Friday, December 23, 2011

Will a Little Bit of Radiation Hurt You? Berkeley Scientists Break the Hold of Linear Non Threshold Reasoning.

Berkeley Scientists Watch Cells Repairing Themselves after Exposure to Low-Level Radiation

Lawrence Berkeley laboratory scientists recently announced results of innovative radiation experiments. The scientists watched cells repairing themselves after being exposed to low-level radiation. Here's the press release: New Take on Impacts of Low Dose Radiation: Berkeley Lab Researchers Find Evidence Suggesting Risk May Not Be Proportional to Dose at Low Dose Levels

Here's a quote from the press release:

“Our data show that at lower doses of ionizing radiation, DNA repair mechanisms work much better than at higher doses,” says Mina Bissell, a world-renowned breast cancer researcher with Berkeley Lab’s Life Sciences Division. “This non-linear DNA damage response casts doubt on the general assumption that any amount of ionizing radiation is harmful and additive.”

Mina Bissell is a member of the National Academy of Sciences, and is known throughout the world for her breast cancer research. Her awards include being elected a Fellow of the National Academy of Sciences, a Fellow of the Royal Society of Chemistry, and further honors from the Curie Institute in France and the American Cancer Society. She has so many awards that Portugal has named an award FOR her: the Mina J Bissell award to a scientist who has transformed our perception of a topic.

Though the techniques were innovative, and the scientists very distinguished, these recent low-level radiation results are not actually surprising. From common sense, we would expect that a body could heal itself well from several low-dose radiation exposures, while having a harder time healing from a single high dose exposure. After all, we expect ourselves to heal after any number of skinned knees, but that says nothing about our ability to heal from stab wounds.

Why Is This A Big Deal? Well, The Official Theory of Radiation is Response is Strictly Linear

In other words, the official theory of radiation response is that a couple of skinned knees is the same as a stabbing.

Risks from low levels of radiation are generally assessed with the Linear No Threshold (LNT) theory. This theory states that the risks of cancer are in direct proportion to the amount of exposure, even for very small exposures. This methodology of risk assessment is supported by a series of reports called the BEIR reports: Biological Effects of Ionizing Radiation. BEIR I through BEIR VII have been issued under the auspices of the National Academy of Sciences. I include a link to the BEIR VII summary document from the National Academy.

Despite this heritage, the BEIR reports are controversial. For one thing, the assumption that there is no threshold for radiation damage is impossible to prove in a world where everyone is exposed to background radiation. For another thing, the linearity of damage at low levels always struck many scientists as unlikely. In general, small amounts of toxins are handled well by organisms, while large amounts can kill. The relationship between zero effect and lethal dose is not generally linear.

Or, as one of my friends put it: BEIR says that if 100 aspirins taken at once is lethal, and 100 men each take one aspirin apiece--one of those men is going to die. That would be a linear response to low dosage aspirin consumption.

Actually, BEIR VII waffles a bit at low dosages. Here's a quote from the summary:

There are two competing hypotheses to the linear no-threshold model. One is that low doses of radiation are more harmful than a linear, no-threshold model of effects would suggest. BEIR VII finds that the radiation health effects research, taken as a whole, does not support this hypothesis. The other hypothesis suggests that risks are smaller than predicted by the linear no-threshold model are nonexistent (sic), or that low doses of radiation may even be beneficial. The report concludes that the preponderance of information indicates that there will be some risk, even at low doses, although the risk is small.

Note that the report does not actually claim that the risks are linear at low doses, but merely that there is some risk though the risk is small.

If you would like to think about some painful consequences of LNT, I encourage you to listen to this podcast about Fukushima or read my blog post about it. Rod Adams invited me and Cal Abel to be on a podcast to discuss the very real question of when Fukushima evacuees could go home. I especially encourage you to read the comments on both the podcast and the blog post.

Following LNT Would Lead to Really Weird Cancer Therapies

"When all is said and done, there's a lot more said than done."

I have always followed the policy of attempting to look at what people do, rather than what they say. People still get dental x-rays. People move to Denver, where background radiation is high. More people move to Denver than move to the Mississippi delta, despite the low background radiation in Mississippi. But let's look at a stronger example.

Let's look at radiation therapy for cancer. Many people undergo radiation treatmentS. The plural is important.

Radiation oncologists know that the healthy flesh heals after a treatment, so treatments are spaced out: six weeks of three times a week, two months of twice a week, etc. I remember one man in my writing group who moved to Boston for two months to be treated. Why did he have to do this? Why didn't he just visit Boston, get one massive treatment, and come home to Vermont? Because, even at high doses, the body heals between treatments. There is a repair mechanism, as there is for everything else our body encounters.

Giving radiation therapy in one dose would hugely damage the surrounding tissues, compared to giving the same amount in smaller doses over a long time. Even at the relatively high doses of radiotherapy, response to one dose versus response to many doses is not linear. Wade Allison has an excellent description of the planning behind radiation therapy in his book, Radiation and Reason.

LNT and Small Doses

From low-level radiation exposure research at Berkeley to the standard methodologies of radiation oncology specialists to seems to me that it is time to stop using LNT as the gold standard for assessing risk.

LNT is wrong and therefore will lead to incorrect results. Even BEIR doesn't really defend it, and yet it is the de facto law of the land.

A major tip of the hat to Rod Adams of Atomic Insights blog, for his post Lawrence Berkeley National Lab announces breakthrough study of low radiation dose effects


Howard Shaffer said...

Great Blog. But will Sally Shaw be reassured?

In addition to "any amount of radiation is dangerous" opponents of nuclear power act as though radiation comes only from man-made sources. Thus we have two wrong ideas to correct.

The Japanese government is supposedly wrestling with how much radiation exposure to allow citizens each year when they move back to the homes evacuated. There is a competing health effect. It is the stress of living as an evacuee. Medical science knows that stress has real detrimental health effects.

I've seen the figure a 20 msv per year as the amount of exposure discussed. I have received more than twice that this year for medical purposes (2 CT scans and operating table fluoroscope). The highest exposure per year from the environment is in Ramsar, Iran at 300 msv.

Scientifically, for the Japanese government to allow 20 msv per year is straight forward and safe. Politically it will be difficult.

Meredith Angwin said...

Hi Howard. As you and I well know, nothing will convince Shaw, or Shadis, or Katz, or...

They are not my target audience! I hope to inform people of these new research results...people who are not terrified of radiation. People who make a career of fearing radiation will continue with their career path, no doubt.

Fourteen elderly Japanese died during the evacuation, as a direct result of their bus trips, problems with IVs and hydration, etc. The Japanese government should think about that, also.

Rod Adams said...


Thank you for producing a well reasoned post questioning our continued acceptance of the Linear No Threshold dose response ASSUMPTION and for pointing out the damage that gets done due to the resulting statement that there is "no safe dose of radiation".

I have a stalwart member of the radiation protection guild who regularly comments on my blog in defense of the LNT. He acknowledges that the risk at low doses is small and that the risk at very low doses is very small, but he believes that it is irresponsible to tell people that there is NO risk. His science cannot measure the risk, but he and his colleagues assumes that it is PRUDENT to acknowledge the possibility that some risk exists.

I strongly disagree. If I can tell someone that it is safe to take an automobile trip with me, I should be able to just as honestly say that it is safe to live in an area with background radiation doses of 20 mSv (2 Rem) per year. There is no need to confuse the issue or the target audience with details about how safe it is. It is safe enough.

As you pointed out, the LNT is based on an assumption. Calabrese recently told the story of how Muller ignored evidence of a threshold dose that came from one of his own students because he had an agenda of trying to stop nuclear weapons testing.

I believe that Shaw, Shadis, Lovins, Katz, Lyman, Bradford, Caldicott and dozens of others will also ignore the evidence because it conflicts with their agenda. However, their agenda is much less moral than Muller's; I can agree on the need to use whatever techniques are needed to halt atmospheric nuclear weapons testing; I cannot agree with those who apply those same techniques to halt nuclear energy development.

The primary beneficiaries of excessive radiation fear are the people who sell coal, oil and natural gas. There are also beneficiaries in the radiation protection guild and among the corporations that build the exceedingly expensive structures that are designed to reduce radiation levels to absurdly low levels, even under worst case accident conditions.

TheHealthPhysicist said...

This study doesn't refute LNT, it supports it. LNT includes a DDREF (Dose & Dose Rate Effectiveness Factor). This study shows that DNA damage is better repaired at low doses, which is what the DDREF accounts for.

Meredith Angwin said...

Bob. Thank you for the comment. I think we are talking about different LNTs.

My view of LNT is that it means Linear Non Threshold, even at low doses. That's what Linear means. I am a chemist, and a linear function has one slope over the range of the function. I mean, that's the definition.

You say that your understanding of LNT is that there are DDREFs that acknowledge that DNA damage is better repaired at low doses. That would mean there is a different slope at low doses. Which means the function is not LINEAR down to zero. Now, the function may still be called Linear Non Threshold in the vernacular, but you are saying it isn't really linear at low doses because the DDREFs change the slope.

If that is the case, I DO think they should rename it. However, I am not good at getting functions renamed! My husband is a mathematician, and he once threw one of my chemistry books across the room because "The Dirac Delta Function is NOT a function!" They never renamed it though ;-)

I guess we are stuck with the name LINEAR Non-Threshold!

TheHealthPhysicist said...


You don't seem to understand the epidemiology. One observes cancer incidence as a function of dose. There are error bars in fitting the data. Overall, the best fit is linear-quadratic. However, in the low dose range, there is no statistical difference between linear modeling and linear-quadratic modeling. This is because the quadratic term decreases as the square of the dose reduction (ie, reduce the dose by 2, the quadratic term significance decreases by 4). So, parsimonously, we just say LNT in the low dose range. Linear is linear...but with error bars one can draw lines of different slopes within the error bars. In the "big picture" the response is still LNT.

Meredith Angwin said...

Huh? In other words, Linear Non Threshold means "this function includes various terms of higher orders, such as exp2".

With all due respect, I don't think I am not "getting" the epidemiology. I am very familiar with all sorts of functions. I am familiar with least-squares fitting of functions. However....Functions with higher-power terms are not linear functions.

I am beginning to think the name is quite misleading.

TheHealthPhysicist said...

No the term is not misleading. It's just that many people don't understand the context. It reminds me of people who ask, "if evolution is true, why are there still monkeys?". Those people don't understand evolution.

See the figure in this link:

Overall, the dose response is L-Q. Below 1.5 Sv, the linear curve and L-Q curve can be drawn through the same data points. They are very similar (linear) through the region labeled "Low Dose Range", which is the region of typical occupational and public exposures.

I hope that helps.

Joffan said...

LNT doesn't include DDREF; it has been tacked on as an alteration to avoid the embarrassment of being disproved.

ICRP pub'n #99 gives the game away: "Unless the existence of a threshold is assumed to be virtually certain, the effect of introducing the uncertain possibility of a threshold is equivalent to that of an uncertain increase in the value of DDREF, i.e. merely a variation on the result obtained by ignoring the possibility of a threshold."

As I interpret it - admitting a threshold can be avoided by introducing DDREF. But of course that model is no longer linear with high doses, so has no experimental justification for assuming other than zero effect.

Bob Applebaum seems to have recovered from his initial dismay at my pointing this out to him and has now embraced DDREF, ignoring all attempts to point out its ill-foundedness.

TheHealthPhysicist said...

Joffan doesn't seem to understand the DDREF or threshold. DDREF is introduced because of observations. The observed cancer risk is less when the total dose is less or when the dose rate is less. If the cancer rate were the same regardless of dose or dose rate, that's what we would observe, and the current risk estimates would be much higher.

There is a radiation threshold. It is an energy threshold. Below the lower energy range of UV there is no cancer induction. Above it, the photons possess the energy to cause changes which contribute to cancer. Above the x-ray energy range, each photon has a non-zero possibity (ie, risk) of contributing to cancer. Therefore, at those energies and above, there is no threshold.

Finrod said...

I expect that at some point hormetic effects will be confirmed beyond any reasonable doubt. I guess the LNT priesthood will then set out to convince the rest of us that the resulting dose-response curve is still nonetheless mystically linear, confounding the ignorant assumptions of us non-theologians.

TheHealthPhysicist said...

Anyone (including Finrod) is free to look at the data-generated dose response curve which I provided earlier. I don't see anything mystical. I see a line, more or less.

What's mystical is why someone thinks it's mystical. Actually that's not mystical. There's a set of psychological phenomenon known as anchoring, confirmation bias, and motivated reasoning. In light of evidence contrary to emotionally held "beliefs" cognitive dissonance results.

And so, climatologists are referred to as a "priesthood" by global warming denialists, biologists are referred to similarly by evolution denialists, doctors are the "priesthood" according to folks who wanted to link vaccines to autism even though there was zero evidence.

And so, like all the denialists each has no doubt that at some point there "beliefs" will be confirmed. And yet, as decades pass, that time never comes. Sort of like messiahs.

Huw Jones said...

Perhaps the risk at low dose/dose rates in non-zero. Or perhaps it's not. Perhaps we'll never truly know. From what I've read studying the topic for my degree, finding credible evidence on the topic is incredibly difficult. A more important question is, does it really matter? If biological effects are this hard to detect at low doses/dose rates, then they are obviously not worth worrying about in relative terms. Money spent on zealous radiological protection would be better spent on anti-smoking campaigns, safer driving schemes etc - which cost many millions of lives per year.

As nuclear advocates, we are spending far too much time on this debate. Specifically, in my opinion we are too interested in the 'NT' in LNT, and not enough in the 'L' - Linear. When I first read about DDREF I faced a similar confusion to what Meredith describes above - with this factor included, the 'Linear' Non-Threshold becomes more like the 'Geometric-proportional' Non-Threshold.

Bill said...

"The other hypothesis suggests that risks are smaller than predicted by the linear no-threshold model are nonexistent (sic), or that low doses of radiation may even be beneficial."

The quoted sentence is missing a comma after "model":
'The other hypothesis suggests that risks are smaller than predicted by the linear no-threshold model, are nonexistent, or that low doses of radiation may even be beneficial.'

Or more clearly,
'The other hypothesis suggests that risks (A) are smaller than predicted by the linear no-threshold model, (B) are nonexistent, or (C) that low doses of radiation may even be beneficial.'

Meredith Angwin said...

Thank you all for your recent comments.

Bill: yes, I noticed the BEIR statement was missing a comma. In my post, I did a cut and paste from the official summary document, and included (sic) showing that I knew there was a problem with the language or punctuation or something. This LNT area is very controversial, and I was concerned that if I ADDED anything, even a comma, someone would say I had misquoted BEIR!

Bob. The standard definition, learned in high school of mathematical terms is that a LINEAR function has no higher is in the form y =a +bx. QUADRATIC functions have square terms for the variables. For example y= a +bx+cx^2. There is no such thing as a linear-quadratic equation in mathematics or standard statistics. The equation either has or does not have terms in x^2.

I discovered that in one area, and one area only, the term linear-quadratic is used. That is in dose/oncology calculations. Now, I don't see any reason why this area has to have unique mathematical constructs. Radiation dose response is a statistical area, and mathematicians and statisticians derive tables for life expectancies, LD 50 calculations, crop yields, gene cross-over probabilities, tax revenues and stock market volatility without linear-quadratic. This term is simply is some kind of obfuscation of the term quadratic. The functions either have x^2 terms, or they don't.

At any rate, Bob, I am NOT going to continue this conversation unless you are willing to acknowledge that there are standard descriptions of mathematical functions and equations. That linear and quadratic are separate standard descriptions, depending upon the power of X in the equation.

If you insist that there is some entirely new type of equation, used only in your specialty, called "linear quadratic"...well, there is really no arguing with that sort of obfuscation of standard mathematics.

No matter what the equation is about, x squared makes it a quadratic equation. This isn't about doses or responses. It is about mathematics.

Huw Jones. Yes. We are spending way too much time on this debate. It reminds me of when people thought high-voltage lines might cause leukemia. Huge numbers of studies were done, looking for some subtle effect, which was never found.

Anonymous said...

I agree with the posters who come to the conclusion that the naming is misleading/confusing. I'm wondering if next it will be called the "Maybe-Sorta-CouldBe-If-We-Pretext-That-There-Is-No-Threshold-LNT". Of course, we are taught in logic class that in cases like this it is probably best just to whip out Occam's Razor and say the simplest explanation is best: we can make it linear if we recognize there is a threshold for harmful effects. That sure beats thrashing around with linear-quadratic and DDREF fudge factors simply to justify an initial assumption (LNT).