Hacker News Reader

Comment by msie | Hacker News Reader

They should give it to some people with fatal stages of cancer.

I agree, or at least I would stress that people should be allowed to consent to that. I don't know what the prevailing medical ethics of doing that kind of thing in consenting patients in that state, but my uninformed intuition is I would disagree with it.

Though one thing that I might think researchers might not want is people may be too sick to recover even if their cancer disappeared tomorrow.

In the US, the FDA has a Compassionate Use exemption to clinical trials for exactly this circumstance!

There must be informed consent, no reasonable alternatives (which, in cases we deem terminal, is often the case), and some evidence pointing to the treatment possibly being helpful. It's an excellent ethical program that gives patients a choice and advances science.

In my experience most legitimate biotech companies working on promising drugs and therapies don’t want to touch the exemption with a 30 foot pole. Since they raise most of their money from the public to fund clinical trials, a single bad reaction could generate enough bad PR to derail fundraising and kill the drug. Sticking to clinical trials allows them to control that blast radius so even though the FDA approves >95% of applications, in practice very few drugs are available that way.

The biggest exception is oncology. Since everyone knows that chemotherapy is hell, cancer drugs tend to get a pass and pre-approval companies are (slightly) more willing to work with compassionate use exemptions.

Both patient participation in clinical trials and compassionate use of experimental treatments are fairly common for cancer patients, with various accessibility barriers. (One issue with the latter, for example, is that the incentives aren't lined up for companies to provide unapproved drugs to dying patients, you're way more likely to get a horrible complication that leads to bad press than a miraculous recovery).

Here's an insightful blog series about Jake Seliger's experience participating in clinical trials. He was a regular HackerNews user who passed away in 2024: https://bessstillman.substack.com/p/please-be-dying-but-not-...

What is the success rate of a clinical trial? Just to see things in perspective.

It's around 10-15% for the whole drug I-III flow (13.8% according to [1]), but that varies dramatically based on therapeutic area. On the order of a third of infectious disease vaccines might be approved but only maybe 5% of oncology therapies because the latter often have a different standard for approval so it's cheaper to run trials.

[1] https://pmc.ncbi.nlm.nih.gov/articles/PMC6409418/

That's interesting, but I was talking about the success rate of someone with a terminal illness going the clinical trial route. Sorry, I now see that my question was not so precise.

For cancer, it doesn't seem to impact survival odds at all [1]. In other fields it may improve metrics a small bit but that's largely because in clinical trial patient selection, they're very careful to exclude anyone with an even remotely confounding factor (like weight/BMI).

[1] https://www.science.org/content/article/joining-cancer-trial...

This is why people begging to take untested, unknown drugs in the extreme off-chance of they work is generally a bad approach. It almost never works, and it encourages the earlier release of ineffective drugs to a wider audience.

If someone is about to die and you save them at the last moment, aren't you basically reviving a zombie at that point? He has eight new tumors. You can pop almost all of them and still be left with a terminal patient.

Even if you're buying time with every trial, all you've done is turn the patient into a lab rat for physicians to play around with. The ideal patient needs to be dead enough to have no human rights, but alive enough to participate in the trial. The hope of a miracle cure means the patient doesn't believe himself to be dead enough to not have human rights anymore. It's a paradox.

Signing the documents for such a trial is equivalent to signing your consent for euthanasia. It shifts the blame of death from the cancer to the company performing the trial. It's an extended form of organ donations where you donate your entire body while you're still alive.

Jake Seliger died ten weeks before his daughter was born. For many people, a chance at buying some time is worth a lot.

Trials are also by no means exclusively performed on terminal patients.

Both of my parents have benefited from access to early medical trials. One is currently very late stage IV cancer. Access to trials is usually proxied through respected doctors/oncologists affiliated with major hospitals rather than offered broadly. I assume for reasons of experimental protocol and integrity the overseeing doctors are typically not the same as the conceiving research team.

That is exactly how clinical research works. My mother worked running clinical trials for two decades.

When she was diagnosed with leukemia she was able to get into a research study herself that gave us 10 more years together.

One of the horrible but necessary parts of trials is the control group, who receives placebo. This is only done in a few of the trial phases but is essential in measuring efficacy. If someone wants to throw their brainpower and a little bit of AI/tech at the problem, you could end up eliminating a lot of suffering.

AI and tech won't help, but if the threshold to try a drug were adjusted to exactly the right threshold, where enrolling in a study would be expected-value neutral (this is by marginal reasoning), taking a placebo would not be worse than not.

I'd think AI and tech could solve the problem pretty easily, assuming the study authors could get access to the health records of everyone that was not in the study (and could therefore generate control cohorts by looking at large samples of comparable patients outside the experimental group).

This is done for targeted advertising all the time. Frustratingly, the surveillance capitalism industry is precisely the reason the dataset you'd need probably shouldn't exist.

Maybe we'll get some decent lawmakers sometime soon, and problems like that will be fixed via legislation. They'd need to ban the root-cause of the problem. I'm guessing it's more likely the current congress will let private companies steal + sell everyone's info instead.

Well yeah, that's the plan. Every medicine needs testing in humans before going widespread. That's... how it always worked.

Reply on news.ycombinator.com