Today I saw a video clip of Larry Ellison discussing the promise of artificial intelligence and mRNA vaccines. He thinks it's going to cure cancer. That's not going to happen. Here's why.
I'm confident AI will have numerous uses in healthcare. I've described many of those at length in prior videos and podcast. Most recently on plenary session podcast you can find a 1-hour lecture I gave on artificial intelligence to the California Hospital Association with many ideas, including making the EMR more usable, and responding to patient queries.
mRNA vaccines I'm less excited about. They clearly have unique and idiosyncratic toxicity. Because they were pushed so hard for covid-19, there's a huge fraction of the public who does not want them. They do have unexplored long-term safety questions. I'm not going to be standing in line to get any.
Cancer therapeutic vaccine refers to a category of vaccine, and it doesn't have to be mRNA, where researchers take a piece of a patient's tumor, and combine it with some immunologic stimulation, and inject the patient with it. The idea is that you create an immune response within the patient against their own tumor. This is different than the HPV or hep b vaccine which prevents you from getting a virus that may cause a cancer.
Over the last 50 years in oncology, there has been tremendous enthusiasm for cancer therapeutic vaccines. In fact, hundreds have been studied. Only one has been FDA approved. The rest failed.
What was the one success? It was a drug called provenge that improves survival by 4 months. But here's the problem with this study. If you are on the control arm you got a saline injection instead of a vaccine. But then if your cancer got worse, you got the vaccine and not the standard of care therapy. So it's a trial of an upfront vaccine in one arm, and a huge delay to treatment in the other arm. The most likely possibility is the control arm was harmed by delaying a useful salvage therapy And that's why there's a survival benefit. Notably not a single person's tumor shrunk.
So what I'm telling you is cancer therapeutic vaccines do not work. Why would MRNA make it better?
MRNA might help you target a specific antigen more easily, but we've already learned that targeting antigen is not the issue with these products. Probably the reason why they don't work is that people with these cancers have an immune system that's just not capable of being revved up in this manner.
And how precisely is artificial intelligence going to diagnose cancer early? There are lots of blood-based tests to find cancer early, but that doesn't solve the problem. Because you can't just have a positive blood test. You need to scan the person afterwards and find the spot to cut out. And if you can't find the spot then that blood test is really scary and you can't do shit about it.
To date, not a single company has shown that they can identify all cancer, and cut it all out. But now here's the trickiest part of all. Not every lesion should be cut out. Some things look like cancer under the microscope but they're never going to kill you. You can cut all of those out and you'll just have a bunch of holes but you won't live any longer. And some cancers can be cut out, but they've already spread. You can cut all of those out, but you're going to die anyway. Not a single one of these companies has solved this problem. The problem is finding the right type of cancer to cut.
Worse, artificial intelligence cannot be trained to solve the problem because no one has a properly coded data set for artificial intelligence to train off of. It's a huge conundrum.
One last point. Some people point to a Moderna randomized trial of a cancer therapeutic vaccine as if it works. That's study is an underpowered phase 2 study with massive censoring. It is so unreliable.
I'm not confident AI will find the right type of cancer, and I'm even less confident that mRNA vaccines are going to do anything. Spending $500 billion on AI might be a good investment. But it probably is not going to cure cancer.
I want to end with a final message.
I'm interested in the truth. On every single issue I comment on, I'm trying to give you my most accurate impression. That's what I'm interested in providing. I'm not always going to support your preconceived notions, and many of my views will span both political parties. If you like that, become a subscriber.
Your truth is what I’m here for. I don’t expect perfect prescience or judgment.
You’re the person I wanted to hear from as soon as I saw the press conference. Hard to listen to Ellison talk about future scientific oncology discoveries when he’s neither an oncologist nor a researcher.