Studies of Radium-Exposed Humans:
The Fallacy Underlying a Major "Foundation of NCRP, ICRP, and AEC

Guidelines for Radiation Exposure to the Population-at-Large"

by

John W. Gofman and Arthur R. Tamplin

Division of Medical Physics (Berkeley)
and
Bio-Medical Research Division
Lawrence Radiation Laboratory (Livermore)
University of California

oN
win

Supplement to Lee

Testimony presented before
The Sub-Committee on Air and Water Pollution
Committee on Public Works
United States Senate
Qlst Congress
November 18, 1969
Studies of Radium-Exposed Iumans:
The Fallacy Underlying a Major "Foundation of NCRP, ICRP, and AB

i

Guidelines for Radiation Exposure to the Population-at-La>re

John W. Gofman and Arthur R. Tamplin

Introduction

The guidelines which specify the maximum limits of exposure of
humans to ionizing radiation from peaceful uses of atomic energy represent
a set of numbers having as great an impact upon the future of the human
race as any set of numbers ever could. Therefore, society must demand , as
an item of the very highest priority, that such guidelines be absolutely
above reproach and question, for the consequences of error can even mean
the deterioration of the human race on earth.

Recently we have attacked the Federal Radiation Council Guidelines
for such exposure on the grounds that if everyone received the Guideline
dosage, some 16,000 additional cases of cancer plus leukemia would occur
each year in the United States (1)(2),

It is the purpose of this communication to demonstrate that one’

of the purported major foundations of guidelines established by the ICRP,

 

the NCRP, and the FRC ig totally without basis in fact and rests upon tne
overtly erroneous interpretation of some otherwise extensive careful observa-
tions on humans. We refer to the belief that a threshold (practical or
absolute) was demonstrated through the studies of radium dial painters,
chemists exposed to radium, and persons receiving radium or related alpna
emitters medically.

The chief proponent cf the belief that the data accumulated through
the study of such individuals leads to a valid "threshold" below wnich no
injury occurs is Professor Robley D. Evans of the Massachusetts Institute
of Technology. Dr. Evans is to be commended for a beautiful series of
investigations extending over 30 years which have greatly increased our
knowledge concerning radium and its effects wpon man, However, we shall.
develop the evidence here to prove that Dr. Evans' conclusions from his
own and from other data are totally erroneous with respect to demonstrating »—
or even suggesting a "safe threshold" of lonizing radiation.

We can best start this evaluation by a series of quotations of
Professor Evans, quotations of such deep consequence as to possibly affect

the future of every living human and those unborn.

 

rene:

ci nme

LTE PO RIN REN NR Ne ORE

Fe RA on yatmane  ne Pp

he ee
Quotation 1 (Reference 3)

"The effects of skeletally deposited radium and mesoti:orium arc
of immediate relevance here. These studies have provided the permissible
body burden for radium in humans. It is the only NCRP, ICRP, Atomic Bnerrsy
Commission permissible dose based directly upon observations on humans, and
is the pivot or reference point for the permissible burdens of plutonium
and of strontium-90." |

Quotation 2 (Reference 1)

"Tt is my conviction that there does exist an absolute threshold
and a practical threshold for inhaled radon daughters, below which these
nuclides are innocuous."

Quotation 3 (Reference 5) ,

"Thus it will be seen that the present RPG of 0O.1uC Ra contains
a large safety factor and would appear to be a satisfactory value even if
applied to large populations."

Quotation 4 (Reference 6)

"In the present series of hearings this committee has been exposed
to the conservative, oversimplified, incorrect, linear and non-threshold

model of radiation carcinogenesis."

These represent four quotations of great assurance and of far-
reaching implications. We shall now, through analysis of the data upon
which Professor Evans bases these conclusions, demonstrate that the con-
clusions implied in these quotations are not correct, and are in no way

supported by the evidence upon which they rest.

The Experimental Observations

 

This analysis will address itself to the data concerning the oc-
currence of cancer (carcinomas plus sarcomas) in persons carrying various
measured residual body burdens of radium. Evans has presented the data
for one series of cases (269 persons in all) with the occurrence of cancer
in individuals in relationship to the residual radium burden (5). of
Hasterlik has presented an entirely separate series (264 women, some 26
years after occupational exposure to radium) with the occurrence of cancer |
in individuals in relation to residual radium burden (7), These data are

reproduced in Table 1 (Evans data) and Table 2 (Hasterlik data). As

Evans correctly pointed out, there is remarkably good agreement between
>
-3-

the two sets of data (8), However, we must add there is remarkably good
further agreement in the fact that neither set of data supports the con-

clusions drawn by Evans.

Table 1 (Reference 5)

 

Data for 269 cases where a pure radium equivalent (residual burden in uC Ra)

was estimated (Dial Painters, Chemists, plus medically treated persons

 

uC Ra equivalent residual

 

 

No. of Cases Dose Range Median Dose Numbex of Cancers~~
te) <0. 001 <0.001 0
61 0.001l- 0.01 0.0055 oO
80 0.01 - O.1 0.055 a 0
32 O.1 - 1.0 0.55 3
LO 1.0 - 10.0 5.5 14
14 10,0 -100.0 55 2

Table 2 (Reference 7)

 

Data for 264 women (~ 36 years after occupational exposure)

 

pC Ra equivalent residual

 

No. of Cases Dose Range Median Dose Number of Cancers
23 <0,001 . ~ <0.001 0
36 0,001-0.01 0.0055 Oo
102 | 0.01 -0.1 0,055 Oo
62 O.1 -1.0 0.55 3
41 >1.0 (1-10) 5.5 14 .

t

Analysis of Both Sets of Data

 

The hypotheses that have been set forth by Evans, exemplified in
the quotations above, are: .

(1) These data indicate that there exists a threshold value
below which radium deposition in the skeleton does not produce cancer in
humans.

(2) These data indicate that the linear model of radiation car-
cinogenesis is incorrect. -

Let us approach both of these hypotheses, since they are closely re-

lated. At first glance, it is to be noted, in these extremely small series

SD
=.

of humans, that none of the observed cases of cancer occurred in any of

the dosage ranges below 0.1 pC Ra residual burden in either series of cases.
We can admit even further that in the Evans series (Table 1), the lowest
dosage where a cancer occurred is 0.6 yC, and in the Hasterlik series, the

lowest dosage with cancer is 0.45 yC. But such a first glance observation

 

 

does not even remotely resemble an analysis and does not bear at all upon
the validity of the Evans hypotheses listed above. We must, therefore,

proceed with an analysis.

(a) Analysis of the Evans data (Table 1)

 

The first step is to determine the probability of finding cancer
in these subjects in relationship to dose of residual Ra burden. This can
be done either using only the group of cases (1.0-10.0 WC Ra) with the
largest number of cancers, since it is most reliable, or by using all the
data for groups where cancers occurred (0.1-1.0, 1.0-10.0, 10.0-100.0 uC Ra).
We shali do the analysis both ways, for the sake of completeness.

For the group of cases with burdens of 1.0-10.0 uC Ra there were

14 cases of cancer out of 40 total persons.

i is, therefore, the probability of cancer for a median dose of 5.5 uC Ra.

So, per pC Ra, oe = 0.064 is the probability of cancer.

Expressed alternatively, 6.4 cases per 100 people are found for a burden of
L yc Ra.

Now, we can look at the three low dose ranges where no cancers
were observed. The linear thesis would expect, for such low dosages, 6.4
cases per 100 persons per pC Ra residual burden.

The 0.01-0.1 uC Ra range

 

We have 80 persons in this group with a median residual burden of
0.055 uC Ra.
For 80 persons, therefore, our expectation is:

oY x (6.4) x (0.055) = 0.28 cases of cancer expected.

Cancer in humans cannot occur as fractional cases. Therefore, in our group

 

of 80 persons, occurrence can be O cases, 1 case, 2 cases, ete. If our ex-
pected number of cases is 0.28, then there are at least 72 chances’ out of
100 of observing O cases. So the probabilities are strongly in favor of

observing O cases, which happened.
-5-

Conclusion: The data are completely consistent with the lincar
thesis and completely consistent with the absence of any threshold "safe"
dose in this range.

The data provide nothing at all to indicate we

should accept either of Dr. Evans hypotheses.

The 0.001-0.01 uC Ra range

 

We have 61 persons in this group with a median residual burden of
0.0055 uC Ra.

For 61 persons, our expectation is:

.
es (6.4) x (0.0055) = 0.021 cases of cancer expected.

With this expectation, there are at least 98 chances out of 100 that 0 cases
would be observed. So the probabilities are extremely strong in favor of
observing O cases, which happened.

Conclusion: The data are completely consistent with the linear
thesis and completely consistent with the absence of any threshold "safe"
dose in this range of Ra burdens, also. -

The data afford no support whatever to either of

Dr. Evans hypotheses.

The <O.001 uC Ra range

 

We have 42 persons in this group with a residual burden of <0.001. To
favor Dr. Evans, let us use 0.001 as the median burden.

For 42 persons, therefore, our expectation is:

(Ge.) x (6.4) x (0.001) = 0.0027 cases of cancer expected.

With this expectation, there are at least 997 chances out of 1000 that Oo
cases would be ‘observed. So the probabilities are enormously in favor of
observing O cases, which happened.

Conclusion: The data are completely consistent with the linear
thesis and completely consistent with the absence of any threshold "safe"
dose in this range of Ra burdens, also.

No support is obtained for either of Evans hypotn-
eses.

Summarizing, we can state, for all dosages below 0.1 uC Ra,
there is not a shred of scientific evidence that should lead anyone to
accept either of Dr. Evans hypotheses. If evidence favoring his hypoth-

eses exists, it certainly must be elsewhere than the data he has provided
XN
from persons with residual Ra burdens. The lincar thesis and the absence
of any "safe" threshold emerge totally unscathed from this analysis. They
are not proved by this analysis, but there is no suggestion whatever that

they are incorrect, in contrast to Dr. Evans claim. (see Quotation 4, avove)—

(b) Analysis of the Hasterlik data (Table 2)
' The procedure of analysis of these data is identical with that
provided above. For the group of cases with burdens of 1.0-10.0 uC Ra

there were 1) cases of cancer out of 41 total persons.

14

Tt is, therefore, the probability of cancer for a median dose of
5.5 wC Ra residual burden. .

So, per uC Ra, — = 0.062 is the probability of cancer.

This means 6.2 cases of cancer per 100 people are found for a residual

burden of 1.0 uC Ra. This is spectacularly good agreement with the value

6.4 found for the Evans cases. .
We can go through each individual group now as previously, and

the results of such analysis are presented in Table 3.

Table 3

Analysis of Expectation vs. Observation in the Hasterlik Series of Cases

 

(These are the groups where O cancers were observed)

Expected No. Probability of observing
No, of Cases Dose Range Median Dose of Cancers 0 Cancers in this series

 

uU :
23 <0,001 6001  . ~—-0, 0014 998 out of 1000
36 0.001-0.01 0.0055 0.012 99 out of 100
102 0.01 -0.1 0.055 0.35 65 out of 100

" Clearly, from these analyses, we can state the data are compietély con-

sistent with the linear thesis and completely consistent with the absence

of any "safe" threshold range of Ra burden. | |
These analyses provide nothing at all to indicate we should

accept either of Dr. Evans hypotheses.

(c) Analysis Based wpon Use of All Cancer Cases to estimate the

 

probability of cancer per uC Ra

 

In order to explore every possible way of analyzing the data to
see if any support can be developed for Evans hypotheses, we thought it

worthwhile to estimate the cancer probability by using all groups where
-f-

cancer did occur. Using both the Hasterlik data and the Evans data, we

have the combined totals shown in Table 4.

NN
aa

Table

Combined Data for Estimation of Cancer Probability Associated with Residual

=~

Ra Burden (Hasterlik + Evans data)

 

No. of Cases Dose Range Median Dose No. of Cancers Observed
94 O.1- 1.0 0.55 6
81 1.0- 10.0 5.5 rs)
14 10.0-100.0 55.0

To estimate the probability of cancer per uC residual Ra burden, utilizing

all cases, we need first the average burden for the overall group of persons.

(94) (0.55) + (81)(5.5) + (14) (55-0) _ 1267.2 6 7

Average Burden = ok + 61 + 14 189

Therefore, probability of cancer per uC Ra burden is:

6+ 28+3 _ 37
(189)(6.7) ~ (189)(6.7)

But this is much lower than the 0.064 we used above. Therefore, if we

= 0.029

used 0.029 as the probability of cancer per uC Ra, the analysis would lead

to the conclusion that it is even far less likely that any support for

ae
~

Evans hypothesis exists within these data.

Lastly, we may exclude the people with the very high Ra residneal
burdens (10-100 pC Ra) on the grounds that a very high prior death rate
may have left an unrepresentative group. -

In this case, we exclude 14 subjects with burdens of 10 uC or

more, and we calculate:

(94)(0.55) + (81)(5.5) _ 497.2 _
“oh ot = “75 = 2.6 yc

The probability of cancer per uC Ra residual is:

Average Burden =

6 + 28 3
(75)(2.8) = Tipsy(e.8) = 2-069

But this number is so close to the 0,064 already utilized, that no material

 

support for the Evans hypotheses will derive from its use instead of 0.064.

nee
8-

(d) Analysis of the Evans Series and the Hasterlik Series Combined

 

As Evans has correctly stated, the data from his serics are in
remarkably good agreement with the data of Hasterlik. In the hope that
possibly, having a larger series through combining both sets of data, it
might be possible to give a fairer trial to the Evans hypotheses, we
have calculated the expectations using all cases from both series. As
the probability of cancer per wC residual Ra burden, the mean of the
values derived from Evans data and from the Hasterlik data, namely,
0.063 per uC Ra residual burden is used. The "combined" analysis is

presented in Table 5.

Table 5

Analysis of Expectation vs. Observation in The Combined Series of Cases

 

(Hasterlik + Evans).
(These are the groups where O cancers were observed)

Expected No. Probability of observing
No. of Cases Dose Range Median Dose of Cancers O Cancers in this series

 

use
65 <0. 001 0,001 0.0041 996 out of 1000
97 0.001-0.01 0.0055 0.034 966 out of 1000
182, 0.01 -O.1 0.055 _ 0.63 37 out of 100

For the dosage ranges up through 0.01 pC Ra residual burden, the answer
is abundantly clear -- no support whatever for either of the Evans hypoth-
eses. Even for the higher dose range 0,01-0.1 uC residual Ra burden, the
results fall far short of acceptable support for the Evans hypotneses.
If we use the minimum statistical criterion of p=0.05, the analysis snows
a probability 7 times too high compared with what it would take to make
us accept the Evans hypotheses. On matters of such grave importance,
one certainly should insist on using p=0.01, and in this case the proba-
bility is 37 times too high compared with what it would take to argue
for acceptance of the Evans hypotheses.

Again, even using the combined series, the data are consistent
with the linear thesis and are consistent with the absence of any "safe"

threshold of residual Ra burden.
Discussion
It is now important to return to the four quotations of Evans
presented in the introduction and to show, in turn, the error in cach one.
Quotation 1 (see above) claims, "these studies have provided
the permissible body burden for radium in humans". The analyses presented _
above show that "these studies" provide nothing in the way of support

for a'

"safe" threshold body burden with respect to cancer induction. If

it is true that NCRP, ICRP, and AEC have, as Evans suggests, used these

studies to decide permissible burdens of radium, plutonium, and strontium-90,

they would be well advised to cease and desist from any further such use.
Quotation 2 (see above) claims. it is Evans “conviction tioc

an absolute or practical threshold exists, below which radon daughters

are innocuous", A "conviction" is, of course, a strange phenomenon. It

can be based upon scientific evidence, upon intuition, upon hunch, upon

religious belief, or upon hope. We would be the first to defend staunchly —

Professor Evans' right to hold convictions based upon intuition, hunch,

religious persuasion, or hope. Our analysis does not address itself to

these areas. We can state that his conviction cannot rest upon scientific

iw
o>

evidence, for our analysis shows that no such evidence exists.

Quotation 3 (see above) claims that "the RPG of O.1 uC Ra con-
tains a large safety factor and would appear satisfactory even if applied
to large populations". This contention rests in part upon the fact that
Professor Evans' studies are of residual radium burdens, and the sug-
gestive evidence that the initial burden was probably 20 times higher,
Thus, he suggests that if 0.1 uC Ra residual burden is "safe", then 2.0 uC
Ra initial burden would be safe. So, he calculates that 0.1 uC Ra
initial burden is "conservative". But 0.1 uC Ra initial burden corres-
ponds to 0.005 uC Ra residual burden. In the analyses above we have
demonstrated that Evans data offer no support that 0.005 uC Ra residual ae
burden is below any kind of threshold. Therefore, there is no evidence

at all to support his contention that 0.1 uC Ra initial burden is at all

 

safe, to say nothing of being conservative.

; This being the case, his assertion that such a value would bed
satisfactory even if applied to large populations could lead, if accepted
by responsible authorities, to a public health disaster unparalleled in

the history of mankind.
-1O-

Quotation 4 (see above) claims, "the linear, non-threshold model
of radiation carcinogenesis is conservative, oversimplified, and incorrect",
But our analysis shows that Evans data and his analyses do not

(a) even remotely suggest the linear, non-threshold model to be
conservative,

(b) even remotely suggest the linear, non-threshold model to be
oversimplified,

(c) even remotely suggest the linear, non-threshold model to be

incorrect.

It is conceivable that the linear, non-threshold model of radia-

tion carcinogenesis may be conservative, oversimplified, and incorrect.
If so, this remains for future science to demonstrate. -Evans' work simply
does not bear upon this issue. It can be stated that the linear, non- ~
thresnold model does make excellent sense in setting Public Health
Standards for radiation exposure. ‘

It would be irresponsibility of the highest order, repugnant'
to any competent bio-medical scientist, to set Public Health Standards
based upon a hope, unfounded in evidence, that somehow a poison will

turn out to be less toxic than conservative sound estimates would indicate,
-1l1-

References

(1)

(3)

(4)
(5)
(6)
(7)

(8)

Gofman, J. W. and Tamplin, A. R. "Low Dose Radiation, Chromosomes,

and Cancer", Presented at "1969 Institute of Electrical and Electronic
Engineers Symposium: Nuclear Science", Oct. 29, 1969, San Francisco,
California (To be published in IEEE Proceedings, February, 1970).
Gofman, J. W. and Tamplin, A. R. "Federal Radiation Council Guidelines
for Radiation Exposure of The Population-at-Large--Protection or
Disaster?", Testimony presented before The SubCommittee on Air and
Water Pollution Committee on Public Works, United States Senate, Slst
Congress, November 18, 1969.

Evans, R. D. in "Radiation Exposure of Uranium Miners", Hearings of
the Joint Committee on Atomic Energy, 90th Congress, Part 1, May 9 thru
August 10, 1967, p. 265.

ibid p. 274.

Evans, R. D. "The Effect of Skeletally Deposited Alpha Emitters in
Man", Brit. J. Radiol. 39, 881-895, 1966.

Reference 3, p. 274.

Hasterlik, R. in "Radiation Standards, Including Fallout", Hearings

of the Joint Committee on Atomic Energy, 87th Congress, Part 1,

June 4-7, 1962, p. 325-331.

Evans, R. D. "The Radium Standard for Boneseekers - Evaluation of the
Data of Radium Patients and Dial Painters", Health Physics, 13, 267-278,
1967.