. Definitely better than when we didn't have the current approval process.

> It's the counterfactual that makes the argument, the people who do not receive treatment due to delay seems to vastly outnumber the people harmed by experimental treatments.

I don't think this is true and I have not seen evidence to support this claim. There is ample evidence of failed treatment options or treatment causing more harm than good. An example I'm well-versed in is HIPEC for advanced ovarian/peritoneal cancers (we've been doing this for years and it turns out it doesn't do much other than excessive morbidity) or Y-90 for HCC/colorectal metastases to the liver (first trials negative and we recently found out "oops we've been underdosing this and measuring non-target dosing to the lungs incorrectly") both of which are for palliative patients with no good alternatives.

Going back to osimertinib as the example, the main criticism of earlier TKIs (same class of drug, and even osimertinib for that matter) was that the promise of limited data (progression free survival rather than overall) led some patients to get TKI therapy rather than platinum based chemo which was standard of care and has proven OS benefit. Except when the OS data came out it was an "oops, OS didn't pan out like PFS" and people were harmed.

That these happen in the current system strongly suggest patients have access to experimental therapies and that we're not being overly stringent, or this data wouldn't exist.

I would also say that a large proportion of the patients I've interacted with that have disseminated disease are on some trial for experimental therapy with the caveat that I've only worked at tertiary care cancer centers. It's really not that hard to get enrolled if you meet criteria. I believe the stories we read of access challenges are largely a vocal minority with challenging circumstances or a misunderstanding of treatment options.