MAY 11 1971

DEPARTMENT OF HEALTH, EDUCATION, AND WELFARE
PUBLIC HEALTH SERVICE
NATIONAL INSTITUTES OF HEALTH
BETHESDA, MARYLAND 20014

 

May 6, 1971

Professor Joshua Lederberg
Professor of Genetics
Stanford University School
of Medicine

Stanford, California

Dear Professor Lederberg:

Your newspaper piece on the hazards of (additional) atomic radiation
(Washington Post, May 2, 1971) was very interesting appearing as

it did three days after a seminar at the annual meeting of the
American Physical Society on the Biologic Effects of Atomic Radiation.
John Gofman spoke there, as did Leonard Hamilton of the Brookhaven
Laboratory.

I'd like to talk about some issues that your piece brings to mind
(which represent my thoughts, and those of a few colleagues - and
not necessarily those of the National Cancer Institute). First,

I have found many people, scientific and lay, upset over computations
of risk based on "average" exposure. One of my colleagues points
out the importance of the distribution of doses among the population
through an extreme example: say we want to minimize the risk to

the total population for a given average exposure in the population.
If one individual received the total radiation necessary to give
this defined average, then the risk to the total population would

be a minimum. This one individual would (very likely) die, and no
one else would be at any risk at all.

Any meaning that working with average exposures would have is

also dependent on the linear, no-threshold model of radiation dose-
response. For example, if the dose-response curve (for the effects
of added radiation looked like this: (I use cancer as the response,
as an example, and not intending to slight your concerns about
genetic damage)

 

RESPONSE:
additional
% of persons
dying of
cancer
as
AVG PERM. MAX PERMISSIBLE
EXPOSURE (APE) EXPOSURE (MPE)

DOSE
Page 2 - Prof. Joshua Lederberg

then the people at dose MPE would be at considerably more than
proportionally higher risk (in proportion to their dose) than the
people at dose APE,

There is another way of under-estimating risk, even if the "true"
dose-response curve were linear, no-threshold. Say there really
is a linear, no-threshold dose-response in the number of somatic
mutations produced; e.g.,

Q)

RESPONSE:
no. of
somatic
mutations

DOSE

Even if all these somatic mutations were to lead to cancer, since
a person can die of cancer only once and since some persons would
have more than one somatic mutation, and some would die of other
causes before the cancer killed we would observe a non-linear
dose-response for cancer deaths; e.g.,

(1)
RESPONSE:
no. of
individuals Q—
dying of
cancer
DOSE

The data in curve (2) are the kind of data that is normally observed
and usually only at high doses. If one now takes responses (cancer
deaths) at some high doses (curve (2) ) and extrapolates back to

the o, o point with a straight line one gets curve (3); e.g.,

RESPONSE:

  

-7 Estimated

  
     

 

 

>» a
GP .
abe aed arches OBSERVED
PERMISSIBLE PERMISSIBLE

EXPOSURE EXPOSURE
Page 3 - Prof. Joshua Lederberg

Now estimating risk from average exposure (using curve 3) would
generally underestimate total risk to the population and the
greatest underestimates are at the lowest doses. I think the
most recent publication from the ABCC on Hiroshima-Nagasaki
(25-69, Ishimaru, et al) possibly the uranium miner data
(Lundin, et al - not yet published) and some of Miriam Finkel's
recent animal data are in line with a dose-response curve like

(2).

You are certainly correct in saying that the battle over the legitimacy
of the 170 millirad standard* may be the wrong battle. Hamilton
suggests that no more than 5 millirem of this 170 need be set aside
for atomic power production. If the Seaborg predictions, or
the Hamilton allowances are correct then obviously one does not
need to have levels anywhere nearly as high as 170 millirems
(for any practical purposes). But no matter what the average
levels are, or are set at, the greatest risk is to the persons
with greatest exposure and these are the people who need to be
protected. If the risks of added radiation are not linear - and
conceptually I think we could fit this idea in with our usual
conceptions of sigmoid dqse-response curves this way ~

I

RESPONSE
"Natural" level
of Response

  

 

‘0 Background Added
Radiation 1 Man-made Radiation
DOSE

|
I
I
I
{
I
{
|
I
t
|
|

and for the reasons I cited above for being concerned about the
distribution of radiation, I think we have to concern ourselves

more about the high exposures. I think most people would accept

the low costs inherent in Hamilton's maximum of 5 millirems, if

they also believed that very great precautions (and always improving
ones) were being taken to prevent an atomic Takoma Narrows Bridge
accident. Most people go along with Karl Morgan's suggestion that
atomic power plants be located outside densely populated areas.

I think I may also be saying here that because of fears of accidents

 

My preliminary computations have Gofman too high by a factor

of 10 even accepting all his assumptions. One of my associates
is now going through detailed actuarial computations to check
me (and Gofman) out.
Page 4 - Prof. Joshua Lederberg

the arguments based on average exposure are emotionally unsatisfying
as well as possibly scientifically wrong.

Perhaps this is all academic. We have excluded medical radiation.
Why? As the AMA has recently said rather strongly in an editorial,
physicians are using diagnostic radiation far too promiscuously.

This radiation increases the total exposure very much in a very
small number of persons - often not for the patient's direct benefit.
Perhaps our discussions of reducing radiation should concentrate
more on reducing unnecessary diagnostic radiation. Is this another
reason for saying that the 170 millirem battle is the wrong battle?

Very truly yours,

Lp A

JAA € Ch ein,
Marvin A. Schneiderman, Ph.D.
Associate Scientific Director
for Demography (Acting)
National Cancer Institute