> Currently, the most plausible theory emerging from her team’s research points to metabolism: Healthy and cancerous cells may process reactive oxygen species—unstable oxygen-containing molecules generated during radiation—in very different ways.
There was also a study that showed that chemotherapy efficacy was enhanced by fasting before treatment.
It seems that when calories are scarce, healthy cells turtle up while cancer cells keep consuming, so fasting reduces absorption rates in healthy tissues and thus collateral damage.
Healthy cells CAN turtle-up, whereas cancer cells engage in unregulated reproduction. Also, some cancer cells can only consume glucose. Which, in a fasted state, would mean that the majority of energy would be in ketones(if the individual were metabolically healthy), starving the cancer cells to death.
Because the cancers cells adapt! (fast reproduction and high mutation rate of the cancerous cells make that process quicker than antibiotics resistance)
Another interesting part of the story is the user element. The issue was most often triggered by fast, experienced technicians who were able to key commands more quickly than Therac engineers anticipated:
> After strenuous work, the physicist and operator were able to reproduce the error 54 message. They determined that speed in editing the data entry was a key factor in producing error 54.
Some years later, I interviewed at Knight Capital, just a couple of weeks before their blowup. (Dreadful interview at which I did dreadfully, being asked to write C _over the phone_ by a supremely uninterested engineer. Quite a red flag in retrospect.)
> Therac-25 is a great case study for software engineers too, recommend reading the Wikipedia article for anyone who hasn't, it's not too long.
I re-read the original paper every few months, more frequently if I'm working on Safety-of-Life-Critical equipment. Which, given my day job, means I'm re-reading it every couple of weeks at most.
The audience of this website is disproportionately aware of the Therac-25 compared to the general public. For the obvious reason, engineering, but also geographically: The Therac-25 being a North American incident that affected Canada and the US. Whereas Theryc is a French company.
While I do agree with your point, as a Swede not even born when the incidents happen I still knew about it, was brought up in a computer science class.
Exactly what I thought as soon as I learned the name.
It's like, man, how to kill a product?
No pun intended.
It could even work? But you put yourself behind such a poorly placed 8 ball when you do these things. Even among researchers, people are a little superstitious about stuff like this. It's always in the back of everyone's mind.
> Even among researchers, people are a little superstitious about stuff like this.
Being superstitious is not common in the medical treatment world, where weird product names are common.
A doctor isn’t going to include the device’s brand name in their decision process for treating a cancer patient.
The Therac-25 case study is noted in the medical world but not to the same extent as in engineering. The case was a tragedy of bad engineering, but the doctors involved in directing the treatments were not at fault for the radiation over exposures.
I doubt any of that is valid. Therac-25 happened 44 years ago, that's a very long time, and many people involved in cancer research today weren't even alive when it happened.
"Theryq" and "Therac" are not quite the same either. The word "therapy" and derivatives of it using "thera" are still used widely across the medical industry.
So I'm not really sure why anyone here is making a big deal about the name of the company being "Theryq".
Hey, FLASH finally hit Hacker News! I remember my professors talking about this in graduate school. It's a fairly well-established effect: the tumor selectivity of radiation is much better at ultra-high dose rates. It is still unclear exactly why. But there are a lot of studies about it:
I was starting to infer there was a better focusing ability so it could start and exit as a broad cone of radiation and keep the peak intensity at the tip of the focal cones at the tumor-tissue, and the short pulse also helped the healthy tissue.
But the way this sounds, it's more like a straight beam delivering similar intensity to healthy and tumor tissue but the biological effect strongly differs between healthy vs tumor tissue?
Yes, the radiation dose under the conventional metric (energy divided by mass) is the same, but the effects on biological systems change. I included a little speculation on the chemistry in my response to a sibling comment.
My guess would be that the radiation doesn't itself care but that tumors have some other characteristic (like multiplying rapidly) that makes them more susceptible to it. Similarly to how you can sometimes attack them with medication that inhibits cell division.
Yeah that's the conventional dose rate effect, not the FLASH effect. The FLASH effect happens on timescales so short that ordinary considerations like the cell cycle or DNA repair mechanisms are inherently ruled out. Instead it might have to do with the type of radical species that form in normal cells versus tumors, possibly related to oxygenation, pH, glycolysis byproducts, etc.
The first interaction of radiation with tissue is usually this:
H2O + ħv >> H2O+ + e- (fugitive)
The radical ion H2O+ is extremely reactive and usually protonates another water molecule immediately:
H2O+ + H2O >> H3O+ + OH*
The hydroxyl radical has a half life of about a nanosecond and will usually be the main "reagent", diffusing until it runs into an organic molecule which will be oxidized and thus degraded. At high enough dose rates, the peak concentration of hydroxyl radicals and more stable radicals like superoxide could be much higher, leading to "nonlinear" effects, i.e. byproducts of multiple radicals interacting with each other or a protein.
One thing I found confusing about the nature article is that it mostly discusses conventional linear accelerator + bremsstrahlung X-ray radiation versus very high dose rate FLASH in the form of electron beams, proton beams, or even carbon ion beams.
Do we know that what the chemical mechanism for damage from charged particle beams is? Is it similar enough to compare directly like this? Are the timescales short enough that charge deposition might matter?
The article is a bit unclear, but we have both a very wide range of X-ray vs charged particle studies, and increasingly of conventional vs FLASH studies with a range of modalities (e.g. the seminal FLASH paper was FLASH electrons vs conventional electrons). FLASH photon vs conventional photons are also increasingly being generated, although they've been more of a pain to generate.
So it's clear there is a temporal FLASH effect, which is not purely a question of radiation type.
That's not to say it's necessarily exactly the same effect - we still don't have a perfect quantitative understanding of the effects of different radiation types even at normal dose rates, let alone when FLASH differences are added into the mix.
The major issue isn't the speed of delivery, and the cancer.
The key question is how do you spare normal tissue, and how do you prove the normal tissue is spared in the long term. Current answer is: You break it apart into multiple sessions, the anti-thesis of FLASH.
from the article (pay attention to the part in italics):
FLASH radiotherapy flips the conventional approach on its head, delivering a single dose of ultrahigh-power radiation in a burst that typically lasts less than one-tenth of a second. In study after study, this technique causes significantly less injury to normal tissue than conventional radiation does, without compromising its antitumor effect.
That phrasing isn't perfectly clear, as there's two things at play.
If you're delivering a large dose D all at once, FLASH spares normal tissue compared to conventional rate irradiations, with maintained anti-tumour effect.
But, you can instead deliver your treatment in a number of smaller doses, say n "fractions" of dose d. This also spares normal tissue (1). This latter approach - fractionation - is the way radiotherapy was delivered for most of its history. But at these low doses, FLASH sparing is small to negligible.
So, we have two demands in tension - and its unclear which is actually optimal. Some of the early results in FLASH showed huge sparing, but lots of more recent studies have shown more modest effects which may not be worth giving up benefits of fractionation for(2). And to date I think we have basically no meaningful in-human data to guide this, so there's still a lot of uncertainty.
1 - Fractionation also spares tumours, a bit, but you can offset this by increasing the total dose a bit and still see benefit.
2 - There is a general move to somewhat larger, fewer fractions even in normal radiotherapy, although almost all of these are still below the threshold where FLASH sparing is seen.
The sidebar mentions heavier particles having a pronounced Bragg Peak[0] and also existing approaches like multi-beam targeting. The FLASH effect in the article is yet another tool to limit the surrounding damage.
I recall someone was analyzing the refractive index of various tissues in order to tighten the target area for multi beam radiation therapy. Particularly for brain cancers. By hitting from multiple angles the dosage in surrounding tissues is lower, and by calculating how the head lenses the beam you reduce the high dose area in the middle, like a 3d Venn diagram.
But I don’t remember is whether that experiment became SOP or not.
Seems not in this case. But I believe the use case was deep brain tumors, like the hippocampus, where any beam alignment problems could be life altering.
I generally don't trust cancer-communication if it's juiced up like this incredible headline. There has been huge amounts of progress. We don't need silicon valley idiots starting to make proclamations. It's doing fine without your mediocrity.
51 comments:
> Currently, the most plausible theory emerging from her team’s research points to metabolism: Healthy and cancerous cells may process reactive oxygen species—unstable oxygen-containing molecules generated during radiation—in very different ways.
Reminds me of this which I (think) was linked here a while ago: https://www.nature.com/articles/s12276-020-0384-2
It really does feel like all these piecemeal cancer treatments are converging on something resembling a cure.
There was also a study that showed that chemotherapy efficacy was enhanced by fasting before treatment.
It seems that when calories are scarce, healthy cells turtle up while cancer cells keep consuming, so fasting reduces absorption rates in healthy tissues and thus collateral damage.
Healthy cells CAN turtle-up, whereas cancer cells engage in unregulated reproduction. Also, some cancer cells can only consume glucose. Which, in a fasted state, would mean that the majority of energy would be in ketones(if the individual were metabolically healthy), starving the cancer cells to death.
Why wouldn’t a strict keto diet not be a cure for those cancers?
Because the cancers cells adapt! (fast reproduction and high mutation rate of the cancerous cells make that process quicker than antibiotics resistance)
There was a study that chemotherapy works best in the _morning_. Derek Lowe had an article about this:
Hopefully, this will turn out better than proton therapy, which held similar promises of improvement. https://pmc.ncbi.nlm.nih.gov/articles/PMC11506991/
Theryq - why would they go with this name when everyone in the field knows about the Therac-25 radiation overexposure incidents?
Therac-25 is a great case study for software engineers too, recommend reading the Wikipedia article for anyone who hasn't, it's not too long.
> Previous models had hardware interlocks to prevent such faults, but the Therac-25 had removed them, depending instead on software checks for safety.
https://en.wikipedia.org/wiki/Therac-25
Another interesting part of the story is the user element. The issue was most often triggered by fast, experienced technicians who were able to key commands more quickly than Therac engineers anticipated:
> After strenuous work, the physicist and operator were able to reproduce the error 54 message. They determined that speed in editing the data entry was a key factor in producing error 54.
Therac is the first one I list and Knight Capital is the second. It is in fact possible to bankrupt your company by misusing feature toggles.
I learned about Therac at college in the 90s.
Some years later, I interviewed at Knight Capital, just a couple of weeks before their blowup. (Dreadful interview at which I did dreadfully, being asked to write C _over the phone_ by a supremely uninterested engineer. Quite a red flag in retrospect.)
I feel like you should get yourself a merit badge printed for that, sewed onto your laptop bag.
> Quite a red flag in retrospect.
No pun intended?
> Therac-25 is a great case study for software engineers too, recommend reading the Wikipedia article for anyone who hasn't, it's not too long.
I re-read the original paper every few months, more frequently if I'm working on Safety-of-Life-Critical equipment. Which, given my day job, means I'm re-reading it every couple of weeks at most.
Keeps you sharp, doesn't it?
The audience of this website is disproportionately aware of the Therac-25 compared to the general public. For the obvious reason, engineering, but also geographically: The Therac-25 being a North American incident that affected Canada and the US. Whereas Theryc is a French company.
While I do agree with your point, as a Swede not even born when the incidents happen I still knew about it, was brought up in a computer science class.
I feel like I have heard about this in a million different management anecdotes in my corporate trainings about management stuff/QA.
> The Therac-25 being a North American incident that affected Canada and the US
CGR who provided the accelerators and basic PDP11-based computing platform were a French company.
> Whereas Theryc is a French company.
I have been a Citroën enthusiast for about 30 years. I love French cars.
I have repaired lots of Valeo electronics modules for vehicles.
I'm not sticking my head in a French fucking particle accelerator.
Redemption arc.
Exactly what I thought as soon as I learned the name.
It's like, man, how to kill a product?
No pun intended.
It could even work? But you put yourself behind such a poorly placed 8 ball when you do these things. Even among researchers, people are a little superstitious about stuff like this. It's always in the back of everyone's mind.
> Even among researchers, people are a little superstitious about stuff like this.
Being superstitious is not common in the medical treatment world, where weird product names are common.
A doctor isn’t going to include the device’s brand name in their decision process for treating a cancer patient.
The Therac-25 case study is noted in the medical world but not to the same extent as in engineering. The case was a tragedy of bad engineering, but the doctors involved in directing the treatments were not at fault for the radiation over exposures.
I doubt any of that is valid. Therac-25 happened 44 years ago, that's a very long time, and many people involved in cancer research today weren't even alive when it happened.
"Theryq" and "Therac" are not quite the same either. The word "therapy" and derivatives of it using "thera" are still used widely across the medical industry.
So I'm not really sure why anyone here is making a big deal about the name of the company being "Theryq".
It’s an s-tier case study for UX research though. Maybe the doctors don’t remember but we do.
> It's like, man, how to kill a product?
"This name makes me uncomfortable. I think I'd rather die of cancer."
First thing that leapt out at me.
Hey, FLASH finally hit Hacker News! I remember my professors talking about this in graduate school. It's a fairly well-established effect: the tumor selectivity of radiation is much better at ultra-high dose rates. It is still unclear exactly why. But there are a lot of studies about it:
https://www.nature.com/articles/s41571-022-00697-z
> It is still unclear exactly why
It'll be nice when we figure it out, then we can understand the unintended consequences better.
Not that it should prevent its use or anything; fuck cancer.
Interesting the effect's reason is still unclear.
I was starting to infer there was a better focusing ability so it could start and exit as a broad cone of radiation and keep the peak intensity at the tip of the focal cones at the tumor-tissue, and the short pulse also helped the healthy tissue.
But the way this sounds, it's more like a straight beam delivering similar intensity to healthy and tumor tissue but the biological effect strongly differs between healthy vs tumor tissue?
Yes, the radiation dose under the conventional metric (energy divided by mass) is the same, but the effects on biological systems change. I included a little speculation on the chemistry in my response to a sibling comment.
My guess would be that the radiation doesn't itself care but that tumors have some other characteristic (like multiplying rapidly) that makes them more susceptible to it. Similarly to how you can sometimes attack them with medication that inhibits cell division.
Yeah that's the conventional dose rate effect, not the FLASH effect. The FLASH effect happens on timescales so short that ordinary considerations like the cell cycle or DNA repair mechanisms are inherently ruled out. Instead it might have to do with the type of radical species that form in normal cells versus tumors, possibly related to oxygenation, pH, glycolysis byproducts, etc.
The first interaction of radiation with tissue is usually this:
H2O + ħv >> H2O+ + e- (fugitive)
The radical ion H2O+ is extremely reactive and usually protonates another water molecule immediately:
H2O+ + H2O >> H3O+ + OH*
The hydroxyl radical has a half life of about a nanosecond and will usually be the main "reagent", diffusing until it runs into an organic molecule which will be oxidized and thus degraded. At high enough dose rates, the peak concentration of hydroxyl radicals and more stable radicals like superoxide could be much higher, leading to "nonlinear" effects, i.e. byproducts of multiple radicals interacting with each other or a protein.
One thing I found confusing about the nature article is that it mostly discusses conventional linear accelerator + bremsstrahlung X-ray radiation versus very high dose rate FLASH in the form of electron beams, proton beams, or even carbon ion beams.
Do we know that what the chemical mechanism for damage from charged particle beams is? Is it similar enough to compare directly like this? Are the timescales short enough that charge deposition might matter?
The article is a bit unclear, but we have both a very wide range of X-ray vs charged particle studies, and increasingly of conventional vs FLASH studies with a range of modalities (e.g. the seminal FLASH paper was FLASH electrons vs conventional electrons). FLASH photon vs conventional photons are also increasingly being generated, although they've been more of a pain to generate.
So it's clear there is a temporal FLASH effect, which is not purely a question of radiation type.
That's not to say it's necessarily exactly the same effect - we still don't have a perfect quantitative understanding of the effects of different radiation types even at normal dose rates, let alone when FLASH differences are added into the mix.
For the other readers in this thread, this poster really knows their stuff.
The major issue isn't the speed of delivery, and the cancer.
The key question is how do you spare normal tissue, and how do you prove the normal tissue is spared in the long term. Current answer is: You break it apart into multiple sessions, the anti-thesis of FLASH.
Source: my wife is a radiation oncologist.
from the article (pay attention to the part in italics):
FLASH radiotherapy flips the conventional approach on its head, delivering a single dose of ultrahigh-power radiation in a burst that typically lasts less than one-tenth of a second. In study after study, this technique causes significantly less injury to normal tissue than conventional radiation does, without compromising its antitumor effect.
That phrasing isn't perfectly clear, as there's two things at play.
If you're delivering a large dose D all at once, FLASH spares normal tissue compared to conventional rate irradiations, with maintained anti-tumour effect.
But, you can instead deliver your treatment in a number of smaller doses, say n "fractions" of dose d. This also spares normal tissue (1). This latter approach - fractionation - is the way radiotherapy was delivered for most of its history. But at these low doses, FLASH sparing is small to negligible.
So, we have two demands in tension - and its unclear which is actually optimal. Some of the early results in FLASH showed huge sparing, but lots of more recent studies have shown more modest effects which may not be worth giving up benefits of fractionation for(2). And to date I think we have basically no meaningful in-human data to guide this, so there's still a lot of uncertainty.
1 - Fractionation also spares tumours, a bit, but you can offset this by increasing the total dose a bit and still see benefit.
2 - There is a general move to somewhat larger, fewer fractions even in normal radiotherapy, although almost all of these are still below the threshold where FLASH sparing is seen.
What is the intensity at the focal point versus areas surrounding it?
The sidebar mentions heavier particles having a pronounced Bragg Peak[0] and also existing approaches like multi-beam targeting. The FLASH effect in the article is yet another tool to limit the surrounding damage.
[0] https://en.wikipedia.org/wiki/Bragg_peak
I recall someone was analyzing the refractive index of various tissues in order to tighten the target area for multi beam radiation therapy. Particularly for brain cancers. By hitting from multiple angles the dosage in surrounding tissues is lower, and by calculating how the head lenses the beam you reduce the high dose area in the middle, like a 3d Venn diagram.
But I don’t remember is whether that experiment became SOP or not.
Won't the effective index of all materials be basically 1 for the high energy electrons involved here?
Seems not in this case. But I believe the use case was deep brain tumors, like the hippocampus, where any beam alignment problems could be life altering.
Does this have to do with cell division?
Cellular metabolism from the look of it. Cellular division and metabolism are linked but not synonymous.
However that’s the current theory, of a long line of theories that did not pan out.
Sounds a little too close, in both name and concept, to Therac for my comfort.
How superstitious are you? Do you avoid black cats and ladders?
When corporate arrogance is involved we called it Bayesian inference not superstition.
I avoid any strange animal and walking under ladders.
Those aren’t superstition, just common sense.
I wouldn’t call a ship “Titanic”.
I generally don't trust cancer-communication if it's juiced up like this incredible headline. There has been huge amounts of progress. We don't need silicon valley idiots starting to make proclamations. It's doing fine without your mediocrity.
(they're French, not Silicon Valley)