Very interesting, I didn't know it was the production of ROS that caused DNA damage from radiation. It seems to clamp down on any claim that normal repair mechanisms might not work as well for radiation. I suppose a direct hit would have a strong tendency to be unrepairable and simply result in cell death
Comparisons between the internal rate of ROS production and repairs and that caused by different rates of radiation then become pretty good for contextualizing how much risk there really is from even meltdown levels of radiation for the public
Not necessarily. The repair mechanism based on comparing two homologous strands of DNA could repair more than a single missing nucleotide pair successfully. It simply splices in a good section to replace the damaged fragment. As long as the single hit does not affect two closely spaced regions, I think the effect would be similar to a DSB caused by an ROS.
Good point. I probably should have thrown in a qualifier But for most of the ceil cycle we don't have a separate template, and if we have a DSB we have to rely on Non-Homologous End Joining which does have a problem with closely spaced DSB's.
I had the same thought. An alpha particle with 7273 times the mass of an electron and twice the charge would seem to be a bowling ball compared with an electron marble, and a massless, chargeless photon. But remember when we are talking about direct interactions between these kind of particles, we are getting into quantum weirdness, which is far beyond the ken of this metal bender. "Direct hit" is a possibly misleading analogy.
It would help the reader to come to their own conclusion if one could plot dose rate on the X axis and total dose on the Y axis. Another two plots might show estimated total dose due just to neutrons and estimated total dose due to gammas. Also it would help to know what the dose is from in figure 1 - is this a plot from gammas, neutrons, a mix?
The Figure 1 dose was all photon (gamma). It would be nice if we had similar plots for alphas and electrons, but I hav enot found any.
Not following your request for a dose/dose rate plot. Are we talking about the Green Table? Perhaps we need a column indicating what particle(s) produced the dose? or maybe some kind of color scheme since we are out of room.
For LNT falsification, the four most important exposures are:
1) The dial painters (almost all alpha),
2) The Radithor drinkers (almost all alpha)
3) The Taipei apartment dwellers (all photon).
4) The Karunagappally cohort (almost all photon)
In the case of neutrons, only the bomb survivors and a few of the Chernobyl first responders got any neutrons. But for most of the bomb survivors most of the dose was photons, although the fraction is debatable.
Excellent description of the DNA damage process. I too had been wondering about the difference between direct damage and damage via ROS.
A bit off topic, but I have to quibble with your statement "the signal to noise ratio is just too small to separate the effects of radiation from the myriad of confounding factors. At near average background dose rates, it is impossible to falsify any semi-reasonable model."
I believe the Lung Cancer vs Radon data does exactly that, and actually provides one of the more understandable falsifications of LNT. (The jury will believe scatterplots over statistical sophistry from even the most prestigious expert.)
With enough data, you CAN separate the confounding factors. I won't fill this space with the entire analysis, but please have a look at my initial draft.
We can rule out the most obvious confounder - smoking. The one variable, elevation, which DOES show some correlation with radon, goes the wrong way for LNTers. Counties with higher elevation have LESS lung cancer, the opposite of what must happen if we believe in LNT.
I agree that the Cohen study is farily strong evidence against LNT, but it is far from falsification. At these dose rate profiles any effect would be tiny. Smoking is orders of magnitude stronger. Cohen only had smoking data by state. Suppose smoking is higher in urban areas when radon DRP's are small and lower inrural areas where radon DRP's are larger. That wouldmake radon look hormetic.
Scatter diagrams canbe manipulated just as easily as any other graphic.
Should have looked at your paper before opening my big mouth. It seems you have smoking data by county, which Cohen did not. Where did you get it?
However the same objection applies but much more weakly. Basically, if you maintain you can measure radiation induced cancer in near background dose rate profiles you are assuming radiation is much more powerful than it is.
Your data does not disprove the null hypothesis of no effect. But we know there is cancer if the dose rate profiles becomes harmful enough. The issue is how do we combine undetectable harm for some dose rate profile with detectable harm for others. I suggest we have to do it smoothly. SNT does that.
I've imported the CSV data to a Google Sheet, added notations, column headers, etc. to make it easy for anyone wanting to do their own analysis, or challenge mine.
So if we see a 30% variation in lung cancer, depending on radon, that is significant.
Lung cancer shows a similar variation with smoking, but not orders of magnitude stronger than radon. Also there is no correlation between smoking and radon, so smoking cannot be a confounder.
This is where the debate in FaceBook ends. The anti's simply refuse to look at the data, and just repeat their arguments from authority.
Cohen did a good job of disproving LNT, even with his limited data, but his argument depends on advanced statistics, easy to ignore in a FaceBook debate.
The purpose of my paper is to make the argument without any advanced statistics, no "meta analysis", just raw data.
I'm not suggesting we discard SNT. It doesn't model these radon responses, but is perfect as a reasonable upper bound on what can happen in an NPP release.
But of course smoking can be confounder. If people in low radon situations smoke more than people in high radon locations, and you don't somehow properly correct for it. you will make radon seem hormetic. If you only have smoking data at the county level, you can't correct for possible intra-county relationships.
I don't know the exact numbers, but smoking increases lung cancer by a whole, whole lot. Lung cancer was an extremely rare disease among women until they were conned into starting smoking, and then it jumped up I think by orders of magnitude among smokers.. If you think radiation can do something similar, you should probably rethink your support of nucelar power.
I should have been much more complimentary. The use of county wide smoking data instead of state wide is a big improvement over Cohen's work. I think your work needs much broader dissemination.
There are some statements which I don't think are supported by the data. I'll send these by email.
Thanks for the warning. I'll take a pass.
Very interesting, I didn't know it was the production of ROS that caused DNA damage from radiation. It seems to clamp down on any claim that normal repair mechanisms might not work as well for radiation. I suppose a direct hit would have a strong tendency to be unrepairable and simply result in cell death
Comparisons between the internal rate of ROS production and repairs and that caused by different rates of radiation then become pretty good for contextualizing how much risk there really is from even meltdown levels of radiation for the public
Not necessarily. The repair mechanism based on comparing two homologous strands of DNA could repair more than a single missing nucleotide pair successfully. It simply splices in a good section to replace the damaged fragment. As long as the single hit does not affect two closely spaced regions, I think the effect would be similar to a DSB caused by an ROS.
Matthew,
Good point. I probably should have thrown in a qualifier But for most of the ceil cycle we don't have a separate template, and if we have a DSB we have to rely on Non-Homologous End Joining which does have a problem with closely spaced DSB's.
I had the same thought. An alpha particle with 7273 times the mass of an electron and twice the charge would seem to be a bowling ball compared with an electron marble, and a massless, chargeless photon. But remember when we are talking about direct interactions between these kind of particles, we are getting into quantum weirdness, which is far beyond the ken of this metal bender. "Direct hit" is a possibly misleading analogy.
It would help the reader to come to their own conclusion if one could plot dose rate on the X axis and total dose on the Y axis. Another two plots might show estimated total dose due just to neutrons and estimated total dose due to gammas. Also it would help to know what the dose is from in figure 1 - is this a plot from gammas, neutrons, a mix?
Lars,
The Figure 1 dose was all photon (gamma). It would be nice if we had similar plots for alphas and electrons, but I hav enot found any.
Not following your request for a dose/dose rate plot. Are we talking about the Green Table? Perhaps we need a column indicating what particle(s) produced the dose? or maybe some kind of color scheme since we are out of room.
For LNT falsification, the four most important exposures are:
1) The dial painters (almost all alpha),
2) The Radithor drinkers (almost all alpha)
3) The Taipei apartment dwellers (all photon).
4) The Karunagappally cohort (almost all photon)
In the case of neutrons, only the bomb survivors and a few of the Chernobyl first responders got any neutrons. But for most of the bomb survivors most of the dose was photons, although the fraction is debatable.
Excellent description of the DNA damage process. I too had been wondering about the difference between direct damage and damage via ROS.
A bit off topic, but I have to quibble with your statement "the signal to noise ratio is just too small to separate the effects of radiation from the myriad of confounding factors. At near average background dose rates, it is impossible to falsify any semi-reasonable model."
I believe the Lung Cancer vs Radon data does exactly that, and actually provides one of the more understandable falsifications of LNT. (The jury will believe scatterplots over statistical sophistry from even the most prestigious expert.)
With enough data, you CAN separate the confounding factors. I won't fill this space with the entire analysis, but please have a look at my initial draft.
https://docs.google.com/document/d/1bTrkJSvrq-hzHaiE5WJBcPZOdkUoLHJb7Y3p13BLyUw/edit?tab=t.0#heading=h.54fwqtl8kxaw
We can rule out the most obvious confounder - smoking. The one variable, elevation, which DOES show some correlation with radon, goes the wrong way for LNTers. Counties with higher elevation have LESS lung cancer, the opposite of what must happen if we believe in LNT.
Here we go again.
I agree that the Cohen study is farily strong evidence against LNT, but it is far from falsification. At these dose rate profiles any effect would be tiny. Smoking is orders of magnitude stronger. Cohen only had smoking data by state. Suppose smoking is higher in urban areas when radon DRP's are small and lower inrural areas where radon DRP's are larger. That wouldmake radon look hormetic.
Scatter diagrams canbe manipulated just as easily as any other graphic.
Should have looked at your paper before opening my big mouth. It seems you have smoking data by county, which Cohen did not. Where did you get it?
However the same objection applies but much more weakly. Basically, if you maintain you can measure radiation induced cancer in near background dose rate profiles you are assuming radiation is much more powerful than it is.
Your data does not disprove the null hypothesis of no effect. But we know there is cancer if the dose rate profiles becomes harmful enough. The issue is how do we combine undetectable harm for some dose rate profile with detectable harm for others. I suggest we have to do it smoothly. SNT does that.
The data is from a study by Simeonov et al. https://dfzljdn9uc3pi.cloudfront.net/2015/705/1/county-data.txt
A good review of this data is found in section 7 of a paper by SARI:
Evidence for Radiation Hormesis.
https://www.x-lnt.org/evidence-for-radiation-hormesis
I've imported the CSV data to a Google Sheet, added notations, column headers, etc. to make it easy for anyone wanting to do their own analysis, or challenge mine.
https://docs.google.com/spreadsheets/d/1rHhUX7frt0ZM_E_JRzhlBHwymXHB8PulgZtYIm-JaXA/edit?gid=736776170#gid=736776170
Background radiation is small, but radon is the largest part of it, cosmic rays a distant second.
https://www.epa.gov/radiation/radiation-sources-and-doses
So if we see a 30% variation in lung cancer, depending on radon, that is significant.
Lung cancer shows a similar variation with smoking, but not orders of magnitude stronger than radon. Also there is no correlation between smoking and radon, so smoking cannot be a confounder.
This is where the debate in FaceBook ends. The anti's simply refuse to look at the data, and just repeat their arguments from authority.
Cohen did a good job of disproving LNT, even with his limited data, but his argument depends on advanced statistics, easy to ignore in a FaceBook debate.
The purpose of my paper is to make the argument without any advanced statistics, no "meta analysis", just raw data.
I'm not suggesting we discard SNT. It doesn't model these radon responses, but is perfect as a reasonable upper bound on what can happen in an NPP release.
Nice job methodologically.
But of course smoking can be confounder. If people in low radon situations smoke more than people in high radon locations, and you don't somehow properly correct for it. you will make radon seem hormetic. If you only have smoking data at the county level, you can't correct for possible intra-county relationships.
I don't know the exact numbers, but smoking increases lung cancer by a whole, whole lot. Lung cancer was an extremely rare disease among women until they were conned into starting smoking, and then it jumped up I think by orders of magnitude among smokers.. If you think radiation can do something similar, you should probably rethink your support of nucelar power.
David,
I should have been much more complimentary. The use of county wide smoking data instead of state wide is a big improvement over Cohen's work. I think your work needs much broader dissemination.
There are some statements which I don't think are supported by the data. I'll send these by email.
I made it to the end, not too bad. I guess I am an outlier in the choir!
This makes sense from my point of view, someone with 40 years working at nuclear power plants.
LNT and ALARA need to be replaced by something more reasonable, like SNT as advocated by our highly educated author.