For once, this isn’t hyperbole. Metastatic pancreatic cancer hasn’t had a meaningful new treatment in literally decades - it’s been brutal chemotherapy or slightly less brutal chemotherapy.
The new generation of KRAS inhibitors, of which daraxonrasib is just the first to make it to approval, is genuinely a practice-changing breakthrough.
It won’t cure people with metastatic disease, but it will help a lot live better lives for longer.
>Metastatic pancreatic cancer hasn’t had a meaningful new treatment in literally decades
Not true. The folfirinox chemo regimen was invented a couple of years ago, and it significantly increased the life expectancy and quality of life for terminal pancreatic cancer patients.[0]
> It won’t cure people with metastatic disease, but it will help a lot live better lives for longer.
Cancer is a problem that can be solved, and it needs to be solved asap. Cancer too often gets either relegated to "Actually, statistically it's very unlikely" by people who don't have it or stylized to the God of Death that unlike other illness we won't solve anytime soon. There is no time to be lost tackling with AI. Dario gets it. Had cancer himself and his father passed away from it.
The thing that makes cancer such a hard problem is that cancer is really a collection of diseases. There's, what, like over a hundred of them? And one organ can produce several unrelated cancers too. I do hope AI will actually be able to figure out the commonalities between all of them and a potential path for a possible even somewhat-universal solution, but I at least don't even know if a universal "cure" is even possible. Granted, my knowledge of biology with respect to cancers is pretty limited.
This is exactly the mindset I meant. Look up how people dismissed Neuralink 5 years ago and now it's precise enough to play GTA via. A month ago Tao prognosed that the Riemann hypothesis will be solved in the next 25 years.
I have a close relative currently in the last stage of a different type of cancer, would be pretty happy to extend their life by another six months. Or even better, improve quality of life because what I'm seeing is no life at all. This might or might not be a breakthrough by some strict definitions of the term, but sure is for those for whom it is important.
Quality of life on chemotherapy is not great. Sadly, I know this firsthand through my father. I hope it gets approved in my country in time to reach him, even if it only gives him another year and a half.
Often pancreatic cancer is discovered only after it has spread. If this can treat it at that point it turns the disease from a death sentence to a lifeline until more research can come to fruition
Everything is relative. A few more months of life with less pain and more energy to spend it with family surely is a breakthrough for many patients and for the doctors treating them.
You're awfully negative about what a lot of people here see as a breakthrough. Can you be more descriptive about why you don't agree this constitutes a substantial difference other than saying you disagree and dismissing others' opinion?
I don’t think you have ever had to witness somebody undergoing chemotherapy. An alternative, if available is almost always better, and certainly less painful.
> Treatment-related adverse events that led to treatment discontinuation occurred in 1.2% of the patients in the daraxonrasib group and in 11.2% of those in the chemotherapy group.
That's a huge reduction in side effects bad enough to stop treatment.
AI will, but the most important thing we need is directors to guide it, same as GRH
Right now AI can check images and give you accurate diagnosis. But that is something already there. We need to prevent, better than cure. And to prevent we need more info about the causes, we need more strict rules about environmental contamination, same as we did with smoking. We need to understand the influence of drugs, chemicals, food, stress, in the development of cancer. We need to put all knowledge in one place in order to allow specialists and geniuses to find solutions. Creating walled silos of knowledge is in benefit of none but a few, that's why open weights, open models, and open knowledge is extremely important for us all
I think the future will be better for those who prevent it, not just cure it. Accurate prevention will make us immune to cancer
The new generation of KRAS inhibitors, of which daraxonrasib is just the first to make it to approval, is genuinely a practice-changing breakthrough.
It won’t cure people with metastatic disease, but it will help a lot live better lives for longer.
Not true. The folfirinox chemo regimen was invented a couple of years ago, and it significantly increased the life expectancy and quality of life for terminal pancreatic cancer patients.[0]
[0]https://www.nejm.org/doi/10.1056/NEJMoa1011923
Cancer is a problem that can be solved, and it needs to be solved asap. Cancer too often gets either relegated to "Actually, statistically it's very unlikely" by people who don't have it or stylized to the God of Death that unlike other illness we won't solve anytime soon. There is no time to be lost tackling with AI. Dario gets it. Had cancer himself and his father passed away from it.
there doesn’t even need to be a commonality, all will get solved
I hear ML researchers are even picking up smoking again
Unless we can reliably sign our cells like we do our binaries, we ain't "solving" cancer.
It's always impossible until it's done
Am I missing something?
From other sources:
> Median overall survival: 13.2 months versus about 6.6 to 6.7 months with chemotherapy.
Merely counting months does not do much - you'd have to count the quality of these months too.
I don't see that as a breakthrough.
> Treatment-related adverse events that led to treatment discontinuation occurred in 1.2% of the patients in the daraxonrasib group and in 11.2% of those in the chemotherapy group.
That's a huge reduction in side effects bad enough to stop treatment.
https://www.nejm.org/doi/full/10.1056/NEJMoa2605555?
If we can prove GRH with AI we really should be able to cure cancer.
Right now AI can check images and give you accurate diagnosis. But that is something already there. We need to prevent, better than cure. And to prevent we need more info about the causes, we need more strict rules about environmental contamination, same as we did with smoking. We need to understand the influence of drugs, chemicals, food, stress, in the development of cancer. We need to put all knowledge in one place in order to allow specialists and geniuses to find solutions. Creating walled silos of knowledge is in benefit of none but a few, that's why open weights, open models, and open knowledge is extremely important for us all
I think the future will be better for those who prevent it, not just cure it. Accurate prevention will make us immune to cancer
Pancreatic cancer halted by virus injection in three patients - https://news.ycombinator.com/item?id=50563066 - Nov 2026 (157 comments)