The following content is
provided under a Creative Commons license. Your support will help
MIT OpenCourseWare continue to offer high quality
educational resources for free. To make a donation or to
view additional materials from hundreds of MIT courses,
visit MIT OpenCourseWare at ocw.mit.edu. MICHAEL SHORT: All right, guys. So today I'm not going to be
doing most of the talking. You actually are,
because, like I've said, we've been teaching you
all sorts of crazy physics and radiation biology. We've taught you how
to smell bullshit, taught you a little bit
about how to read papers and what to look for. And we're going to spend the
second half of today's class actually doing that. Well, we're going to have a
mini debate on whether or not hormesis is real. And you guys are going to
spend some time finding evidence for or against it. Instead of just me telling
you this is what hormesis is or isn't. So just to finish up the
multicellular effects from last time,
we started talking about what's called the
bystander effect, which says, if a cell is
irradiated, and it dies or something happens to it,
the other cells nearby notice. And they speed up
their metabolism, their oxidative
metabolism, which can generate some of the
same chemical byproducts as radiolysis does,
causing additional cell damage and mutation. And there was an interesting-- yeah, I think I left-- we
left off here at this study, where they actually
talked about most of the types of mutations
found in the bystander cells were of different types. But there were mutations
found, in this case, as a result of what's called
oxidative-based damage. This is oxidative
cell metabolism ramping up and producing more of
those metabolic byproducts that can damage DNA as well. What we didn't get
into is the statistics. What do the statistics look
like for large sample sizes of people who have been exposed
to small amounts of radiation? I'm going to show
you a couple of them. One of them is the folks within
3 kilometers of the Hiroshima. So I want you to notice
a couple of things. Here is the dose in gray,
maxing out at about two gray. And in this case
this ERR is what's called Excess Relative Risk. It's a little
different than odds ratio, where here an
excess relative risk of 0 means it's like
nothing happened. So anything above 0 means
extra excess relative risk. So what are some of the features
you notice about this data? What's rather striking
about it in your opinion? Yeah? Charlie? AUDIENCE: [INAUDIBLE]
so in the [INAUDIBLE] timeline from [INAUDIBLE]
timeline here. MICHAEL SHORT: This one? AUDIENCE: Yeah. MICHAEL SHORT: Oh, yeah,
these are the errors. Yep. What does it say here? Is it-- more than one
standard error Yeah. AUDIENCE: There's a
lot of variability? MICHAEL SHORT:
Yeah, I mean, look at the confidence in
this data at high doses. And then while you may say,
OK, the amount of relative risk per amount of
radiation increases with decreasing dose, which
is the opposite of what you might think, our
confidence in that number goes out the window. Now what do you think of the
total number of people that led to each of these data points? How many folks do you
think were exposed to gray versus milligray of radiation? AUDIENCE: A lot less for
gray than [INAUDIBLE].. MICHAEL SHORT: That's
right, the sample size. I thought it was cold
and loud in here. The sample size for the folks
in gray is much smaller. And yet the error bars
are much smaller too. That's not usually the
way it goes, is it? Usually, you think larger
sample size, smaller error bars, unless the effects themselves
and confounding variables are hard to tease out
from each other. If you then look at
another set of people, all of the survivor--
oh. yeah, Charlie? AUDIENCE: How did they determine
the-- the doses [INAUDIBLE]?? MICHAEL SHORT: This would
have to be from some estimate. This would be from models. It's not like folks had
dosimeters everywhere in Japan in the 1940s. But this-- these
would be estimates depending on where
you lived, let's say in an urban,
suburban, or rural area, let's see, things
like milk intake right after the bomb, or
anything that would have given you an unusually high
amount of radiation, distance where the
winds were going. This is the best you
could do with that data. And now look at all
of the bomb survivors, including the ones outside
3 kilometer region, but still got some dose. What's changed? AUDIENCE: It seems like
they're less likely to get more risk for less dose. MICHAEL SHORT:
Yeah, the conclusion is almost flipped for
the low dose cases. If you put them side by side,
depending on the folks living within 3 kilometers of
the epicenter of Hiroshima versus anyone exposed,
all the bomb survivors, you get an almost opposite
conclusion for low doses, despite the numbers
being almost, you know, within each
others confidence intervals for high doses. So what this tells us is
that the effects of high dose are relatively easy to
understand and quite obvious even with low sample sizes. What is different between
these two data sets? Well, it's the only difference
that's actually listed here. Distance from the
epicenter, right? So before I tell you
what's different, I want you guys to try
to think about what could be different
about the folks living within 3 kilometers of
the epicenter of Hiroshima versus anyone else in the
city or the countryside? Yeah? AUDIENCE: Would it
be like [INAUDIBLE]?? It seems like a the
closer, like, it would be a lot more instances
where you get a higher dose. So they're underestimating
[INAUDIBLE].. MICHAEL SHORT: Could be, yeah. It might be harder to figure out
exactly how much dose folks had without necessarily
measuring it, right? But what other major factors
or confounding variables are confusing the data here? Yeah? AUDIENCE: Wouldn't a lot
of people who lived closer, like, not inside
the radiation, like, the actual shockwave and heat
from the bomb [INAUDIBLE]?? MICHAEL SHORT: So in this case,
these are for bomb survivors. So, yes, that's true. If you're closer, you
get the gamma blast. You get the pressure wave. AUDIENCE: But like, even if
you survive that, it still like would affect them in
addition to radiation. Is it counting for people who
got injured from that too? MICHAEL SHORT: It should just
account all survivors, yeah. AUDIENCE: So if
they were injured, that could change how they
reacted to the radiation exposure. MICHAEL SHORT: Sure. Absolutely. And then the other
big one is, actually, someone's kind of mentioned it,
but in passing, urban or rural. The environment that
you live in depends on how quickly, let's say, the
ecosystem replenishes or not if you live in a city or
what sort of other toxins or concentrated
sources of radiation you may be exposed to by
living in a city that's endured a nuclear attack
or something else. It could also depend on
the amount of health care that you're able to receive. If you show some
symptoms of something, if you live way out
in the countryside, and there weren't
a lot of roads, then maybe you can't get
to the best hospital, or you go to a clinic that
we don't know as much. The point is, there's a lot
of confounding variables. There's a lot more people. But anything from
like lifestyle, to diet, to relative exposure,
think about the differences in how folks in the city
and out in the countryside may have been exposed
to the same dose, because, again, dose is given
in gray, not in sieverts. That's the best we can estimate. But would it matter
if you were to exposed to let's say, alpha-particle
containing fallout that you would
then ingest versus exposed to a lot of gamma
rays or delayed betas. It absolutely would. So the type of radiation and the
route of exposure in the organs that were affected are not
accounted for in the study because, again, the
data is in gray. It's just an estimated
joules per kilogram of radiation exposure, not
taking into account the quality factors for tissue, the quality
factors for type of radiation, the relative exposure,
the dose rate, which we've already
talked about. How much you got as a function
of time actually does matter. So all these things
are quite important. And for all these
sorts of studies, you have to consider
the statistics. So let's now look at a-- I won't say, OK, a
cellphone-like study where one might draw a
conclusion if the error bars weren't drawn. So based on this, can you
say that very low doses of radiation in
this area actually give you some increased
risk of, what do they say, female breast cancer? No. You can't be bold enough to
draw a conclusion from the very low dose region from, let's
say, the-- the 1s to 10s of milligray, that whole
region right there that people are afraid of getting,
we don't actually know if it hurts or it has
nothing, or if it helps. That's a kind of weird
thing to think about. So the question is,
what do we do next? These are the actual
recommendations from the ICRP. And I've highlighted
the parts that are important, in my opinion,
for everyone to read. And the most important
one, probably we'll have to come to terms
with some uncertainty in the amount of damage that
little amounts of dose do. So this is the ICRP saying
to the general public, you guys should chill out. There's not much we can do
about tiny amounts of exposure. They happen all the time. You can either worry about it,
and get your heart rate up, and elevate your
own blood pressure, and have a higher chance
of dying on your own, or you can just chill
out because there is not enough evidence to say
whether a tiny little amount of radiation, and we're talking
in the milligray or below, helps, or hurts, or does
nothing, which leads me into the last set of slides
for this entire course, they're not that long because
I want you guys to actually do a lot of the work here,
is radiation hormesis, real or not? There are plenty of studies
pointing one way or the other. And I want to show you a few of
them with some other examples. The whole idea here is that
a little bit of a bad thing can be a good thing,
much like vitamins, or, let's say, vitamin A in
seal livers, a little bit of it you need. It's a vital micronutrient. A whole lot of it can do
a whole lot of damage. You don't usually think of that
being the case for radiation. But some studies may have
you believe otherwise with surprisingly
high sample sizes. So the idea here is that if
you've got anything, not just element and diet, but
anything that happens to you, there's going to be some
optimum level where you could die or have some ill
effects if exposed to too much or too little. We all know that this happens
with high amounts of radiation. The question is, is
that actually happened? So let's look at
some of the data. In this case, I mentioned
selenium and actually have a fair bit of this
data that shows some, let's say, contradictory
results in this case, where a whole lots of
different people were exposed to a certain amount
of selenium accidentally. I don't think these were
any intentional studies. But some folks received
massive doses of selenium and tried-- folks tried
to figure out, well, what how-- oh, yeah, if you
want to see how much they got. Remember that you want about 5
micrograms per day on average. That's a pretty crazy
amount of selenium that ended up killing
this person in four hours. But let's look at a sort of
medium dose, something way higher than you
would normally get. Two different studies published
in peer-reviewed places-- this one says, "taking
mega doses of selenium," so enormous doses, "may
have acute toxic effects and showed no
decreased incidence of prostate cancer and
increased prostate cancer rates. 35,000 people. The same supplements
greatly reduced secondary prostate cancer
evolution in another study." Kind of hard to wrap your
head around that, right? Both these studies were done
with, I'd say, enough people and came to absolutely
opposite conclusions, showing that there's
definitely other confounding variables at work here. So there's kind of two
solutions to this problem, increase your sample
size to try to get a most representative
set of the population or control for other
confounding variables. And then the question
is, how do you model how much is a
good thing to go over what these models mean. The one that's described
right now in the public is called the linear-no
threshold model. This means that if this
axis right here is bad and this is axis
right here is amount that any amount of
radiation is bad for you. What I think might be a
little bit more accurate is called the linear
threshold model. If you remember from
two classes ago, the ICRP recommends
that, I think, 0.01 microsieverts is
considered nothing officially. That would mean there is
a threshold below which we absolutely don't care. And if there are
any ill effects, they're statistically
inseparable from anything else that would happen. And that would suggest here
this linear threshold model, where this control
line right here would be the incidence of whatever
bad happens in the control population not exposed to
the radiation, the selenium, the whatever. There's also a couple of other
ones like the hormesis model, which says that if
you get no radiation, you get the same amount of ill
effects as the control group. If you get a little
radiation, you actually get less ill effects. In this case, this
would be like saying getting a little bit of
radiation to the lungs could decrease your
incidence of lung cancer. Does anyone believe that idea? Getting a little bit
dose to your lungs could decrease lung cancer? OK. And then you reach some
point of crossover point where, yeah, a lot of
this thing becomes bad. And the question is,
is radiation hormetic? Does this region where
things get better actually lead all the way to x equals
0 as a function of dose? And I want to skip
ahead a little bit to some of the studies. No, I don't want to skip ahead. There are some non
hormetic models that have been proposed
in the literature. It's easy to wrap your head
around a linear model, right? It's just a line. More is worse. But the question is, how much? So folks have proposed
things like linear quadratic, where a little bit
of dose is bad. And then a lot more dose is
more bad as a function of dose. That's actually kind of what
we saw in the Hiroshima data. And I'll show you
again in a sec. So the history of this LNT,
or Linear No-Threshold model, states the following
four things-- radiation exposure is harmful. Well, does anyone disagree
with that statement? I think we all know that
even large-- you know, at least large amounts of
radiation exposure is bad. It's harmful at all
exposure levels. That's the one you
have to wonder. Each increment of exposure
adds the overall risk, saying that it's an always
increasing function. And the rate of
accumulation exposure has no bearing on risk. The first one's easy. We know this is true
because you expose people to a lot of
radiation, bad things tend to happen,
deterministically. The second one, we
already know is false. If you look at large sample
sets of data, like, the data we showed before,
there's definitely a non-linear sort of
relationship going, where each incremental
amount of exposure has the same amount
of incremental risk. We know from a lot of studies
that's not typically true. Then the question is,
what about these two? So now it's going to--
we're going to find and who you some fairly
interesting studies. In this case, leukemia as a
function of radiation dose, what do you guys think
about this data set before I seed any ideas into your heads? So here is dose and
sieverts, not gray. And here is odds ratio, relative
risk of contracting leukemia. If you were to look at
the data points alone, what would you say? AUDIENCE: A little bit
of dose is good for you. MICHAEL SHORT: Yeah,
you might think that. But look at all
the different types of models you can draw
through the error bars. As you could draw
anything going, let's say, down and then up. You could draw a linear
no-threshold model, as long as it got through
this line right here or a linear quadratic model. So a study like
this doesn't quite give you any sort of
measurable conclusion. A study like this might,
especially considering the number of people involved. In this case, this
is the activity of radon in air as related
to the incidence of lung cancer per 10,000 people. Notice the sample size
here, 200,000 people from 1,600 counties that
comprise 90% of the population. Chances are you've then
passed the urban-rural divide. You've then passed any
region of the country. So by including such a
gigantic sample size, you do mostly eliminate
the confounding variables. So, location, you know,
house construction, urban versus rural,
age, anything else are pretty much smeared out
in the enormous sample size. And what do you see here? AUDIENCE: Looks pretty
good for low dose. MICHAEL SHORT: Yeah,
you see a fairly statistically-significant
hormesis effect, where, you know,
the route of exposure is very well-known. Everything else seems to
be controlled for by-- I mean, we've included
something like almost 0.1% of the US population. That's not bad. Other ones for people that get
more specific, targeted dose, in this case, women who
received multiple x-rays to monitor lung collapse during
tuberculosis treatment, a group of people that can
be tightly controlled and followed very well. These are numbers with
one standard deviation. And that, right there so
you can see, is centigray. So this dose right here
is one gray worth of dose. That's a pretty toasty
amount of radiation. But below that,
again, statistically significant-looking data. I don't know how many
people were in the study because I didn't extract
that information. But it's something you might be
doing in the next half an hour. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Oh, it does. It says deaths per 10,000 women. But how many people
were in the study? The question is, what
is your sample size? So like in the last study,
it was just 200,000 people in the samples. That gives you some pretty
good confidence that you've eliminated confounding results. So I don't know how many
folks get tuberculosis these days in the US, or whether
this was even a US study, chances are the sample
size is smaller. So than even if the data
support your idea of hormesis, you have to call
into question, is this a large enough, and
a representative enough, sample size to draw
any real conclusion? So then let's keep going. More data needed. Evidence for a threshold model. This is probably the
most boring-looking graph that actually gives
you some idea of, should there be a threshold
for how much radiation is a bad thing? In this case, it's
very careful data. It's a very carefully-controlled
data set, lung cancer death from radon in miners. And folks that are going
down underground probably have a higher incidence
of lung cancer overall from all
the horrible stuff they're exposed to, whether
it's coal or, you know, if you're mining gypsum. Oh, there's lots of
nasty stuff down there. But there is an
additional amount of deaths responsible
from radon. Here's your relative
list risk level of 1 and up to 10
picocuries per liter, which was around the
maximum of the last study. It's as boring as it
gets, which helps refute the idea of a linear
no-threshold model, because if there was a
linear no-threshold model, this dose versus risk would
be reliably and significantly going up. So there's data out
there to support this. And even-- even better
ones, lung cancer deaths from radon in homes. The study was
careful to look at. If you look at the
legend here, these are different cities ranging
from Shenyang in China, to Winnipeg in Canada, to New
Jersey, which is apparently a city, to places in Finland,
Sweden, and Stockholm, which are somehow
different places. Yeah. So when you see a study like
this where they actually control and check
to make sure they're not getting any
single locality as an unrepresentative
measurement, and the data just looked like a crowd-- a cloud along relative
risk equals 1, this either refutes the idea
that there is no threshold or supports the
idea that there's got to be some threshold lying
beyond 10 picocuries per liter. So, again, to me, it supports
the ICRP's recommendation of chill out. You're going to have a little
bit of radon in your basement. But pretty big studies,
and quite a lot of them, show that a little bit isn't
going to add any risk to you. So if you're worried
about risk, they're statistically is none based on
quite a few of these studies. And in order to enable you to
find these studies on your own, I wanted to go through five
minutes of where to look. And the answer is not Google
because Google is not very good at finding every study. It also picks up a
whole lot of garbage that's not peer reviewed because
it just scrabbles the internet, you know? That's what it does really well. Instead, I want us to
take the next half hour, split into teams for
and against hormesis, and try and find
studies that confirm or refute the idea that
radiation hormesis is an actual effect. So how many of you have
some sort of computer device with you here? Good. Enough so that there is
equal amount in each group. I'd like to switch
now to my own browser. And I want to show you
guys the Web of Science. Web of-- yeah, [INAUDIBLE]
I use Pine on my phone. It's much better science. So if you just Google search Web
of Science, and you're at MIT, it will recognize
your certificates and send you into the actual
best scientific paper indexing thing out there. AUDIENCE: Better
than Google Scholar? MICHAEL SHORT: Oh, my god, it's
better than Google Scholar. Yeah. If you think you've
found everything by looking at Google Scholar,
you're only fooling yourself. You're not fooling anybody else. It's getting better. But it doesn't find anything. And Google Scholar is
really good at finding things that aren't
peer reviewed, self-published stuff,
things on archive, things that you can't
trust because they haven't passed the muster
of the scientific community. So instead, let's
say you would just do a simple search for
radiation hormesis. You can all do this. Don't worry. I'm not showing
you how to search. I'm showing you some
of the other features of Web of Science. And you end up with 534 papers. You can, let's say, sort
by number of times cited, which may or may not be a factor
in how trustworthy the data is. It might just correlate
with the age of the paper. It might also be controversial. So if people cite it as an
example of what to do wrong, it might be highly cited. You know, people have made
tenure cases and like careers on papers that ended
up being wrong. And all you see is 10,000
citations saying this person is an idiot. If the committee val-- you know,
judging you for a promotion doesn't read that
far into it, they're like, oh, my god,
10,000 citations, right? Boom! Tenure, that's all
you have to do. I think I have it
a little tougher. The important part is while
with a title like that, oh, man, the more-- the real
fun part though is you can see who has
cited this paper. So if you want to then go see,
why has this paper been cited 260 times, you can instantly
see all the titles, and years, and number of additional
citations of the papers that have cited it. So this is how you get started
with a real research, research. Yeah, that's what
I meant to say, is starting from a paper and
a tool like Web of Science, you can go forward and
backward in citation time, backward in time to see what
evidence this paper used to make their claims,
forward in time to see what other people thought about it. So who wants to be for hormesis? All right, everyone, all you
guys on one side of the room, all you guys, other guys on
the other side of the room. And I'd like you
guys to try to find the most convincing
studies that you can to prove the other side wrong. I suggest using Web
Science, not Google Scholar. It's pretty easy to figure
out how to learn how to use. And let's see what
conclusion we come to. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Yep,
hormesis by the wall-- yeah, anti-hormesis
by the window. There we go. And I'm going to hide this
because I don't want to give anyone an unfair advantage. AUDIENCE: So [INAUDIBLE]. SARAH: So this is a graph
showing the immune response in the cells of mice showing
that after they were given doses from 0 to 2 gray,
or 0 to 7 on the right, the response of
the immune system. So at the lower doses below like
0.5 gray, which is in the range that we're looking at, well,
the immune system in the mice had a stronger response
at low doses of radiation and then very
quickly tapered off, supporting the claim the
low doses are good for mice. [LAUGHTER] MICHAEL SHORT: [INAUDIBLE] SARAH: I have another graph too. MICHAEL SHORT: So this
percentage change in response, I'm assuming 100
years is no dose. OK. SARAH: Yes. So at higher doses, the
response of the immune system was suppressed, which follows
with what all the other studies show about giving doses in
excess of like 1 gray to cells. MICHAEL SHORT: Cool. So anti-hormesis group. SARAH: Oh, I have
another graph, but-- MICHAEL SHORT: Oh, you do? SARAH: Yeah. MICHAEL SHORT: Oh, I wasn't
going to call them out. I was going to have them
criticize what's up here. SARAH: Oh, no. I have another graph. MICHAEL SHORT: [INAUDIBLE] next. SARAH: I have two
of the same ones. No, I have another
one somewhere. I'll find it in a sec. This one. All right, so this
one is incidences of lung cancer based
on mean radon level and corrected for smoking. So you can't say that it was
just from people smoking. So for radon levels up to
7 picocuries per liter, the incidence of
fatal lung cancer actually decreased as
you had more radon. MICHAEL SHORT: Oh. AUDIENCE: SARAH: Yes. MICHAEL SHORT:
Anything else you guys want to show before we let the
anti-hormesis folks poke at it? SARAH: That's what I got. MICHAEL SHORT: OK. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: What
are your thoughts? AUDIENCE: OK, could you
go back to the last one. SARAH: I will try, yes. AUDIENCE: Do you have
any other [INAUDIBLE].. AUDIENCE: [INAUDIBLE] response. AUDIENCE: So-- so
a mouse is twice-- almost twice as effective
at fending off disease? OK, I-- I am not
a mouse biologist, but the smell test
makes me think that-- that perplexed me. And I guess you didn't
do studies [INAUDIBLE].. SARAH: I am not personally
offended by this. So you're good. AUDIENCE: Enormous--
enormous change. And if radiation hormesis
has such a strong effect on these mice, then why isn't it
something everywhere as a thing now. Like, if radiation-- if
hormesis is responsible for 80% [? movement ?] in mice,
[INAUDIBLE] like where-- SARAH: I don't know
that it was improvement. I think it was just in the
amount of response they saw. I don't know if
that means it's-- well, that doesn't
always mean it was effective at doing something. Right. MICHAEL SHORT: [INAUDIBLE]
you guys have comments too? AUDIENCE: Additionally,
that's like an extremely small of a dose for such
a massive response in like a field that is
so based on probability. Like, how can something
like the dose range that small have that much
of an impact on mice? SARAH: Well, from 0 to half
a gray is pretty significant. AUDIENCE: But [INAUDIBLE] SARAH: [INAUDIBLE] AUDIENCE: --before you
get to the 0.6 gray. AUDIENCE: You're
also only looking at the cells from
[INAUDIBLE] it seems like. And it like looked
varied depending on the kind of tissue. So you can't do it for overall. MICHAEL SHORT:
OK, I want to hear from the pro-hormesis team. What makes your-- what makes
your legs a little shaky trying to stand and hold this up? AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Aha. SARAH: Didn't read the study. [LAUGHTER] MICHAEL SHORT: I like this-- I like this idea
that, yeah, you're only looking at one type
of cell, which may or may respond differently to
different types of radiation. There are no error bars. SARAH: No, not even
a whole mouse either. AUDIENCE: [INAUDIBLE]
in the mouse. MICHAEL SHORT: Oh, oh to
trigger an immune response. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: It's
like-- there are-- there's other cells nearby. But they're like, oh,
you're not my cell. I'm going to [INAUDIBLE]. AUDIENCE: [INAUDIBLE] mice. MICHAEL SHORT: Yeah. So that's-- that's
a valid point. But, yeah, did it say
in the study how many? SARAH: Again, did
not read the study. [LAUGHTER] Read the conclusion. MICHAEL SHORT: The data alone,
just taken it at face value, make it look like hormesis is a
definite thing, Yeah, Kristin? AUDIENCE: I'm saying if
there is [INAUDIBLE].. MICHAEL SHORT: Yeah. SARAH: True. Nine mice cell samples. MICHAEL SHORT: Let's
go to the other study. SARAH: All right,
the-- the lung one? MICHAEL SHORT:
Yeah, it seems to be more controlled and more legit. SARAH: Yeah. This one has error bars. MICHAEL SHORT: Yeah, 1 has error
bars, 2, corrected for smoking. So let's see what
the caption says. Lung cancer fatality rates
compared with mean radon levels in the US. SARAH: And for multiple
counties because it talks about counties plural. So-- MICHAEL SHORT: So
multiple counties helped control for
single localities, or-- AUDIENCE: So the 0 level
there is theoretical. So the data that
you have down here, like, we don't know what
actually happens [INAUDIBLE].. SARAH: Past what? AUDIENCE: Like-- like below
1, the mean radon levels because everyone is
exposed to radon. SARAH: Well, it says average
residential level of 1.7. So I think that means maybe
some people have less, maybe some people have more. I don't know what the
minimum radon level is. MICHAEL SHORT: It's
not going to be 0. SARAH: It's not 0. MICHAEL SHORT:
Yeah, no one gets 0 unless you live in
a vacuum chamber. SARAH: I don't know what
kind of scale that's on. AUDIENCE: Me too. MICHAEL SHORT: Yeah. Cool, yeah. So this-- this is
fairly convincing. If the point here
was to show there is the theory of
linear no threshold, and here's what's an actual
data with error bars shows. It does a pretty
good job in saying, the theory is not
right, in this case. Can you say that in all cases? It's hard to tell. In the first study you found
that was on the cellular level. Maybe the multicellular level-- multicellular level, certainly
not the organism level, like we said, how many mice. This is just parts of mice. Just-- SARAH: It could
be the same mouse. MICHAEL SHORT:
Some cells-- yeah. This one is definitely
at the organism level. It's for-- for gross amounts
of exposure, how many of them resulted in increased
incidence of lung cancer? The answer is pretty much none. They all showed a
statistically-significant decrease, which is
pretty interesting. So thanks a lot. Sarah. And the whole team. Now one of you guys come
up and find [INAUDIBLE].. SARAH: Carrying the team. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: So
who wants to come up? Or does no one [INAUDIBLE]? SARAH: Let's throw down, right? Fixing to scrap. MICHAEL SHORT: OK, you
can just pull it out. SARAH: OK, Are you sure? MICHAEL SHORT: Yeah. SARAH: OK. I don't want to break things. MICHAEL SHORT: No,
pulling it out's fine. If you jam it in, you
can bend the pins. And that's happened here before. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Yeah, if you
want to take a minute to send each other the links, go ahead. No, I like this,
though, is you can-- you can find a graph
that supports something. And you can cite it in a paper. And you can get that
paper published. But looking more
carefully at the data does sometimes call
things into question. AUDIENCE: Just like [INAUDIBLE]. MICHAEL SHORT: Like,
I think you guys found a good example of
that mouse cell study that looks like it
supports hormesis, but you can't say so for sure. Make sure no one's
waiting for their room. No one's kicking us out. AUDIENCE: Have we got a
paper that I found here but we can't open up on there. MICHAEL SHORT: Interesting. Can you send me the link? AUDIENCE: [INAUDIBLE] AUDIENCE: Wait, that
wasn't an option. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Yeah. I mean, we can continue this. There's-- we're not-- since
we're not going to the reactor since that valve was
broken, let's keep it up. AUDIENCE: Hey,
[INAUDIBLE] workbook and [INAUDIBLE] put
it in the log book. AUDIENCE: That's your fault. AUDIENCE: [INAUDIBLE] AUDIENCE: I wasn't
even [INAUDIBLE].. AUDIENCE: [INAUDIBLE] Email us by name. AUDIENCE: [INAUDIBLE] AUDIENCE: It's not over yet. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Yeah,
actually, I like this. This will be a good-- quite a good use of recitation. I'll keep my email
open in case folks want to send things to present. AUDIENCE: That's
the whole title. GUEST SPEAKER: So one--
one of the main problems that we had with
the hormesis effect was that all of the
studies that we've seen seem to cover a large
scope of like tissues, different effects, and
all sorts of things, like, yeah, there's
a lot of studies. There's a lot of trends. But, like, the
things in particular that they're studying
are all over the place. And a lot of the-- a lot of the research
done, like these studies here, are not actually
meant to study hormesis. It's kind of like
recycled data that's used from some other study. And they're kind of like pulling
from multiple sources, which increases the uncertainty. Then, additionally,
we have conflicting epidemiological
evidence of low dosages. So we're, in one instance,
you may see a reduction in breast cancer mortality. You'll see excess thyroid cancer
in children, other, which is-- MICHAEL SHORT: That's the same
study that was just shown, the Cohen 1995
residential radon study. GUEST SPEAKER: Yeah. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: [INAUDIBLE] [LAUGHTER] GUEST SPEAKER: And so I think-- we're not-- I don't think we're
trying to disqualify hormesis as, like, completely wrong. I think one of
the biggest issues that we're taking with it
is that it's a small effect, if anything. It's something that we
really don't know about. It's hard to quantify. And it's, at the end of the day,
really just not worth it, not worth looking into because
of all of the variable-- variables that go into it. And the effects that, like,
we just don't know about. We don't understand it. So, yeah, fire away. MICHAEL SHORT: That's a a
great viewpoint, actually. Yeah, Monica? AUDIENCE: [INAUDIBLE] OK, so it says support for
radiation hormesis [INAUDIBLE] cell in animal studies, OK? And then it cites an example. Can you tell me how
that, like, you know, supports what you're saying? AUDIENCE: Can you just
highlight the part? MICHAEL SHORT: Oh,
right-- right up here. AUDIENCE: OK. GUEST SPEAKER: We haven't
seen it in humans. AUDIENCE: Well, often,
biological studies are done on rats because they
have similar effects to humans. But it's a lifespan of, like,
1/10 a human's lifespan. So, biologically,
that's accepted. GUEST SPEAKER:
Medicine also is not accepted until it works
on humans, not on animals. AUDIENCE: [INAUDIBLE] GUEST SPEAKER: So we can
cure cancer in rats all day. But, like, if it doesn't
work in like the human body, then it just-- we still don't use
it, like, it needs to clear the hurdle
of human usefulness before we actually use it. MICHAEL SHORT: Let's actually
look at this paragraph. They relate to carcinogensis
in different tissues and the dose-response
relationships [INAUDIBLE].. AUDIENCE: So there's
a line that says the evidence for
hormesis in these studies is not compelling since
the data may also be also be reasonably interpreted to
support no radiogenic effect in the low dose range. MICHAEL SHORT: Oh,
that's interesting. Now, how would one interpret--
because you showed the Cohen data. So how would one interpret
that to mean no effect? I'm trying now
determine in this-- are the claims of this paper
that you've been [INAUDIBLE]?? And this brings up,
actually, another point. They do agree that there's been
hundreds of cell and animal studies. They cite three human studies. So since we have
the time, you guys may want to look for more
than three human studies, done at the time of this writing. It's not fair to take ones
that were done afterwards. AUDIENCE: [INAUDIBLE] GUEST SPEAKER: What? Let's find out. AUDIENCE: After 2000. MICHAEL SHORT: It might say at
the bottom of the first page. AUDIENCE: Oh, wait, in
the-- in the [INAUDIBLE].. MICHAEL SHORT: 2000, yep. Yeah. So if you want to
refute that point, you may want to find more
human studies pre 2000. It wouldn't be fair
to do otherwise. But, actually, I
liked what you said. So what you're proposing-- if there's a mostly
blank board, is that most people should
adopt the model that looks something like this. This is the axis of
how much bad or that 0. And this is dose in gray. And whether your model does
this, or this, or this, it sounds to me like
you are defining a-- like you're defining
a kill zone. [INAUDIBLE] maybe the-- GUEST SPEAKER: Yes. MICHAEL SHORT: The point isn't
whether or not hormesis exists. The effect may be so
small that who cares. But the bigger discussion
is how much is that, not is a little bit good. Is that what you're getting at? GUEST SPEAKER: Yeah, the like,
maybe it does look like this. But the dip is small,
really not that different from the linear
threshold model, we noticed. MICHAEL SHORT:
Oh, so in addition to being a basic
science question, could the issue
of hormesis almost be a sidetrack in getting
proper radiation policy through? That's a point I hadn't
heard made before, but I quite like it. Because it's not
like you're going to recommend everyone smokes
three cigarettes a day or, you know,
everyone gets blasted by little bit of radiation once
a year as part of a treatment. I don't think anyone
would buy that. Even if it did help, I don't
think anybody would emotionally buy that. But by focusing on-- you know, that-- there's
a nice expression is the most important thing is
make the most important thing the most important thing. It means don't lose sight
of the overall goal, which is if you're making policy
on how much radiation exposure you're
allowed, do you focus on saying, a little
bit is actually good, or do you focus on saying,
here's the amount that's bad? And anything below
that, we shouldn't be regulating or overregulating
because there's no evidence to say whether it's good or
bad outside the kill zone. I quite like that
point, actually. It means that the
supporters of radiation should chill out as well. Cool, all right, so any other
studies you want to point out? GUEST SPEAKER: We had
a couple of abstracts. MICHAEL SHORT: Yeah, let's see. GUEST SPEAKER: But I don't-- I'm not sure. AUDIENCE: [INAUDIBLE] GUEST SPEAKER: OK. AUDIENCE: Some of the other ones
don't compare hormetic models. But they look at-- they say [INAUDIBLE]. It's like-- GUEST SPEAKER: Do
you want to come up? AUDIENCE: Yeah, this
one says [INAUDIBLE].. GUEST SPEAKER: All right. AUDIENCE: [INAUDIBLE] AUDIENCE: [INAUDIBLE] AUDIENCE: It basically
compares threshold models with no-threshold
models in [INAUDIBLE].. AUDIENCE: [INAUDIBLE] So perhaps hormetic is
still better for you, but they-- the [INAUDIBLE] was
good enough with [INAUDIBLE].. MICHAEL SHORT: So what
they're saying is the-- the choice of model
really doesn't matter, as long as it
fits through the data that we've got. And it seems to be, again, what
happens in the low-dose regime is less important, right? AUDIENCE: And it will--
they were satisfied when it fell from the [INAUDIBLE]. MICHAEL SHORT: So they're saying
the best estimate of this-- interesting. AUDIENCE: They prefer no
threshold [INAUDIBLE].. MICHAEL SHORT: That's funny. "If a risk model with
a threshold is assumed, the best estimate
is below 0 sieverts. But then how is their
confidence interval from-- oh, less than 0 to 0.13. They don't quantify
how much lower it goes because a negative
dose doesn't make sense. No. So, yeah, it's a
strong conclusion. But it looks-- looks
fairly well supported to say that we can't say with
those confidence intervals that they give if there
is or isn't a threshold. Interesting. What do you guys think of this? So what would you
delve into the study to try to agree with
or refute this claim? AUDIENCE: They use a linear
quadratic model only, it looks like. So they're not considering
any of the other proposed models, which is a little-- maybe not sketchy,
but it just seems like it'd be very easy
to consider other models and why didn't they do that. MICHAEL SHORT: Sure. You know, what no study
has gotten into yet is, what's the mechanism of, let's
say, ill effect acceleration. This is something that, at
least at the grad school level, we try to hammer to
everyone constantly is not just what's the data,
but what's the mechanism. What's the reason for an
acceleration of ill effects? So if you guys had to think with
increasing radiation exposure, let's say we wanted this linear
quadratic model idea, what could be some reasons or
mechanisms for an increased amount of risk per unit dose
as the dose gets higher? Yeah? AUDIENCE: Well, your
body [INAUDIBLE].. But then-- so at some--
you get more dose-- you get more dosing [INAUDIBLE]. It just keep fixing itself. And once you get
past a certain point, then it can't [? fix itself ?]
[? fast enough. ?] The additional damage keeps
snowballing events. And they're giving
it more damage to curb more radiation
because you would run out of-- of various [INAUDIBLE]. MICHAEL SHORT: Sure. Works for me. Yeah, I like that--
the idea there was that you've got
some capacity to deal with damage from radiation. And then once you exceed that
capacity, you don't also-- with a higher dose,
you don't also ramp up your capacity
to deal with that dose. So in the linear
region, let's say, you're somewhat absorbing the
additional ill effects of dose by capacity to repair
DNA or repair cells. Then once you exceed
that threshold, you're beyond that point. So that could be a
plausible mechanism for why there could be a linear
quadratic model that could be tested, certainly with single
cell or multi cell studies, like these-- these radiation
microbeams or, you know, injecting something
that would be absorbed by one cell
[INAUDIBLE] irradiated, and seeing what
the ones nearby do. So you could count
that as number of mutations, number
of cell deaths, anything, something that could
be quantitatively tested. So that's pretty cool. I actually quite
like this study. It's awfully hard
to poke a hole in-- in the logic used here. The claims aren't outrageous. They're saying, this is
what the data is saying. If you change the model, you
can or not have a threshold and still get an acceptable fit. Can we actually look
in the study itself? One thing I want to
know is, what sort of-- do they do meta
analysis, or did they-- yeah, so this was on the
Japanese atomic bomb survivors. So did they analyze
previous data, or did they get their own. And then if so, what
was the sample size? Somewhere it'll be,
like, yeah, [INAUDIBLE].. So where [INAUDIBLE]. GUEST SPEAKER: Where am I-- where should I be
looking for this-- MICHAEL SHORT:
Probably further down in any sort of
methodology section-- materials and
methods, here we go. OK, here it is, 86,500
something survivors. Oh, yes, with lots of follow up. AUDIENCE: But how are you
able to determine the dose? Like-- MICHAEL SHORT: That
is a good question. AUDIENCE: Because
especially for-- if we're looking like low
dose, and you're estimating, it's very easy to, like,
estimate wrong, or, like, because then-- then it calls
into question you have-- [INAUDIBLE] modeling
they're using. MICHAEL SHORT: Mhm. So that's a great
question is, how do they know what
those people die? So how would we go about
trying to trace that? This is when you
dig back in time. They reference this,
the data appears et al, whatever, whatever. So if you can go
to Web of Science, pull up this Pierce
et al Web paper. Look at cited references. Yeah, right there. And look for that
1996 Pierce study. Let's see if it has it. You can just like control F
for Pierce, and we'll find it. Pierce and [INAUDIBLE]. Yeah, 1996, that's the one. GUEST SPEAKER: Where? Which one? This one? MICHAEL SHORT: [INAUDIBLE]. This is the 1996 one. Yep. So let's see if we
can trace this back and find out how they estimated
the dose of these folks. GUEST SPEAKER: So I
just go to full text? MICHAEL SHORT: Yeah. AUDIENCE: How [INAUDIBLE]. MICHAEL SHORT: OK. So interesting, this LLS cohort. So there was some
life span study, which was also referred to
actually in the lecture notes as one of the original
studies, says, who met certain conditions
concerning adequate follow up. Although estimates of the-- OK, I want to see the next page. Although we
estimate-- that might be what we're looking for. Number of survivors, let's see. AUDIENCE: It's 92%. MICHAEL SHORT: OK, here we
go, materials and methods. The portion of the
LSS cohort used here includes the same
number of survivors for whom dose estimates
are currently available, et cetera, with estimated doses
greater than 5 millisieverts is [INAUDIBLE]. Table 1 summarizes the
exposure distribution. So let's go find table 1 and
see where the data came from. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: So it turns out
that this is specifically-- DS-86 weighted colon
dose in sieverts. Interesting. AUDIENCE: It [INAUDIBLE]. So how did they get that? MICHAEL SHORT: I don't know. But it sounds like we need to
find this LSS, this Just LSS. So let's look at the things
that this paper cites. Find this LSS. So I'm walking-- what I'm doing
here is walking you through how to do your own research. And if someone comes to you
with some internet emotional argument of, this and that
about radiation is wrong, instead of yelling
back louder, which means you lost the
argument, you hit the books. And this is how you
do the research. AUDIENCE: LSS-85, does that
mean it was [INAUDIBLE].. MICHAEL SHORT: Probably. Version of-- title
not available. I hope it's not that one. Can you search for LSS? Nothing? So let's go back to
the paper and find what citation that was. If you go up a little bit, I
think there was like a sup-- a superscript up to the
last page, I'm sorry. There was a superscript
on LSS stuff. AUDIENCE: So general
documentation of the selection of LSS
cohorts [INAUDIBLE].. MICHAEL SHORT: Thank you. All right, let's find
references 9 and 10 in the-- yeah, [INAUDIBLE]. AUDIENCE: Can you click
one of the References tab? MICHAEL SHORT: Oh, yeah,
up there, References. Awesome! 9 and 10, OK. Let's find them. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: So
let me show you quickly how to
use Web of Science to get what you're looking
for if I could jump on? GUEST SPEAKER:
[INAUDIBLE] up here? MICHAEL SHORT: You
don't have to, yeah. But thank you for being up here
for so long and running this. So we're looking for-- where was-- the
article was here. Went into references. I guess that was like the last-- I don't want to
close all your tabs. Here we go. So GW, is that
Beebe and Usagawa. So we'll go to Web of
Science, look for authors, any paper with those authors. So you can do a more
advanced search. This is where things get really
interesting and specific. So ditch the topic. Search by Beebe and
add a field, Usagawa. And then anything
with these two folks in the author field that is
indexed by Web of Science will pop up. Nothing. Did I spell anything wrong? Usagawa, of course. That's unfortunate. Last thing to try
is Add Wild Cards. Interesting. This is actually one place
where I would use Google to find a specific report. So because you're not looking
to survey a field that's out there, but you're
looking for any document that you can confirm
is that document. Let's head there. Oh, it looks like
Stanford's got it. That's something
that references it. So at this point,
we've hit the maximum that we can do on the computer. But if you finally
want to trace back to see how were the
Hiroshima data acquired, take these citations,
bring it to one of the MIT librarians like Christ Sherratt
is our nuclear librarian. AUDIENCE: He's a
nuclear librarian? MICHAEL SHORT: And we have
a nuclear librarian, yeah. MIT libraries is pretty awesome. So when you're looking
for anything here in terms of research
or whatever, there's actually
someone whose job it is to help you find
nuclear documents. And chances are, this
is a pretty big one. So I wouldn't be
surprised if we have a physical or electronic copy. So we're now like one
degree of separation away from finding the
original Hiroshima data, where we can find out how
did they estimate that dose. So I think this is fairly-- hopefully, this is
fairly instructive to show you how do you go about
getting the facts to prove or disprove something, knowing
the-- not just the physics that you know, but how to
go out and find that stuff. Now, I did see a
bunch of sources from the pro hormesis team. You still want me to show them? AUDIENCE: [INAUDIBLE] MICHAEL SHORT: OK. Thanks. All right, you just want to
hold this up while your-- let's go to your sources. OK, here we go. AUDIENCE: All right. MICHAEL SHORT: So walk us
through what you found. GUEST SPEAKER: I just
need to open them up. AUDIENCE: Go through
them all, or-- MICHAEL SHORT: Yeah,
let's do them all. GUEST SPEAKER:
There's not too much. Kind of-- OK, so,
I unfortunately was not able to find like too
many pretty graphs, or data, or anything of the sort. But if you look up, what
did I search for this? I think I just looked
up radiation hormesis. And this is one of the
articles that turned up. And it seems to be
pretty well cited. You can see it's
been cited 184 times. And kind of the quick look
through the citations, from what I saw, seemed
to be in support of it. And if you actually look at the
abstract itself, where is it? AUDIENCE: [INAUDIBLE] GUEST SPEAKER: Yeah,
well-- the last sentence is pretty excellent. "This is consistent with
data both from animal studies and human epidemiological
observations on low-dose induced cancer. The linear
no-threshold hypothesis should be abandoned and
should-- and be replaced by a hypothesis that is
scientifically justified and causes less
unreasonable fear and unnecessary expenditure." MICHAEL SHORT: You know what? I want to see what are the human
epidemiological observations that they cite. GUEST SPEAKER: Yeah, so
unfortunately, the MIT libraries does not
have an electronic copy of this article. And I wasn't able to find one. But going through some
of the citations for it-- MICHAEL SHORT: Before
you do, could you go back to the article? GUEST SPEAKER: Sure. MICHAEL SHORT: I want
to point something out. GUEST SPEAKER: Yes. MICHAEL SHORT: Can you tell
if this was peer reviewed? GUEST SPEAKER: I do not
know how to do that. MICHAEL SHORT: It appears
to be a conference. GUEST SPEAKER: OK. MICHAEL SHORT: Not all
conferences require peer review in order to present the papers. So while conference
proceedings will typically be published as a record of
what happened at the conference, we don't know if this one
was peer reviewed and checked for facts by an
independent party. Could you go up a little
bit, and maybe there'll be some information on that? Oh, it did go in the British
Journal of Radiology. OK, that's a good sign. So conference proceedings,
you don't know. But in order to publish
something in a journal, you do because then in
order to get in the journal, things have to be peer reviewed
to meet the journal standards, regardless of whether they
came from a conference or just a regular submission. So, OK, that's good to see. So, now, what else you got? GUEST SPEAKER: And then
one of the key sentences that I found right here,
adaptive protection causes DNA damage prevention,
and repair, and immune system or immune stimulation. It develops with
a delay of hours, may last for days to
months, decreases steadily at doses above about
100 milligray to 200 milligray and is
not observed anymore after acute exposures of more
than about 500 milligray. That's all pretty interesting. Like I said, unfortunately, I
couldn't find the actual paper. So you can't really delve
into some of those claims. But I tried to look at
some of the citations that delved into them. And this is where my
presentation gets a little bit shakier because I'm
not particularly good at parsing some of this
complex stuff very quickly. MICHAEL SHORT: Let's
do it together. GUEST SPEAKER: All right. [INAUDIBLE] MICHAEL SHORT: If you could
click Download Full Text in PDF, it'll just be bigger. GUEST SPEAKER: OK. MICHAEL SHORT: There we go. GUEST SPEAKER: So
it seemed to me this one was more looking
through the statistics of various studies. I'm not entirely sure. But I think the conclusion-- [INAUDIBLE] There we go. So the very last paragraph,
"the present practice assumes linearity in assessing
risk from even the lowest dose exposure of complex tissue
to ionizing radiation. By applying this type
of risk assessment to radiation protection
of exposed workers and the public alike, society
may gain a questionable benefit at unavoidably substantial cost. Research on the p
values given above may eventually
reveal the true risk, which appears to be inaccessible
by epidemiological studies alone. MICHAEL SHORT: So
what are they going on claiming [INAUDIBLE] versus
not being willing to claim it? GUEST SPEAKER: So it
seems like they're saying that at the current,
there's not really a problem-- a statistically
valid assertion of the linear no-threshold
model and that the benefits to society gained from that are
not worth the cost to society from that assumption. MICHAEL SHORT: So
what sort of costs do you think society
incurs by adapting a linear no-threshold
dose risk model? GUEST SPEAKER: I mean, it could
pose unnecessary regulations on like nuclear
power, which could be arguably better for society. MICHAEL SHORT: Sure. Nuclear power plants
emit radiation, fact, to use the old cell
phone methodology. There's always going to be some
very small amount of tritium released. The question is, does it matter? And if legislation is made
to say absolutely no tritium release is allowed,
well, you're not going be allowed to
run a nuclear plant. That's not the question
we should be asking. The question we should be
asking is, how much is harmful? So I think that's what this
study is really getting at is I'm glad to see someone
say, you may have a benefit. But the cost is not
worth the benefit. Like I-- I had a multiple
of the same arguments with different people when
they were complaining, well, how dare would you expose me
to any amount of radiation at any risk that
I can't control. I used to protest
outside Draper Labs for 30 years protesting
nuclear power. I was like, OK, how
did you get there? They were like, oh, I drove. What? In a car? Do you even know the risks per
mile of getting on the road, let alone in Cambridge
specifically? No? Well, I was like, you
should really consider where you put your effort? It's-- again, it's
emotions versus numbers. I'm going to go with
numbers because I tend to make bad decisions
when I follow my emotions, as do most people because
most decisions are more complex than fight
or flight nowadays. Yeah? AUDIENCE: So a lot
of the discussion just seems to be around
like expanding [INAUDIBLE].. But a lot of the
arguments don't seem to like really [INAUDIBLE]. But, yeah, like there's
a certain extent, like, oh, you will
see [INAUDIBLE].. MICHAEL SHORT: Yeah. AUDIENCE: [INAUDIBLE]
are doing the same. MICHAEL SHORT: You
make a great point. That's why I like your--
your chosen idea so much is, well, you didn't say chosen. That's what I-- yeah. Yeah, the question we
should be asking ourself is not what is the dose-risk
relationship, but when should we actually care. It's like both sets of
studies have kind of come to the conclusion
that, nah, right? AUDIENCE: [INAUDIBLE] dose
doesn't really matter. GUEST SPEAKER: Yeah, and
then I found this last one is a little bit more assertive. It's kind of just
hitting the same nail on kind of the elimination of
the linear no-threshold model. But then it does go on to make
some more powerful claim right here. "These data are examined
within the context of low-dose radiation
induction of cellular signaling that may stimulate
cellular protection systems over hours to
weeks against accumulation of DNA damage." MICHAEL SHORT: Was
this the paper cited in the other one that
actually said hours two weeks? GUEST SPEAKER: I
believe so, yeah. MICHAEL SHORT: OK, cool. GUEST SPEAKER: And
then we can actually-- MICHAEL SHORT:
[INAUDIBLE] this one? GUEST SPEAKER: Yes. We can look up the full
text on Google Scholar. MICHAEL SHORT: That's OK. When you know what you're
looking for, you can verify it. That's-- that's a useful thing
for Google is like to find known content. But if you're trying to
survey a field in Google, no. GUEST SPEAKER: That's
not what I wanted. MICHAEL SHORT: Not yet. I'm sure-- I'm sure
they're working on it. But they're not
Web of Science yet. GUEST SPEAKER: All right. AUDIENCE: [INAUDIBLE] GUEST SPEAKER: Does anybody see
a Get The Full Paper button? Oh, wait, right here, right? MICHAEL SHORT: Yep. That's it. GUEST SPEAKER: OK. Sign in? MICHAEL SHORT: Sounds like
we don't subscribe to this. GUEST SPEAKER: Oh, I was
able to get to it somehow. Well, yeah. AUDIENCE: I have another article
supporting this claim, though. MICHAEL SHORT: OK. GUEST SPEAKER: But this one-- AUDIENCE: Submit it, or
bring yours up, or whatever. GUEST SPEAKER: And
then this one-- this one just had
some nice data. If I'm going to
summarize, it had-- it was looking at the amount
of DNA damage instances compared normal background dose
to like very, very low dose. And the very, very low
dose was significantly less than the normal background dose. So that just kind
of shows that like very low levels of radiation are
like no worse for you than just background dose,
which is interesting. MICHAEL SHORT: Cool. GUEST SPEAKER: Yeah. MICHAEL SHORT: I also
want to make sure, do you guys have more
articles you want to show? AUDIENCE: [INAUDIBLE] MICHAEL SHORT: If you
want to send it to me, I'll put it up here. GUEST SPEAKER: All right, I
minimized because I didn't just want to leave your email. MICHAEL SHORT: Oh, I don't care. There's nothing-- GUEST SPEAKER: OK. MICHAEL SHORT: I'll
bring it back up. So that's all the ones you sent? Cool. Actually, this one-- this
debate is turning out a whole lot more interesting
than previously because, well, because you're thinking. It's actually really
nice to see this. And this is the-- AUDIENCE: [INAUDIBLE] MICHAEL SHORT:
I'm not surprised. Don't worry. It's just pleasant to have
a debate about something controversial with a
whole group of people who are thinking and
researching rather than shouting and like throwing plates. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Oh, no, if
you want throw a chair, but I might throw one back. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: I wonder if
anyone's gone out recently and has come up with all of the
pro and anti hormesis studies and actually written
a paper that says, that's not the point,
because, really, what we're getting-- huh? AUDIENCE: You could write that. MICHAEL SHORT: No, I think you
could write that paper now. AUDIENCE: Well, oh. MICHAEL SHORT: It would make for
a pretty cool undergrad thesis, actually. Yeah? Maybe I can tell
you a little bit about what an undergrad
thesis actually entails because the
seniors are all asking. But it's good for you
to know ahead of time. So the main requirement
for an undergrad thesis is it's got to be your work. That doesn't mean
you have to have collected the data yourself,
like done an experiment. But it has to be some
original thought, or idea, or accumulation of yours. So trying to settle this debate
and trying to figure out what would be a proposed
chill region to say, forget the linear
threshold or no threshold. That's for the basic scientists. If you are a government and
want to legislate something that actually captures should
people be afraid or not, defining that region would
be a pretty cool study to do in the meta-analysis of
lots of other studies, tracing back how worthy-- I mean, a lot of people
refer to the Hiroshima data set because that's about
the biggest one we have. In addition to folks with
radon or folks that smoke, they were all exposed
to the same thing in the relatively same area. So it's a good control
group of people. But how was-- how were
those doses estimated? You have to dig that up. And the act of digging
that up and then recasting all of these new studies
in the basis of everything we've learned since would
make for a pretty cool undergrad thesis topic. So as undergrad chair, I
wouldn't say no to that. Threshold and other departures
from linear quadratic curvature in the same data
set appears to-- is it the LSS data set? Let's try to get the full text. Awesome! I think it's looking good. Great! Now I've seen that name before. Interesting. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Interesting. They propose
another model called a power of dose, a power law. And they say,
depending on this-- there's little
evidence that it's statistically different
from one which is a what do they call
one linear threshold quadratic threshold or linear
quadratic threshold, OK? So, again, it seems to be
yet another paper saying, I don't think it matters. Statistics says
it doesn't matter. You could fit any
model to this data. Let's get to the methods. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Interesting. So dose response for
all non-cancer mortality in the atomic bomb survivors. So, also, in this case,
it's mortalities not caused by cancer. AUDIENCE: Like, caused
by radiation disease? Or is that caused
by [INAUDIBLE]?? MICHAEL SHORT: So
this would be-- I think what they're getting
at is is there a response, or is there a change in
the amount of mortality not due to cancer and the-- the-- AUDIENCE: Health benefits other
than decreasing risk of cancer. MICHAEL SHORT: Or in this
case, health detriments, right? Because in this-- you know,
it never goes negative. You can't really
tell in some cases. Let's see. Yeah, quite hard to tell,
especially considering. And so at the low doses,
what would you guys say for the low dose data? AUDIENCE: That doesn't matter. MICHAEL SHORT: I see a pretty
well-defined chill zone right there, right? AUDIENCE: Chill zone? MICHAEL SHORT: We're
definitely still in the chill zone at 0.4
sieverts of colon dose. And that's a pretty
hefty amount of dose. You know, we're talking eight
or nine times the allowed amount that you're able to get in a
year from occupational safety limits. Once the doses
get higher, things seem to get a little more
deterministic or statistically significant. But, yeah, look at all
the different models. The linear threshold,
quadratic threshold, linear quadratic
threshold, power of dose all goes straight through not
just like in the error bars, but almost straight through
most of the data points, except for the
really far away ones. So this is a pretty
neat study, showing, like, hey, the
relationship does not appear to matter for
doses of consequence. I would call 2 sieverts
a dose of consequence based on our earlier discussion
of biological effects. Luckily, it doesn't go
much farther than that. You don't want a lot
of people to have received doses beyond 10 gray. But this is pretty
compelling to me to say, like, we can argue
about what the real model is and what the underlying
mechanism is, but is this a question we really
should be asking ourselves when the total risk-- let's say, when the
total risk to an organism reaches about 100%, once you
reach a a dose where it doesn't even matter, then
is this a question that we should really be
debating in the public sphere? I love the outcome of
this particular debate. Lots of statistics,
don't have time to parse. Is there anything else,
Chris, that you wanted to highlight in this study? AUDIENCE: This appears
to [INAUDIBLE] comments on Professor Donald
Pierce on [INAUDIBLE].. MICHAEL SHORT: Oh, OK, well-- AUDIENCE: Do you think it
could be the same Pierce? MICHAEL SHORT: Maybe. It was a UK Pierce, I think. That's pretty cool. So anyone else have
any other papers they want to show for or
against or for our sort of collective new conclusion? Which is that we
should just relax. Cool. Well, that went-- yeah? Charlie? AUDIENCE: I just had
had a question, like, what would be like a
posed use of radiation hormesis [INAUDIBLE]? [INAUDIBLE] MICHAEL SHORT: So
let's say you could prove beyond a shadow of a doubt
that a little bit of radiation exposure was a good thing. You might then prescribe
radiation treatments in order to reap the benefits. I don't think there's
been a single study that shows that there's like
deterministic benefits from irradiating people. Some of the studies
show that folks that have gotten exposed
via various routes do show a lower
incidence of cancer. So you could almost think
of it like a vitamin, not an injectable vitamin. But-- so back-- there are
lots of pictures online and stories of way up
in the north in Russia and northern countries
that expose you to ultraviolet
radiation to stimulate the production of vitamin
D in your skin cells because in the absence of an
ingestible source of vitamin D, you make it naturally, but not
when there's eternal darkness. So they'd actually have kids
stand in front of a UV lamp, which does have ill effects. That can cause
also skin cancers, but the benefits of the
organism in generating vitamin D that you need
for health are greater. So that might be an example. These-- these sorts of ideas
are not that far fetched. If you put little kids
in front of UV lamps, which you know
can do bad things, but also does more good
things, then who's to say it shouldn't happen for radiation? Well, no one's to
say yet because we have no real conclusive
proof that it is helpful. But that was the-- yeah? AUDIENCE: Have there been any
mechanisms that [INAUDIBLE]?? MICHAEL SHORT: You
mean in-- for radiation or for something else? AUDIENCE: For radiation. MICHAEL SHORT: The mechanisms
of-- so that one study that Chris showed that-- what was the idea? That-- [INAUDIBLE]. The first one that you
showed, the mouse one, and then the one
that Chris mentioned where a little bit
of radiation dose stimulated the immune system. That might be a
potential good thing, where the damage or
death of a few cells may stimulate the nearby ones
to ramp up an immune response, thus snuffing out any other
infection or problem that's coming up. That could be a use. But we have to be proved
with much more confidence than anything I've seen today. So that's a good question. Yeah, like how would you use it? Use it like a vitamin, like
a UV lamp, like a SAD lamp. Although, I don't
think SAD lamps do anything bad, the
Seasonal Affective Disorder, the most unfortunate
acronym in the world. Yeah. AUDIENCE: [INAUDIBLE] MICHAEL SHORT: Yes. I don't know if that
would be easy to swallow. Yeah. Cool. All right, any other
thoughts from this exercise? I think I'll do more
interactive classes like this. It's good to hear you
guys talk for a change. Cool. OK.