At present only PDF and raw text is avaiable for this file at:
                         http://ratical.org/radiation/CNR/ESoCbIR1971.pdf
                         http://ratical.org/radiation/CNR/ESoCbIR1971.txt

Proceedings of the Sixth 
. Berkeley Symposium on 


I 


Mathematical Statistics 


and Probability 

HELD AT THE STATISTICAL LABORATORY, 
UNIVERSITY OF CALIFORNIA 
JUNE AND JULY 1970 
APRIL, JUNE AND JULY 1971 


UNIVERSITY OF CALIFORNIA PRESS 

.. 

EPIDEMIOLOGIC STUDIES OF 
CARCINOGENESIS 

BYIONIZING RADIATION 

John W. Gofman and 

\, 

Arthur R. Tamplin 
July 20, 1971 



CONTENTS 

1. 
Do We Really Need Human Epidemiologic 
Data for Pollutants? ......... p 235 
1.1 
Carcinogenesis and Leukemogenesis in 
Humans Exposed to Ionizing Radiation ¥. 236 
"Three generalizations" ..
. 
237 

1.2 
Dose-response Relationship: Ionizing 
Radiation-Induction of Cancer ahd 
Leukemia. . . . . . . ¥ . . . . ¥ ¥ . 240 
1.2.1 
Time of Onset of Carcinogenic 
Response and Its Duration ...¥ ~ ¥. 241 
1.2.2 
Dose-response Relationships over a 
Range of Doses ¥¥¥.¥¥.¥..¥ 244 
1.2.3 
Variation in Carcinogenic DoseResponse 
Relationship with Age at 
Radiation Exposure . . ¥ . .á ¥. ~ . ¥ 249 
2. 
The Carcinogenic Consequences of 
Population Exposure to Environmental 
Ionizing Radiation ¥¥¥..¥.¥¥ 252 
2.1 
Cancer Hazard for Average Population 
Exposure at 0.17 rads per Year, Total 
Body Irradiation ¥...¥¥¥..¥¥ 254 
3. 
Life-shortening by Radiation-induced 
Cancer .¥¥¥¥. . . . . . . . . . . 259 
4. 
Are There Possible Mitigating Factors 
Which Could Reduce the Estimated 
Hazard of Population Exposure? ¥..¥ 262 
4.1 Thresholds: Absolute and "Practical" 
262 
4.2 Protraction of Radiation .¥.¥¥
¥ 
262 
4.3 Repair of Radiation Injury ¥.¥.
. 
264 
5. 
A Re-look at the Purpose of This 
Symposium after Consideration of the 
Potential Population Consequences of 
Low-dose Radiation Exposure ¥.¥¥.. 264 
Sununary . . . . . . . . . . . ¥ . . . . . 
. 
265 
266

References 
Discussion ¥....¥...¥¥.¥¥.¥ 269 

Figure 1: Dose-response vs. time curve . p 242 

Figure 2: Three generalized dose-
response patterns ¥ . . . . ¥ . . . ¥ 246 

Figure 3: Linear dose-response relationship 
in the plateau region ¥. . ... 248 

Table I: Increase in Childhood Cancer and i 
leukemia from in utero radiation ... 238 

Table II: Change in rate of induced 
malignant disease with duration of time 
since exposure ¥¥¥..¥¥¥.. 

238 

Table III: Ratio of spontaneous cancer á 
mortality rates to leukemia mortality 

rates ¥.¥¥..
¥ 
. . . . . . . . . 239 ¡ 

Table IV: . Variation in cancer induction 
per rad with age ¥¥¥¥¥¥.¥..¥ 

252 

Table V: Case 1, Radiation-induced cancer 
mortality by age and sex from 170 mR 
per year, with plateau constant after 
latency period ¥..¥¥.¥¥¥¥¥. 256 

Table V.I: Case 2, same as Table V except 
plateau lasting 30 years beyond latency 257 

Table VII: Case 3, same as Table V except 
10-year , latency period instead of 15, 
and plateau lasting 20 years beyond 
latency .¥...¥ ¥ ¥¥¥¥¥¥¥.¥ 259 

Table VIII: Loss of life-expectancy from 
radiation-induced cancer ¥¥¥¥..¥ 261 

I 
I 
t 

I 

l 

I 

á 1 

(, 

á



EPIDEMIOLOGIC STUDIES OF 
CARCINOGENESIS BY IONIZING 
RADIATION 


' I 

' 

JOHN W. GOFMAN and ARTHUR R . TAMPLIN 
LAwBENClil LIVERMORE LABORATORY and 
UNIVERSITY OF CALIFORNIA, BERKELEY 


I

I 

I 

¥ 
1. Do we really need human epidemiologic data for pollutants? á1 
\ 
In general, we should like to express our lack of sympathy for the expressed á 
purpose of this Symposium, which is the planning of epidemiological studies for 
the evaluation of effects of major pollutants on humans . Carcinogenesis and 
leukemogenesis are two particularly worrisome long ~nn effects which deserve 
consideration with respect to any pollutant . From our experience with ionizing 
radiation as a pollutant we have derived some iessons that we believe are 
extremely important to understand if society is to avoid paying a very high, 
probably unacceptable, price for the introduction of environmental pollutants. 
One such lesson centers around the prevalent notion that human epidemiological 
evidence concerning carcinogeneeis should be required be/oretechnological 

, 
promoters are willing to admit the serious potential hazards of a pollutant. J 
Ionisdng radiation is a classic example of this fallacious notion. 

I

In our opinion it is neitherappropriate nor good public health practice to 
demand human epidemiologic evidence to evaluate carcinogenic or leukemogenio / ~ 
hazard of a pollutant. First, in a civilized society, there should never exist 11n j 
ideal set of human epidemiologic data . What epidemiologic data do become j 
available are always subject to serious reservations with respect to equivalence 1 
of controls and exposed groups upon variables other than the specific pollutant : 
variable under study. The net result is that controversy persists interminably. 
Peculiarly, but not unexpectedly in the face of promotional bias, the presumption 
is all too commonly made that, where uncertainty exists about the magnitude 
of effect; it is appropriate to continue the exposure of humane to the potential 
pollutant. It would indeed be sad if this Symposium helped contribute to this 
pernicious philosophy, which can only be described as that characteristic of a 
society bent upon ecocide in the name of ostensible technological progress. a 

In the case of radiation as a pollutant, we may consider some of the major 
á epidemiological samples that have become available for study and relate the 
reservations that have been raised concerning acceptance of the results derived 


This work was supported in part by the U.S. Atomic Energy Commission. 

235 

236 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

from the study of these samples. Approximately 100,000 survivors of Hiroshima 

and Nagasaki atomic bombing have been under followup study with respect to 

cancer and leukemia. Dosimetry reconstruction is difficult, at best, considering 

the nature of the event during which the radiation exposure occurred. Further, 

the associated possible injurious factors other than radiation were expected, in 

general, to be highly correlated with radiation exposure. Another large sample 

available for epidemiological study is the series of some 11,000 cases of anky


losing spondylitis in Great Britain, treated with X-irradiation. No satisfactory 

control series of spondylitics, untreated by X-rays, but otherwise equivalent, ls 

available. Hence, questions can properly be raised about using the populatio ~1 

at large as a reference sample. And the use of drugs for pain relief in addition to 

radiation therapy leads to the question of effects due to the drugs alone or to 

synergistic effects between drugs and radiation (see Collen and Friedman ls 

contribution to this Symposium) 
. 
á ! 

It can be pointed out that a vast experience with experimental animals of 

several species has proved cause and effect relationship between radiatioh 

exposure and carcinogenesis and leukemogenesis. Therefore, the real significan~e 

of the human studies is to ascertain comparability of dose response relationships 

for humans versus other species, rather than establishment of whether the ob


served association of radiation and cancer in these human population samples is 

causal. 
j 

We believe the appropriate approach to the study of leukemogenic or carcinogenic 
potential of w._llutants is the study of dose response relationships in several 
mammalian species.á And until or unless scaling laws are established among 
species, including humans, it should be aasumed,for public health purposes, that 
the human is at lÇut as sensitive as the moal sensitive experimental species 
studied . In the ionizing radiation case, abundant experimental animal data have 

.. I'

accumulated over the past quarter century demonstrating that radiation can 
provQke cancers of essentially all organs, provided the radiation is delivered to :; 

() '' 

susceptible ce!ls. Moreover, reasonable dose re~nse data were available through \ 
such studies (Gofman and Tamplin [11]). Had these experiment.al animal data 
been utilized properly, the recent surprise concerning the higher than anticipated 
cancer hazard of ionizing radiation need not have occurred. 

t

Having expressed our serious disapproval of the concept that human epidemio{} 
logical studies should represent an approach to the study of pollutant effects, we 
should like to review here the treachery inherent in such studies, how they led 
to an earlier underestimate of the carcinogenic effect of radiation, and the 
residual uncertainties which still exist in assessment of the magnitude of the 
carcinogenic response to ionizing radiation in humans. 

1.1. Carcinogenesis and leukemogenesis in humans exposed toionizingradiation. \ 
J 

Direct evidence that virtually all forms of human cancer can be induced by 

' 

!

ionizing radiation has accumulated over several decades, often, however, with 
poor assessment of dose response relationships . By now, acute and chronic 
myelogenous leukemia, other acute leukemias, multiple myeloma, bone sarcoma, 


SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN

CARCINOGENESIS BY IONIZING RADIATION 237 238 

skin carcinoma, lung cancer (bronchiogenic and other varieties), thyroid cancer, 

breast cancer, stomach cancer, pancreas cancer, malignant Jymphoma, colon 

cancer, cerebral tumors, neuroblastoma, Wilms tumor, maxillary and other 

sinus carcinomas, and pharynx cancer have all been shown to be inducible in 

humans by ionizing radiation. (Gofman and Tamplin, [7]). One disease (pre


sumed malignant), chronic lymphatic leukemia, does not, thus far, appear to be 

radiation-induced (Lewis [21]). The implications of this finding remain unclear . 

For those remaining varieties of human cancer, other than the ones just listed, 

no evidence indicates they are not radiation-inducible . Within the evidence 

available, fortunately limited, there are simply no adequate data concerning 

radiation-induction. 

Recently we presented three generalizations concerning inducti,on of human 

cancer and leukemia by ionizing radiation. (Gofman and Tamplin, [7], _[9]). 

These generalizations follow: 

GENERALIZATION1. All forms of cancer, in a.IIprobability, can be increased 

by ionizing radiation, and the correct way to describe. the phenomenon is either 

in terms of the dose required to double the spontaneous mortality rate for each 

cancer or, alternatively, of the increase in mortality ra~e of such cancers per rad 

of exposure. 

GENERALIZATION All forms of cancer show closely similar doubling doses 

2. 
and closely similar percentage increases in cancer mortality 
rate per rad. 
GENERALIZATION Youthful subjects require leBB radiation to increase the 

3. 
mortality rate by a specified fraction than do adults. 
Others (Stewart and Knee.le, [31]) had clearly stated the outlines of these 

generalizations based upon the irradiation of infants in utero. Court-Brown and 
Doll [3] had done so based upon irradiation of adults. Additional study (Gofman, 
Gofman, Tamplin, Kovich, [13]) provides no reason to suggest a change in any 
. of these generalizations; rather, it provides supplementary support for the 


generalizations. 

The second of these generalizations led us to predict that for every leukemia 
induced by ionizing radiation, the 8Um of the number of cancers induced would 
stand to leukemia aa doeathe BUmof 8pontanwus cancer mortalities to leukemia 
mortality. Since the sum of spontaneous cancer mortalities is some twenty times 
that of leukemia mortality (Table III) over a fair share of the human adult life 
span,á we predicted the sum of cancer mortalities per unit of radiation would be 
twentyfold that of leukemia. This caused a furor in the "radiation .community," 
since the International Commission on Radiological Protection (IGRP) [17] 
had predicted in 1966 only one cancer mortality per leukemia mortality from 
radiation (exclusive of thyroid carcinoma which shows a low mortality rate in 
the cases which do occur). The error in the ICRP estimate represents a classic 
illustration of the pitfalls in the epidemiologic approach that had been used . 
Leukemia happens to occur earlier, post-irradiation, than do other cancers. Thus, 
since the ICRP was studying population samples in the relatively early yea.rs 
post-irradiation, the cancer mortality was seriously underestimated . 

Data. a.re available for adults from the study of the irradiated ankylosi11g 
spondylitis cases in Great Britain [3]. These subjects were irradiated primarily 
in early adulthood and then followed for periods up to 27 years. This study 
provides a good basis for testing the prediction that the sum of cancer mortalities 
is some 20 times that of leukemia mortality following irradiation. It is obvious 
that such a comparison test requires that radiation dosages be equivalent for all 
eites compared, or that appropriate dosage corrections be made before comparison 
of cancer 1nortalities with leukemia. mortality. The Court-Brown and 
Doll data are presented in Table II, including partial followup through 27 

years. 

TABLE I 

INCREASE IN CHILDHOOD CANCER AND LllUKEIUA J'ROM In Utero RADIA'l10N 
Radiation delivered in the form of X-rays during diagnostic pelvimetry. 
&!timated dose <2 rads. 

Type of cancer 
Stewart-Kneale data (1008) 

Leukemia 
Lymphosarcoma 
Cerebral tumors 
NeuroblBBtoma 
Wilms' tumor 
Other cancers 

MacMahon data 

Leukemia 

(1002) 

Central nervous system tumors 
Other cancers 

! 
I 

Radiation induced increase 

I 

50% increll86 over spontaneous mortality rate 

lá;

50% 
'á~,

50% 
. ,,. 

50% 
60% 
50% I 
I 

50% 
60% 
40% 


TABLE II . 

CHANOII IN R.ATII 101' INDUCED MALIGNANT D1BEABB WITH 0URA'l10N 
OJ' TIMII 8JNCII Exl>08URII IN IRRADIATED ANKYLOBING 8p0NDYLITICS 
q,'rom data in Table VI of Court-Brown and Doll, 1005.) 

~ 

.

Caeee per 10,000 man-years at risk 

Leukemia and 
Years after irradi ation 

aplBBtic anemia 

2.5 
3-5 
0-2 

6.0 

5.2 
9-11 
6-8 

3.6 
12-14 
4.0 
15-27 
0.4 

Total of expected cBBes in 10,000 persons 
in Z1 years calculated from the rates given 67 

Cancers at heavily 
irradiated sites I,.. 
3.0 
0.7 
3.6 
13.0 
17.0 
20.0 
300 


CARCINOGENESIS BY IONIZING RADIATION 239 

TABLE III 

RATIO or SPONTANEOUS CANCER MORTALITY RATES 
TO LBu:ii:&111A MoRTALJTT RATBB 
(Derived from U.S. Vital 8141i,liu for 1060.
) 


Males Age group Raf (ESpontaneoua cancer mortality rates) 
(years) IO, I: Leukemia mortality rates 

40-44 16.9 á, 
45-49 22.9 
50-54 28.5 
55-50 28.7 
60--64 29.2 
65-60 29.1 
70-74 23.5 

In these studies 40 per cent of the total bone marrow (the expected site of 
origin of the leukemias) ie estimated to have received irradiation. Tlie spondy-. 
litis treatment ie directed to the spine, not to other bone ejtee containing marrow. 
The mean bone marrow dose ie 880 rads (for spinal marrow)-.... 

The "heavily irradiated" sites in those studies represent the sites receiving 
spray irradiation incident to the planned 8l)inal irradiation. Dolphin and Eve 
(4] estimated that these "heavily irradiated" sites received approximately 
seven per cent of the mean spinal marrow dose. á 

From Table II, the observed (E Cancer Mortalities/Leukemia Mortality) ""' 
(369/67). 
The E Cancer Mortalities must be multiplied by (100/7), to correct dosage 
for "heavily irradiated" sites to be equivalent to that for the spinal marrow. 
Th e Leukemia Mortality must be multiplied by 2.6 to correct for the fact that 
only approximately 40 per cent of the total bone marrow received irradiation. 

Therefore, for true total body irradiation the Corrected Ratio for radiationinduced 
malignant diseases, (E Cancer Mortalities/Leukemia Mo~ality) -= 
(369/67)(14/2.5) ~31. 

Since the spondylitis patients were irradiated in early adulthood, the period 
of fo11owup is approximately in the 40 to 70 year age region. From U.S. Vital 
Stati8tic8, 1966, we can derive the ratio, (E Spontaneous Cancer Mortality 
Rates/I: Leukemia Mortality Rate) for this age range. These values are presented 
in Table III. 


In the spondylitie patients, the sites designated ae "heavily irradiated" 
include lung, stomach, colon, pharynx, esophagus, pancreas, lymphatic tissue. 
The major contributing sources to cancer mortality are, therefore, included. 
Possibly the ratio (E Radiation-Induced Cancer Mortalities/I: Leukemia 


áMortality), determined here to be approximately 31 might be increased some if 
remaining tissue sites had been irradiated. The ratio CESpontaneous Cancer 
MQrtality Rates/I: Leukemia Mortality Rate) ie in the neighborhood of 20 to 

240 SIX1'B BERKEI,EY SYMPOSIUM : GOFMAN AND TAMPLIN 

30, for the relevant age range. Within the errors of such data as those for the 
spondylitis casea, the similarity of ratios for the spontaneous and the radiationinduced 
cases can be taken as strong support for Generalization 2 presented 
above, and as grosslyat variance with the earlier ICRP prediction. 

Dy now, however, this whole controversy has all but subsided. An ICRP 
Task Force (1969) has presented the Court-Brown and Doll data, together with 
the dose correction shown above (application of the Dolphin-Eve correction). 
Hamilton (15] stated that his own estimate of the ratio, (E Radiation -Induced 
Cancer Morta1ities/Radiation-Induced Leukemias), is within a factor of five of 
that of the authors, but he failed to take into account the dosage corrections 
which are, of course, absolutely essential in the treatment of the ankylosing 
spondylitis data . When the Hamilton estimate ie appropriately corrected for the 
dose difference between bone marrow and the "heavily irradiated" sites (where 
cancers arise), hie revised estimate would be entir~lyin accord with our .own 
estimate. Mole (24) has recently published an estimate that the sum of radiation


,
~ 
induced cancer mortalities is "an order of magnitude" greater than radiationinduced 
leukemias. In a personal communication in 1970, Mole indicated to us 
that he had notapplied the full Dolphin-Eve dosage correction, and this almost 
certainly explains the residual factor of two differences between his estimat~s and 

our own. 
Thus, the so-called "radiation controversy," at least with respect to the ratio 
(E Cancer Morta.lit_ies/E Leukemia Mortality) for total body radiation, is 
essentially over. The á tl<>ntroversy did pinpoint a valuable epidemiological 
pitfall, namely, the serious underestimate of cancer hazard from ionizing radiation 
resulting from the use, by standard setting bodies, of epidemiologic data for 
!. 

a time interval beforethe serious carcinogenic effects had developed. And the 

I/ 

long observation periods required should alert ue to the futility of hopes of learning
á of carcinogenic effects of new pollutants through human epidemiologic 

á: studies on a time scale that can be practicaUy .useful. 
j ¥ 1.2. Dose ruponse relational,ipa: ionizing radiation-induction of cancer and 
' leukemia. The ultimate objective, for a poll~tant such as ionizing radiation, is 
an estimate of the human cost in premature death through cancer and leukemia, 
resulting from fairly chronic low or moderate dose irradiation. It is self evident 
that dose response relationships are required for such an estimate. Less immediately 
evident are some of the more subtle characteristics of the dose response 
relationships, characteristics which are crucially determinative of the magnitude 
of expected human cost. 

One such characteristic is the time of onset of the carcinogenic response 
following exposure. Closely related is the duration of the response period in an 
exposed population. A second characteristic ie the nature of the dose response 
curve over a wide range of doses. Thie becomes especially important because 
much of the available epidemiologic data covers a dose region higher than that 
anticipated for population exposure. Dose rate ie an ancillary feature deserving 

¥á 
\ 

f 

! 

., 
i 

' 

'i

, I 

. I 

,á'

I 

I ,, .. 

L

t' 

r 
i 

i .. 

r 

:,, 


242 SIXTH BERKELEY SYMPOSIUM! GOFMAN AND TAMPLIN

CARCINOOENEBIS BY IONIZING RADIATION 241 
consideration . A third characteristic is the variation in dose response relationship 

~ 

as a function of ageat expo8Ure. 

áa 1.2.1. Time of onset of carcinogenic resvonse and its duration. A valid parameter 
commonly employed to assess carcinogenic response to ionizing radiation is 
the radiation-induced.age specific mortality rate from any particular malignancy 

áI 

or group of malignancies. It would be ideal if this parameter were readily 
available both from the experimental animal and human data, but this is not 
always the case. Following radiation exposure (of humans and experimental 
animals) there is a. period of time which elapses before any provably induced 
mortality from cancer or leukemia. is observed. In short lived mammals, like the 
laboratory rat, this period is on the order of magnitude of months; in the human, 
of years . Most workers have referred to this apparently silent period as a latent 
period. It is not at all certain that such a latent period is truly as long as has 
generally been suspected. What is more likely is that the dose response curve 
shows at first a gentle slope upward with time, fo11owed by a more steep slope, 
and then followed by what may be called a '!plateau" region (Figure la). In 
studies involving relatively few subjects, the low incidence in the gentle slope 
region can appear to be a period free of effects, and this _may well be why the 
impression has arisen of a long latent period. In most of the-da,ta available for 
analysis, the quantitative features of this segment of the response versus time 
curve are poorly defined. 

Of additional great importance would be knowledge concerning durati<mof 
the "plateau" region of the response versus time relationship. Unfortunately, 
the available data simply do not allow, for any particular malignancy, satisfactory 
construction of this curve to ascertain how long the "plateau" region 
persists . For chronic myelogenous leukemia [36) the data suggest that once the 
excess mortality rat e from radiation is perceived, it persists year after year for 
some 10 to 15 years, whereupon the excess mortality rate drops toward a lower 
value. In that same study the radiation -induced excess acute leukemia mortality 
rate showed no significant decline from the peak (or "plateau") value even after 
20 years beyond irradiation. In the study on patients with spondylitis treated by 
X-rays [3], the 15 to 27 year period post-irradiation showed a higher excess 
mortality rate than any á earlier periods of observation. There is no evidence 
within that study, of a return toward spontaneous mortality rates from malignant 
disease for the irradiated subjects. ; ' 

Both the Japanese studies and the spondylitis studies should, in the next ten 
years, provide very valuable clues concerning th e duration of th e plateau region 
of response. For the present, however, no valid data are avail able to determine 
plateau duration . Indeed, and regrettably, the data for experimental animals, 
with respect to this issue, are no better than the sparse human data. As will be 
noted in the subsequent discussion of estimating long term population effects of 

frl 

a.w
(I)!;:; 
w


C) 0:: 
<l>2 
!:::

-...J

1;


u:::E 

~~ 

~
3 


~ 

~ 

~ 

(l)W

w!;i 

~fl: 

~~ 

~
~ 


WO 

0:: :i:

0 

~ffi 

1z~ 

Wet 

0 (.
) 
0:
: 


~ 

YEARS AFTER IRRADIATION 

'. 

',. 

PLATEAU 

'á 

LATENT. 
PERIOD 
YEARS AFTER IRRADIATION 
FIGURE 1 

I 

I 

.I 

NO RETURN TO 

1---¥BASELINE RATE 
I 

I 

I

L w1rn RETURN TO 
I BASELINE RATE 
I 

. low or moderate dose radiation, the durntion of the plateau region is an extrem ely Dose response versus time curve; actual shown in upper po.nel, ideal shown 
~rucial parameter in determining the human cost expected. Furthermore, th e á in lower panel. 


r 

CARCINOGENESIS BY IONIZING RADIATION 243 

244 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

shape of the early part of the dose response curve (the so-ca11ed latent period 
region) is also an important parameter in determining the total magnitude of 
expected population cost. 

In the absence of definitive data on these two issues, we shall idealize such 
dose response curves using simplifying assumptions which are in reasonable 
accord with what experience is available. Figure lb presents such an idealized 
diagram describing the main featur es of th e dose response relationship . The á 
gently sloping part of the response curve is there replaced by an idealized "zero" 
response; followed by an abrupt rise to a flat plateau region. The duration of the 
flat plateau region is then available as a parameter for study, which is all that can 
be done at this time in the absence of definitive data. 

In order to explore the consequences of variation in major parameters (length 
of "latent period" and duration of "plateau"), the folJowing assumptions wiJI be 
used: 
AssUMP'l'JON 1. A single latent period of five years far in utero irradiation ia 
assumed toagree with the estimates of Stewart and Kneale (30], (31]. 
AssUMP1'ION 2. A single latent perwd of 15 years is assumed for all Jorms of 
cancer Jar all irradiati<m beyond lnrth (except in the extreme case). 

Three general case typ es will be discussed, first , that with no return toward the 
spontaneous mortality rate (plat eau, extending throughout the remaining life 
span for the population at risk); and second, that with an idealized abrupt return 

á to spontaneous mortality rates after a 30 year plateau region . And third, an 
ex'tr emo case with a latency period of ten years (instead of 15 years) for all postnatal 
radiation and a plateau duration of 20 years. Both these changes have the 
effect of reducing the expected consequ ences of irradiation. We refer to such cases 
as "extreme" because it appears doubtful that the gently sloping part of the 
dose response curve is any shorter than ten years for the majority of radiationinduced 
malignancies (aside from leukemia , which appears shorter than á a11 
others), and second, because what evidence is available suggests that the plateau 
region is most likely to be greater than 20 years in duration. 

It is essential to consider tho manner of description of the radiation-induced 
excess age specific mortality rates. Commonly, results are presented either as 
excess cases per 1000 population at risk, or as the percentage increment in cancer 
mortality over the spontaneous age specific mortality rates. In some cases data 
are availabl e for individual malignanci es; in others, all cancers are presented as 
a sum. There is no theoretical reason for preference of absolute or percentage 
increments in age specific mortality rates. Doth expressions suffer the defect that 
data derived from one population sample (for example, Japanese subjects) may 
not be directly applicable to another population sample (for example, United 
States subjects). We are far, far from havi11g sufficient epidemiologic data to 
address such questions. 

We have menti oned earlier the desirability of having age specific mortality 
rates for alJ ages of inter est. We are far from that goal. Instead, available to us 
are radiation -excess mortality rates over a span of years of observation of exposed 

'! 

population samples. Therefore, expressed either as absolute or per cent increment 
in mortalities, the data allow only an average value for this span of years. In the 
absence of further evidence, we are here treating the plateau region as a fixed 
percentage increment in cancer mortality per unit of radiation over the entire 

1 
_ 
plateau regi<m. Only extensive further ádata can definitively test the validity of 
this particular approach . In its favor is the conservative nature of this treatment 
for public health purposes. Let us cone.ider the implications of this treatment. 
Since spontaneousage specific cancer mortality rates change with age (rie.ing 
steeply wiih age beyond 20 years), the assumption of aped percentage increment 
for radiation-induced exceea over the whole plateau implies that the absolute 

á increase in age specific mortality rate induced by radiation alao changes with age. 
Thus, if the plateau region represents a 50 per cent increase in mortality rate, 
there will be 1000 extra deaths per 10' persons per year where the spontaneous 
mortality rate is 2000 deaths per 10' persons per year. At a later age, with a 
spontaneous mortality rate of 4000deaths per 10' persons per year, the abaolute 
increment due to radiation would be 2000 deaths per 10' persons per year . Thus, 
a constant percentage increment in the plateau response region implies that 
absolute radiation-induced age-specific mortality rate increments will increase 
over a span of ages. 

"Spontaneous" cancer mortality rates include all known and unknown causes 
of cancer. Therefore, in an epidemiologic study, radiation-induced cases resulting 
from natural radiation background plus medical radiation exposures are included 
in the "spontaneous~á --~ncer mortality rates for the population sample under 
study. Thus, if calculations are presented concerning the percentage increase in 
cancer mortality rate per rad of additional exposure to such a population, the 
true "spontaneous" base rate must be lower than that which includes the radiation 
effect from such sources as medical or natural radiation. Therefore, the true 
radiation percentage increment per rad is actually larger than that presented. 
For calculational purposes this does not introduce any significant complications. 
However, if th!' effect per rad is ~igh, then the á observed per cent increment per á 
rad is ástated to be lower than it truly is, simply because the spontaneous rate 
already is inflated -by that mortality due to natural plus medical (and other) 
radiation for the wpulation sample under study . 

¥ 
1.2.2. Doae response relationships Oller a range of doses. One cannot becertain 
that the time aspects of the dose response relationship are identical over a11 dose 
ranges to be cone.idered. Earlier impressions have been that the "latent period" 
(the gently sloping region described in Section 1.2.1.) might be longer at lower 
radiation doses . This speculation was weakly supported, if at all. The kind of 
study which led to this impression of a longer latent period at lower doses 
generally included small population samples at the lower doses, such that the 
expectancy of cancer at the low doses was often measured as a small fraction of 
one case (37]. Under these circumstances the probability of observing zero cases, 
in a small population sample , is very high . The obaervati<m
of zero cases led to the 
false impression of a long latent period. In this manner the myth arose , concern



CARCINOGENESIS BY IONIZING RADIATION 245 ! 

Ii 

ing "practical thresholds" at lowdoses, that low doses of radiation might not be 

I

carcinogenic simply because the la.tent period could exceed the life span of the 
I 'I 
exposed population . Finkel and co-workers [6l recently demolished this myth 

11

very effectively, based upon a study of some 3200 mice exposed to radium 226. .\

:,

They saw no evidence of a variation in latent period with dose and indicated that 

ii, 
,,

they believed áno other investigators would see variation either if they had an 
adequatepopulation sample in the low dose region. 
In the absence of any evidence to the contrary we shall assume latent periods 


l

and duration of plateau to be independent of the dose range under consideration . 

l 

We shall, further, consider the á effects calculated in the plat~au region of the ! 
Iidealized diagram of Figure lb . The epidemiologio data available cover a wide 
range of doses of radiation, with much of the human data , at least up to recently, I 
having been obtained at moderate or high doses. Our interest, for purposes of 
evaluation of radiation is, in general, for doses in the low to moderate range. It 
is, therefore , essential to know the nature of the dose response curve over a wide 

á
á 
range of doses if the epidemiologiodata are to be utilized for predictive purposes 
in the case of population exposures. 

A priori, in such problems, there is no way to predict the nature of the dose 
response curve. In principle, three generalized dose response patterns, connecting 
observations at high doses with those to be anticipated at low doses are conceivable 
(see Figure 2). 

áCurve A may be taken as one representative curve of a family of curves that 
are convex upward. Clearly, curves of this family express peBSimiemin that they 
predict a higher response at low doses than would be anticipated from a linear 
dose response relationship, such as curve B. Curve C, on the other hand, is a 
representative of a family of curves concave upward. This curve may be con1. 
sidered the "optimistic" curve from the viewpoint of a radiation-associated I 

i 

technology. The optimism arises because there can be a low dose region where the 
excess mortality due to radiation may be extremely low. 

Early in the history of study of radiation carcinogenesis data were available, 
for humane and experimental animals, only for the fairly high dose region, and 
th e shape of the entire curve down to very low doses was unknown . During that 
period, most responsible scientists and radiation study groups such as the 
International Commission on Radiological Prot ection made the prudent assumption 
of a linear relationship of radiation dose versus excess cancer mortality rate 
(curve B). While this did represent a conservative approach consistent with 
sound public health principles, it must be emphasized that this was by no means 
the most conservative position. Any of the family of curves, represented by ., 
curve A (Figure 2) represents a more conservative relationship for connecting á 
available high dose points with the low dose region. But all the se considerations 
describe an era that is now past. Abundant new data, in humane and experimental 
animals, have now become available, permitting description of the dose 
response relationship over a wide range of doses. These new data all point 
unmistakably to the correctneee of curve B, the linear relationship between . 

246 SIXTH BERKELEY SYMPOSIUM: OOFMAN AND TAMPLIN 

...... 

u 

l1.. 

0 

w 

0. 
U) 

w 

~ 

-

w 

!:c 

O::z 
>-0

.-


:J~ 

~a 

0:: <(

on: 
:E~ 

mg

w

uW 

X::>

WO 

RADIATION DOSE 

, FIGURE 2 

Three generalized dose response patterns: (A} higher response at low doses; (B) 
linear dose response relationship; and (C) l9w response at low doses. 

exceBBcancer mortality and radiation dose, over a very wide range of doses for a 
variety of cancers and benign tumors. While one can understand the disappointment 
of radiati(!n-industry promoters over the disapp earance of the fondly 
regarded curve C, it is not possible to condone their lack of appreciation of the 
existence of all this new evidence. 

Let us consider the specific new evidence that has appeared in recent years . 

(1) Shellabarger, Bond, Cronkite and Aponte [28] have demonstrated linearity 
both for breast adenocarcinoma and breast fibroadenoma development in rats 
exposed to X-rays or gamma rays down to total doses of 15 rads. (2) Upton and 
co-workers [34] have demonstrated linearity for mouse mortality from thymic 
lym'phoma down to total doses of ten rads. Studies at lower doses are in progress. 
(3) Finkel, Biskis, and Jinkins [6] have demonstrated linearity for osteosarcoma 
development in the mouse with radium 226 injection over a wide range of doses. 
This is a landmark study, since it is refreshingly characteriz ed by the experi-
I 

'I 

i 

! 

I 

If

I. 
I. 

CARCINOOENESl8 DY IONIZINO RADIATION 247 

mental design of providing an adequate number of experimental animals in the 

low dose region. The authors [7], [8] have pointed out the fallacious conclusions 

derived from the study of inadequate numbers of humans, exposed to radium 

226, who developed osteosarooma. (4) Hempelmann [16) has indicated Jinearity 

in the production of human thyroid adimomas by X-rays, including data points 

down to 20 rads total dose to the thyr oid gland. (5) Beebe, l{ato, and Land [l] 

have extended the leukemia studies in survivors of the Hiroshima-Nagasaki 

bombings. They have demonstrated linearity in the production of human leuke


mia with radiation dose, down to total doses of 20 rads. (6) Stewart and Kneale 

[31] have demonstrated linearity between cancer and leukei:nia induction in 
children during the first ten years of life and irradiation by X-mys in utero in the 
procees of diagnostic obstetric radiography. Their observations covered the 
range of approximately 2.0 rads, thus providing direct human evide~ce in the extremely 
low dose region. (7) Mays and Spiess [29) have demonstrated linearity 
in the production of osteosa.rooma both in human adults and children asa 
result of radium 224 injection . Their experimental data extend down to 00 rads 
r estimated dose. These studies are grossly at variance with the claims of Evans 

[36] of a "threshold" for osteoearcoma in humans by alpha emitters at a dose of 
1000 rads. 
Taken ~verall, these recent and diverse publications leaveávery little reason to 
doubt a linear dose response relationship for cancer and leukemia induction by 
radiation . It has been an interesting phenomenon, indeed, to observe the antics 
of the promoters of radiation-a.saociated technologies during the evolution of all 
these data. Starting with their hope that linearity would fail below 100 rads, they 
have been forced to retreat steadily to 50 rads, then 25 rads, and now they find 
themselves faced with linearity down to the region of a fraction of one rad. Hope 
springs eternal. 

To be sure, for any particular set of data , one could always argue that perhaps 
there is a deviation from linearity somewhere below the dosage represented by 
the lowest experimental point. There exists, however, no rational support for 
such an assumption, since it would require a fundamental change in the mechanism 
of radiation carcinogenesis in the region below the linearity region. Further , 
such an 888Umption, in the absence of evidence supporting it, represents an 
unsound approach to the protection of the public health . The in utero data [31) 
extending down to approximately 0.3 rads, militate strongly against further 
serious consideration of nonlinearity in the very low dose region. 

From the point of view of mechanism, linearity between radiation dose and 
carcinogenic response suggests that a single event phenomenon is involved in the 
production of the critical change which results in the development of cancer.. 
If a single event produces the carcinogenic change over a wide range of doses, for 
a variety of cancers, for several mammalian species, there appears little reason 
to expect a fundamental change in such mechanism at still lower doses. 

Since linearity appears well established for a variety of cancers, we shall here 
consider the dose. response relationship, in the plateau region, as being linear f of 

,I 

248 SIXTH DERK ELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

every type of cancer and leukemia (Figure 3) for prediction purposes. The excess 

age specific mortality rate , for any cancer, can be expreBSed, for a linear dose 

respouse relationship, as a percentage increase per rad over the spontaneous 

mortality rate for that particular cancer. Such percentage increment is simply 

tho slope of the linear plot of Figure 3. For illustrative purposes, assume the slope, 

for a particular cancer, were determined to be one per cent per rad. It follows 

then, for a linear relationship, tha.t 100 rads will produce 100 X 1, or a 100 per 

cent increase in cancer mortality above the spontan eous cancer mmtality rntc . 

That dose which increases the spontaneous cancer mortality rate by 100 per cent 

is commonly defined as one doubling dose of radiation for production of that 

particular cancer. Thus, if a is the slope of the line in Figure 3, then the doubling 

dose is defined as 100/a (for this particular cancer). The doubling dose notation 

does not in any way imply a geometric progreBBion in excess cancer mortality 

rate with increasing radiation dose. Rather, one doubling dose adds 100 per cent 

to the spontaneous age specific mortality rate, two doubling doses add 200 per 

cent, thr ee doubling doses add 300 per ácent, and so forth . It is simply a matt er of 

convenience as to whether radiation carcinogenesis, for any particular cancer, is 

described as the per cent increment in cancer mortality rate per rad or as the 

300 

<t 

0: 
f2 

w 

~ 200 

AS ESTIMATED IN PLATEAU REGION 

~b 

BEYOND LATENCY PERIOD) 

:JZ 

~~ 

0:: (!)

O::J 

:ii: <i 

o::!E

¥ 

------------~--------------------


. SPONTANEOUS

MORTALITY RATE 

~!a

frl..J 

0. ::> 
U)U 

~~ 

<t~ 

DOSE IN RADS 

FIGURE 3 

I 

Linear dose response relationship in the plateau region whel'e the age specific 
mortality rate is a percentage increase per rad . 


. ! 

CARCINOGENESIS BY IONIZING RADIATION 249 

dose in rads, the doubling dose, required to add an euus mortality rate equal to 
the spontaneous age specific rate. Nothing about the doubling dose notation 
inf ere or euággeets that the doubling dose is the same for all forms of cancer. Thie 
ie a matter for experimental evidence to decide. Before considering the question 
of variation of doubling dose with form of cancer induced, it ie necessary to turn 
our attention to the variatiQn in sensitivity to radiation carcinogenesis with age 
at irradiation . 

a 1.2.3. Variation in carcinogenic dose response re'lationship with age at radiation 
exposure. Our considerations thus far have led ue to deeoription of radiation 
carcinogeneais ae follows: (1) dose reaponse relationship, at aáspecified age, is 
linear (Figure 3), characterizedby a particular percentage increment in cancer 
mortality per rad for a particular form of cancer;and (2) dose response relationshipa 
will be treated for the idealized "plateau" region of the curve of the response 
versus time ajter radiation exposure. 
For any particular cancer, occurring at a apecifiedage, how does the slope of 
the dose response line vary with age at irradiation1 It is clear that any refined 
effort to assese population response .to continuous or intermittent radiation 
exposure required consideration of this particular question . To the best of our 
knowledge there exists no the(jry that provides the answer. to this question. We 
must, therefore, have recourse to empirical data. á á 
First, we have the data for in uteroradiation provided by Stewart and Kneale 
[30], and MacMahon [22]. These data, presented in Table I, describe the increase 
in cancer and leukemia mortalities during the first ten years of life 
followiog irradiation in utero. Inspection of the data leads to a best estimate of a 
50 per cent increase in mortality rates for a variety of cancers and for leukemia 
for radiation aseociated with diagnostic pelvimetry , The similarity in percentage 
increase in cancer mortality for diverse cancers and for leukemia, for such ¥ 
radiation, is striking. Stewart and Knee.le [31] indicate that approximately four 
X-ray films lead to 100 per cent increase insuch childhood cancers and leukemia 
mortality rates. During the period of accumulation of their evidence, each X-ray 
film represented leBB than 0.5 rad delivered to the infant in tdero. Conservatively, 
therefore, one estimates that two films, or 1.0 rads, are requir ed for an approximate 
50 per cent increase either in cancer or leukemia mortality rates . (The true 
value may be somewhat higher than the conservative 50 per cent increase per 
rad.) 
None of the Stewart studies addreBB the iBBue of effects of in utero radiation 
upon the development of cancer or leukemia beyond the first ten years of life. 
Both from the Hempelmann etudies [16) and the Hiroshima-Nagasaki studies 
(Jablon and Belsky, (19), involving the irrl!(liation in early infimcy, we have 
conclusive evidence that carcinogenesis extends far beyond the first ten years of 
life. It would be surprising, therefore, if such were not the case for in utero irradition 
as well. In any event, our treatment of the data for estimating population 
exposure specifica]]y explores the effect of various durations of the plateau 
response region. Utilizing the Stewart-Knee.le and MacMahon data, we shall use 

250 SIXTH BERKEJ.EY SYMPOSIUM : GOFMAN AND TAMPLIN 

a 50 per cent increase in age specific cancer mortality rates per rad for irradiation 
received in utero and shall assume this value holds for all cancers and leukemias. 

For iufaucy and childhood irradiation, there are two major sources of information: 
(a) data for thyroid cancer induction in U.S. children irradiated in early 
infancy; and (l~)data for various cancers in Japanese subjects in Hiroshima and 
Nagasaki, who were between O and O years of age at the time of bombing . 

For the thyroid cancers occurring in irradiated children, J>ochin (26] provided 
an estimate that the absolute increment is one case per 101 persons per year per 
rad of exposure of the thyroid gland. Carroll, Hadden, Handy and Weeben (2] 
reported the spontaneous thyroid cancer rate as approximately five to ten cases 
per 101 persons per year in the age range 10-20 years. Combining these data, we 
have previously estimated 10 to 20 per cent increase in thyroid cancer per year 
per rad for irradiation in infancy. (Gofman and Tamplin, (9]). 

Jablon and Belsky [19] have recently provided data for cancers (other than 
leukemia) in persons exposed to atom bombing at 0-0 years of age. For those 
receiving 100 rads or more, the cancer mortality rates (during the period 10 to 
24 years beyond exposure) was 8.4 times that observed for persons receiving less 
than 10 rads. The mean dose for the (100 rad or more) group was not given, but 
it must lie between 100 and 200 rads. So, 100 to 200 rads represent 7.4 doubling 
doses (8.4 .:_ 1.0 = 7.4). Therefore, the doubling dose for cancer production in 
these 0-9-year old children (at exposure) lies between 14 and 28 rads. This 
corresponds to a 3.5 to 7.0 per cent increase in cancer mortality rate per rad. The 
per cent increment á in leukemia mortality rate per rad was even higher, as 
observed in a group of children Oto 14 years of age at the time of bombing [10]. 
A variety of cancers were represented in the Jablon and Belsky data, but the 
limitations of numbers did not allow for treatment of individual typ es of cancers, 
(see also [20]). 

From several sources, data are available concerning the percentage increase in 
specific site cancer mortality rates per rad for persons irradiated in early adulthood 
(32]. Th~se include data for subjects receiving radiation under widely 
different conditions. Includ ed are: (1) breast cancer (Nova Scotia women (23] 
receiving fluoroscopic radiation and Japanese survivors of atomic bombing); 

(2) thyroid cancer ,(Japan ese survivors of atomic bombing); (3) lung cancer 
(spondylitis cases andJapanese survivors of atomic bombing); (4) leukemia 
(epondylitis cases and Jap anese survivors of atomic bombing); (5) stomach 
cancer (spondylitis cases); (6) colon cancer (spondylitis cases); (7) pancreas 
cancer (spondylitis cases); (8) bone cancer (spondylitis cases); (9) lymphatic and 
other hematopo citie organ cancer (10) miscellaneous cancers (spondylitie cases); 
and (11) pharynx cancer (spondylitis cnsee). 
'The range of values determined for percentage increase in cancer mortality 
rate per rad of exposure was between one and five per cent with an estimated 
best value of approximately two per cent per rad. Ideally, one would want to have 
these values determined for groups irradiat ed at a specified age, and one would 


CARCINOOENESIS DY IONIZING RADIA'fION 251 

wish to be certain that the observations were strictly referable to the plateau 
region of the response VCTSU¤ time curve, rather than possibly including some 
data referable to the Jatent period. Dut such ideal data are unavailable. Hence, 
we shall use a two per cent increase in cancer mortality rate per rad as a "best" 
value, we aha.II consider that it applied to aU cancers (the major ones are aIJ 
represented in the data), and we sha11 relate thi~ value to irradiation at approximately 
20 to 30 years of age. As will be noted below, the overall data indicate the 
sensitivity to cancer induction when expressed as the per centá increase over 
spontaneous cancer mortality rates per rad of exposure, is -a .steeply declining 
function of age at irradiation. Therefore, it is entirely possible' that the range of 
one to five per cent increase in cancer mortality rate per rad might be narrowed 
appreciably but for differences ir~ age at irradiation for the young. adult groups 
tabulated. Inaccuracies in dosimetry may also account for part of the range of 
values observed. In any event, the average value of two per cent per rad-for 

irradiat ion in the age range of 20-30 years will be seen below to be consistent 
with trends noted over a very broad span of ages at irradiation. 

Beebe, Kato and Land [l] have recently presented data for cancer mortalities 
during the 1962-1966 period for Hiroshima-Nagasaki survivors who were 
between 25 and 55 years of age at the time of bombing (1~5). It appears quite 
clear, from their studies, that there is a markedly lower sensitivity for cancer 
induction per rad compared with that for younger subjects. These workers 
estimate a 20 per cent increase in cancer mortality risk per 100 rads, -or 0.2 per 
cent per rad for this older group of subjects. 

Summarizing all the evidence just described, we have the foilowing estimates 
of sensitivity to radiation-induction of cancer and leukemia as a function of age 

.. ; 

at irradiation: 

in utero rv50% increase in mortality rate per rad 
0-9 years of age 3.5-20% 
20-30 years of age rv2% 
rv50 years of age rv0.2% 


There can be no doubt that risk of induction of excess cancer mortality rates 
per rad, described as per cent increase over spontaneous mortality rate, declines 
steeply with increasing age at irradiation. Within the totality of available 
epidemiologic evidence now available the estimates just listed provide about as 
much description of this declining function as is now possible. For purposes of 
estimation of the consequences of population exposure, these estimates can be 
reasonably approximated by the step function presented in Table IV. It can be 
shown that the precise values in the step function are not the dominant parameters 
that determine the consequences of population exposure. Of far greater 
importance is the duration of the plateau region of the response versustime 
curve. 

262 SIXTH BERKELEY 8YMP081UM : GOFMAN AND TAMPLIN 

TABLE IV 

VAJUA'110N IN C.t.NCIIB lNDUCl'ION PIIB RAD WITH Ao¥ 

These eetimat.ee represent a step function approximat.ion 
in reasonable accord with the data point. 
available In the text. 


lncreue In cancer mortality rate 
Age at irradiation Jiff' rad (in Plateau Region) 
(years) (per cent) 

I¥ 
utcro 60 
(HS 10 
(HO 8 

11-16 8 
16-20 4 
21-30 2 
31--40 1 
41-tiO 0.6 

.61-60 0.25 
61 and beyond Assumed negligible :
, 


t: 
I 

) 

ᥠ2. The carcinogenic consequence, of population etposure 
to environmental ionizing radiation 

The major parameters required to-ev~Juate the consequences of population , 
exposures to ionizing radiation have been identified in the foregoing discuesion. 
That the epidemiologic data are far Jesa than ideal for quantitative evaluation 
ie undeniable. A humane society should consider itself fortunate that better data áá 
are ~t available. , 

The various sources of potential ionizing radiation exposure include natural 
radiation, radiation from weapons testing f~lout, radiation from a variety of 
peaceful atomic energy programs, and radiation from diagnostic medical and 
dental exposure. Since the signing of the atmospheric test ban treaty, weapons 
testing fallout hae become a email source, and should decline further, unless 
nonsignatoriee to that treaty increase weapons testing appreciably. 

Peaceful atomic energy programs are currently allowedto deliver an average 
dose of 0.17 rads per year to the U.S. population. At present, 80 far as measurements 
allow dose estimates, it appears that such programs deliver only a small 
fraction of this "ailowable" average dose. Nevertheless, with the burgeoning 
growth of the nuclear electric power industry plus numerous proposals for 
utilization of "peaceful" nuclear explosivesá (Project Plowshare) plus growing 
radioisotope utilization, the exposure to the population from the "peaceful'' 
atom wiU undoubtedly grow. SoJong as 0.17 rads per year remains permiesible 
by Federal Regulations, there is good reason to believe the full exposure may 
ultimately be reached. It is, therefore, of special importance to calculate the 


CARCINOGENESIS BY IONIZING RADIATION 253 

cancer and leukemia expectation for such an average exposure to the U.S. 
population. 

Medical and dental exposures to X-rays have resulted in a steadily increasing 
average population dose of ionizing radiation. Medical diagnostic X-ray exposure 
has recently been estimated to provide approximately 0.10 rads as an average 
population somatic tissue dose (Morgan [25]). We are in full accord with 
Morgan that advantage should be taken of modern technology to reduce such 
exposure drastically, especially since Morgan has estimated that a ten-fold 
reduction in average exposure could be accomplished without any loss. in diagnostic 
X-ray information. ' á 

N atura.l radiation provides an average population exposure in the neighborhood 
of 0.125 rads per year. Such features as radioactivity conte11t of building 
materials, radioactivity in rocks of the earth, and elevation abave sea level 
account for variation in such natural doses among population subsamples. 
Through a strange system of logic, or better, illogic, it is commonplace for 
promoters of radiation-associated technologies to arrive at the wholly absurd 
conclusion that doses comparable to natural radiation cannot be carcinogenic 
because natural radiation "has always been with us." 

The above sources of ionizing radiation represent primarily . low Linear Energy 
Transfer (LET) radiation. Primarily the radiations are X-rays, gamma rays, 
and beta rays. Carcinogenic effect per rad will be essentially identical for aU 
these radiation sources. One could estimate population consequences per 
millirad per year, for natural radiation exposures, for medical exposures, or for 
the 0.17 rads per year permitted as an average population exposure for peaceful 
atomic energy activities. Since the concern of this Symposium is with matters 
related to environmental pollution, it is particularly appropriate to estimate the 
consequences of the 0.17 rad per year average allowable population exposure. 
The U.S. Government [5] has decreed this much population pollution to be 
permissible (Federal Radiation Council, 1000). The scientific and lay communities 
should be especially interested in the carcinogenic consequences of this 
permissible pollution by ionizing radiation. It should be evident that the consequences 
of natural, medical, or weapons faUout exposures can be derive<.! from 
the consequences of 0.17 rads per year by direct application of the linearity of 
dose ver81Uresponse. 

We have previously estimated the cancer plus leukemia consequences of 
exposure to 0.17 rads per year to be approximately 32,000 extra cane-er plus 
leukemia. deaths per year, at equilibrium, for the U.S. population ti.tits current 
size of 2 X 101 persons, [10], [11]. That estimate was based upon tbe average 
two per cent increase in cancer mortality :tate per year per rad of exposure 
observed for young adults, coupled with a 30 year duration of the plateau region. 
With the more extensive data available in the past year concerning sensitivity 
variation with age, a more refined estimate is now possible. Moreover, it is 
important to explore the implications of both a longer and shorter duration of 
the plateau region, aswell as the implications of variation in "lat ent" period. As 

254 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAJIIPLIN 

we shall see, the estimate of 32,000 extra deaths per year is by no means overly 
conservative, since this number can rise several fold if it turns out that the 
plateau region extends throughout the life span of exposed populations. 

¥ 2.1. Cancer hazard for average population expoBUre (total body irradiation) at 
0.17 rads per year. Three genern.l cases will be considered here, the case where 
the plateau persists indefinitely after latent period, where the p)ateau region 
persists 30 years with subsequent return to spontaneous cancer mortality rates , 
and the extreme case where the plateau region persists 20 yea.re with a la.tent 
period of 10 years for post natal radiation (in contrast with 15 years for the first 
two cases). 
CABE 1. Plateau persists indefinitely after latentperiod. 
The calculation is based upon the consideration of the total per cent increm ent 
in radiation-induced cancer mortality rate at a particular specified age as made 
! ; up of tho sum of contributions from radiation received at ages less than the 
specified age. The procedure will be illustrated below. 
! : For in utero irradiation we have stated above that a five year latent period will 
; i 

be assumed. 
In Case 1 ca.Jcula.tione, a 15 year latent period is assumed for all post natal 
irradiation. 
, I Radiation received in any particular year of life begins to contribute to .cancer 
mortality rate only after the )a.tent period is over. Thus, radiation in the first 

i 

year of life startsc~ntributing to cancer mortality in the 16th year of life. Radiation 
in the 10th yti'atá of life starts contributing to cancer mortality in the 25th 
year of life. 

For in utero irradiation at 0.17 rads per year, approximately 0.12 rads would 
be received in the course of a pregnancy . At 50 per cent increase in cancer . 
mortality rate per rad, we calculate 50 X 0.12, or a six per cent increase in 
cancer mortality rate for the in utero radiation exposure. Now, since we have 
assumed a fiv~ year la.tent period for in utero radiation, there is obviously uro 
cancer mortality increment during the first four years of life. For the fifth year 
of life and beyond, however, the six per cent increment in cancer mortality rate 
would apply for each year that the p)ateau region persists. In Case 1, under 
consideration here, á ~s would be for the remainder of the 'life span of the exposed 

; ' 

population. 
ii For irradiation in the first year of life (0.17 rads), the sensitivity factor to be 

I ,' 

ta.ken from Table IV is ten per cent per rad. Thus, 10 X 0.17 = 1.7 per cent 
increase in cancer mortality rate. However, since we a.re taking the la.tent period 
I for post natal irradiation to be 15 years, it folJowe that irradiation in the first 

I 

year of life does not begin to add its increment . in cancer mortality rate until the 
16th year of life. For Case 1, this increment would be effective for all subsequent 
yea.re for the exposed population, since indefinite persistence of the plateau is 
assumed . 

Th erefore, for th e 16th year of life, there is six per cent from the in utero 
irradiation plus 1. 7 per cent from irradiation in the first year of life, áfor a total 

,. 

'iá 
~ 


OARCINOOENESIS DY IONIZING RADIATION 255 

increment of 7.7 per cent in radiation-induced cancer mortality rate . For the 

17th year of life, we have 6 per cent from in utero radiation, 1.7 per cent from 

1st year irradiation, plus 1.7 per cent from the 2nd year irradiation, for a total 

of 9.4 per cent increment in cancer mortality rate from the irradiation received 

in utero plus the first two years of post natal life. . 

The increment in cancer mortality for irradiation in each subsequent year of 
life is calculated in the samo manner as the product of the sensitivity factor from 
Table IV (for that year of life) by the 0.17 rads. The totalincrement in cancer 
mortality rate for any particular year of life is the sum of all contributions to that 
year from irradiation at earlier years, taking into account tliat. no increment is 
derived until the latent period is over for that particular year's irradiation. In 
this manner, a value for total per cent increment in cancer mortality rate becomes 
available for every year of life, taking into accomit, appropriat ,ely, irradiation 
received at all earlier periods of life. For ease of comparison with U.S. Vital 
Stati8tica,these annual values are averaged for five year age intervals. á 

In assessing impact of irradiation upon the population, we can consider just 

the per cent increase in age specific cancer mortality rate . The values just 

calculated provide this resul t. Or, alternatively, and of poBBibly greater interest, 

is the absolute increase in number of cancer deaths per year at each age for the 

population at risk. We are now immediately in a position 'to m!lkc this estimate. 

From U.S. Vital Statiatica,the absolute number of spontaneous cancer deaths 
per year for each age interval is provided (1966 data used here). Now, let us 
suppose for a particular age that the combined increment due to all prior radiation 
is a 15 per cent increment in cancer mortality rate over the spontaneous 
cancer mortality rate . And let us suppose, further, that for this particular age, 
th e spontaneous cancer mortality rate is 1000 cases per year . The radiationinduced 
increment is then (15/100) X (1000), or 150 radiation-induced cancer 
deaths for the population at this particular age. 

In a similar manner, a tabulation of absolute numbers of radiation-induced 
cancers by age interval can be built up, separately for males and females. Finally, 
the total annual number of radiation-induced cancer fatalities can be calculated 
by summation over all age intervals for males plus females. This tabulation, for 
Case 1 calculations, isprovided in Table V. The result, a prediction of some 
104,000 annual additional cancer fatalities is more than three times worse than á 
our earlier estimate. We are, of course, not at all surprised at this result, for we 
bad indicated earlier that taking sensitivity as a function of age into account 
could make for a much more serious prediction. Additionally, Case 1 calculations 
consider the plateau region to extend indefinitely, whereas our earlier calculations 
were based upon a 30 year duration of plateau. 

It can be further noted that if the real effect is as large as shown in Table V 
(and no reason exists to reject the Case 1 analysis), then the contribution of 
natural plus medical radiation must constitute a quite appreciable segment of 
the so-called "spontaneous" cancer mortality rates. One could consider a second 
iteration on the total calculation, (lorrecting the "spontaneous" mortality f!l,tes 

256 SIX.TH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

TABLE V 

RADIATION-INDUCED CANCl!lB MORTAUTJ' BY Aoz AND SEX 

6 year latency for in ulm> radiation. 
15 year latency for all other radiation. 
Plat.eau constant after latency period 
. 
Exposure: 0.17 rads/year. 


Total spont11neous cancer mortality per year -303,691 cases. 

, I 

Total radiation-induced cancer mortality per year -104,259 cases. 
Per cent Increase in cancer which would oocur with 0.17 rads I 
averageannual exposure -34.3 per cent. 

Per cent Annual Annual 
increase in Annual radiation Annual radiation 
cancer spontaneous induced spontaneous induced 
Age mortality cancers cancers cancers cancers 

interval rate (male) (mal e) (female) (femnle) 

(Yea111) 
0-4 0 827 0 720 0 
6-9 0 820 60 006 30 

10-14 6 673 40 482 29 
16-19 9.4 820 77 640 51 
20-24 17.2 764 130 608 88 
26-29 23.3 790 186 733 171 
30-34 27.8 1,145 318 1,418 304 
35-39 30.5 2,104 641 2,800 881 
40-44 32.2 4,163 1,340 5,565 1,791 
45-49 3~u . 7,109 2,372 8,732 2,914 
60-64 34.2 12,363 4,231 11,950 4,089 
55-69 34.8 17,694 6,123 14,369 4,907 
60-M 35.2 22,469 7,009 15,780 6,565OHO 36.5 26,275 8,968 17,92i 6,358 
70-74 35.7 25,698 9,169 18,746 6,689 .. <76-79 35.8 21,221 7,589 16,650 6,964 i

: !I 

( 
80-84 35.8 13,318 4,763 12,141 4,342 
. 
I 
I 
85 and 
Leyond 35.8 7,793 2,787 8,990 3,217 

i; I 

Total 
164,948 56,703 138,743 47,556 

downward (by subtracting the contrjbution from natural plus medical radiation) i. 
and correcting th~-per cent increment per rad upward as a result of the lower 
true "spontaneous" mortality. These two effects would tend to balance out, so ! I 

; I 

th at the final calculations of population risk would not be seriously altered. It I! 
would, however, point up the major contribution of natural plus medical radiation 
to the existing cancer mortality rate, wholly aside from increments due to 
peaceful atomic energy programs . 

CASE 2. Plateau region persists SO years, with subsequent retum tospontaneous 
cancer mortality rates. 

It is possible that once the increased cancer risk due to irradiation is fully 
developed (the plateau region), such risk may not persist indefinit ely. It is 
difficult to know, wjthin presently availab le epidemiological data, how many 


r . 
I 
I 
CARCINOGENESIS BY IONIZING RADIATION á: i 257. 

years the plateau ]~ts, if it does indeed onJy Jast a Jimited period for cancer. A 

calculation , based upon a 30 yeat p]ateau period ie provided here. In this calcu-, 

lation, the contribution of radiation received in any particuJar year of Jife is 

credited for 30 successive years, following the ]a.tent period. After this, the 

contribution of that particular radiation ie cut off. Thus, for example, the pei: 

cent increment in cancer JJiortaJity rate from radiation received during the let 

year of life begins to be credited starting in the 16th year of life, and is credited 

for each subsequent year of life out to the 46th year of life. Beyond the 46th year 

of life, no crediting toward radiation-induced á cancer mortaJity ie given for 

irradiation in the first year of Jife. Similar calculations are á~e for irradiation 

in each subsequent year of life. Otherwise, procedures of calculation are similar 

to those for Case 1, Table V (five year latent period for in utero radiation; 15 

year latent period for all post natal irradiation). The calcuJations for Case 2 ~ 

presented in Tab]e VI. 

. !, á 

TABLE VI á 

CABB 2: luDIAflON-lNDUCIID CANCIIB MOBTAUTI' BT A011AND Sax 

5-year latency period for in ulero irradiatiol)
, 
.. 

15-year latency period for all other irradiation. 
. !l 

Plateau: 30 years beyond latency period ¥ . 

Expoame: 0.17 radl/year. 
' 

l 
Total epontaneoue cancer mortality per year -á 303,601 caees. 
Total radiation-induced cancer mortality per year ¥ 74,013 cues. 
Per cent increase in cancer which would occur with 0.17 rads average annual exposure ¥ 24.4o/a-

Percent Annual Annual 
increa.se Annual á radiation-Annual radiation-

Age in cancer epontaneoue induced epontaneoue induced 
interval mortality cancen, cancers . cance111 cancera 
(yean,) rate (male) (male) (female) á. (female) 

0-4 0 ffl 0 720 : 0 

5-9 6 826 50 606 36

.

10-H 6 673 40 482 29 
15-19 9.4 820 77 546 5i 

508 .
20-24 17.2 7M 130 87 
25-29 23.3 796 185 733 171 
30-34 27.8 1,H5 318 1,418 39' 
36--39 24.5 2,104 515 2,890 708 
40-44 26.2 4,163 1,091 5,665 1,458 
45-49 26.0 7,109 1,863 8,732 . 2,288 
50-M 25.4 12,363 3,140 11,950 3,036 
55-59 24.9 17,504 4,381 14,359 3,575 
60--64 24.6 22,469 5,527 15,780 3,882 
65-69 24.4 25,275 6,167 17,921 . 4,373 
70-74 24.6 25,608 ~.322 18,746 4,612 
75-79 24.4 21,221 5,178 16,650 4,063 
80-84 24.5 13,318 3,263 12,141 2,975 

85and 
beyond 24.0 7,793 1,870 8,996 .2,159 . 
Total 164,048 40,117 138,743 aa,800 

á
' 


I 
258 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

I 

I It ie evident, on comparison of Table V with Tab]e VI, that reduction of the 

! 
plateau duration provokes a marked drop in the expected mortalities (104,000 
down to 74,000). However, both vaJuee are extremeJy high and should raise 
grave concern about the nature of the societaJ benefits that might be worth 
permitting population exposures ae high as 0.17 rads per year as the average 
exposure. No comfort whatever is to be drawn from repeated aesurances that 


I 

'I abound froi;n nucJear promoters to the effect that "we'JJ never give you the full 
; : allowable exposure" while at the same time they staunch]y defend retaining such 
an aJJowable exposure. Good intentions are matcriaJly aided by codification into 

I, Federal Regulations. 
The calculations shouJd be especiaJly illuminating to the sponsors of this 

ii 

! 
; 
Symposium addressing the issue of designing epidemioJogic studies for tbe 
evaluation of societal impact of environmental pollutants. A quarter of a 
century into the atomic era, the epidemioJogic data indicate that our permissibJe 
doses could lead to a public health caiamity-a 25 to 35 per cent increase in 
annual cancer . mortality rate. No evidence at this time militates against the 
most pessimistic calculation (Case 1). We have commented elsewhere that this 
]ate realization based upon epidemiologic data could aJJ have been averted by 

, . 

i. 
judicious use of experimental animal data decades ago (Gofman and Tamplin 

[11]).

Iá It is of interest to specuJate upon possibiJitiee that might have resuJted in the 
Case 1 or Case 2 calculations leading to a serious overestimate of the cancer 
hazard . For exampJe, one might consider the possibility that dosimetric or other 
errors bad led to an overestimate of the percentage increment in cancer mortality 
rates per rad at aJJ of the ages listed in Table IV. We believe it is unlikely that 
such an overestimate could be ae much as two-fold . Moreover, one might also, 
under such circumstances, consider that the seriousness of the resuJts ie underestimated 
as a.result of dosimetric errors. 
CASE 3. Th e extnme case: plateau regi<m perBists £0 years, lattnt period of 10 
years; post natal irradiation. 
It ie important to ascertain what the prospects for "optimiFm" may be with 
regard to carcinogenic consequences of population exposure to radiation. ThereI 
fore, we may conljider th e poesibiJity that the duration of th e plateau region of 
the response versustime relationship is materially shorter than 30 years. From I¥ ¥ 

' ' ¥

the epidemiologic evidence availabJe, admittedly still scanty, we would estimate I. 
that it is highly unlikeJy for plateau duration to be Jess than 20 years . (Radiation


1 á 

induced cancers have been described occurring 30 to 40 years after exposure.) 

' 
' 
! .....

But since this should lessen greatly the expected consequences, we shall test here 
a 20 year duration for the pJateau region. It is also evident that if the latent 
period were shorter thnn 15 years, the net carcinogenic effect would be reduced 
'further, because the Jarge per cent increments in cancer mortality rate for 
irradiation early in life would not be carried as far fmward into the Jater agf! 
spans where the spontaneous cancer mortality rates are high and, hence, the 
products of per cent increment by spontaneous mortality rates are also high. 
The procedure of calculation is precisely the same as that employed for Case 1 


CARCINOGENEBl8 BY IONIZING RADIATION 259 

and Case 2 except for the alterations in the two parameters, plateau duration 
and latent period for postnatal irradiation. The results are presented in Table 
VII. The final estimate for population exposure at an average of 0.17 rads per 

TABLE VII 

CAe11 3: RADIATION-INDUCllDCANCllR MoRTALITr Br Ami: AND Ssx 

5 year latency period for in ulero radiation. 
10 year latency period for all other radial.ion. 
Plateau: 20 years beyond latency period. 
Expoeure: 0.17 rads/year 
. 


Total spontaneous cancer mortality per year -303,691. 
Total radiation induced cancer mortality per year -9,428. 
Per cent lncreaee In cancer which would occur with 0.17 rads average annual expoeure -3.1% 

'¥ 
I 
Percent. Annual Annual 
mcreaee Annual radiation Annual radiation 
ineanoer apontaneous induced apontaneous induced 
Age 
interval 
mortality 
rate 
cancers 
(male) 
cancers 
(male) 
cancers 
(female) 
cancers 
(female) 
(Years) 
0-j 0 827 0 720 0 
5-9 6 826 50 600 36 
10-14 u 673 63 482 45 
15-19 17.2 820 141 546 04 
20-24 23.3 754 176 508 118 
25-29 21.8 796 173 733 160 
30-34 21.1 1,145 241 l,U8 299 
35-39 16.0 2,104 315 2,1100 434 
4()-j4 10.2 4,163 425 6,565 568 
45-49 6.6 7,109 471 8,732 676 
50-54 4.6 12,363 566 11,950 650 
55-59 3.3 17,594 577 14,359 474 
60-64 2.2 22,469 503 15,780 347 
65-69 1.6 25,275 40'l 17,921 287 
70-74 1.2 25,698 311 18,746 225 
75-79 1.0 21,221 212 16,650 167 
80-84 1.0 13,318 133 12,141 i21 
85and 
beyond 1.0 7,793 78 8,990 00 
Total 164,948 4,837 138,743 4,691 

year is 9428 extra cancer deaths per year. While this is a marked reduction 
compared with the estimates for Case 1 and Case 2, the seriousness of such 
radiation exposure levels is self-evident. We would doubt that a more 11optimistic" 
set of parameters than those for the Case 3 calculation is likely to be 
justified. 

a 3. Life shortening by radiation-Induced cancer 

A variety of pronouncementit have greeted estimates of the serious carcinogenic 
hazard of population exposures to doses in the neighborhood of 0.17 rads 

260 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 
per year . One such we have dealt with above, namely, the statement that, after 
all, this dose is comparable in magnitude with natural radiation, which humans 
have endured on eart h for the entire history of the species. No further comment is 
required. A second ie that even though the calculated cancer deaths may indeed 
occur, they will occur so late in life as to be iuconsequential. Grendon has 
championed this approach, readily provable to be false, [14). A variant of this 
approach is that of Sagan [27) who has pointed out that, even if the calculated 
cancers did occur, the average life shortening for the exposed population would 
be very sman. In fact, it has become fashionable of late to estimate the deleterious 
effect of environmental hazards in terms of average life shortening for the exposed 
populatio~ . We hear "Wouldn't people be willing to give up a few minutes, 
hours, or days of lire span so we can all enjoy 4clean, cheap, and safe' nuclear 
electricity?" This approach to evaluation of life shortening is exceeded in its 
scientific fallacy only by ite immorality in public deception. 
If those who die prematurely of cancer due to irradiation are averaged in with 
those who do not, the apparent loss of life expectancy appears quite small. What 
really matters is the average loss of life expectancy for those who do develop 
radiation-induced cancer. Their loss of decadesof life expectancy is not easily 
: i ¥ I 
recompensed by a "loan" from those who do not become victims . The losses in 
life expectancy for the victims are readily estimated. If the victims of radiationinduced 
cancer bad not been irradiated, there is a priori every reason to assume 
they would have experienced the usual life expectancy associated with their age 
group at victim~ation. Thus, from 1971 estimates, a man at age 25 years has a 
life expectancy of 45.5. years. If he dies at 25 years of age of radiation-induced 
: i cancer, he has lost 45.5 years of life expectancy . In 'fable VIII are presented the 
calculated losses of life expectancy by age group for the persons developing 
á I , 
radiation-induced cancers, as well as the average loss of life expectancy for all 
the cases of radiation-induced cancers as a group. For males developing radiation-
induced cancers, the average loss of life expectancy is 13.1 years. For 
femaies, the loss is 13.7 years. Such average losses hardly are in accord with 
I Grendon's assertion that the radiation -induced ácancers occur so late in life as to 
' i be inconsequential. For men in the age group of 65 to 69 years, the life expectancy 
( 
: l (as of 1971) is 11.5 years. If these men lose their life through radiation-induced 
cancer at 67 years; ,hey have lost 11.5 years. One wonders whether Grendon bas 
I 
! 
checked with such members of the population to ascertain that these "old" 
people need not care about losing 11.5 years of life. 
Let us return to the Sagan view of only a minor loss of life expectancy (hours 
or days). If the man-years of li£e expectancy are distributed into the entireU.S. 
male population of 95,919,000 men instead of into the 56,703 victims of radiationinduced 
cancer, the average loss of life expectancy is computed to be 2.8 days. 
'I This practice of hiding the serious loss in life expectancy for the vict ims of an 
environmental poison by averaging the loss over the larger group of nonvictims 
deserves strong condemnation. The sole effect of the practice is to obscure the 
real hazard of an environmental poison from the public, carried through on 


CARCINOGENESIS BY IONIZING RADIATION 261 

TAB~E VIII 

Lo88 or Lin EXPEO'r.ANCT l'ROI( ftADI.ATION-INDUcm> CANCIIR 
(Data from Table V) 

Life expectanc ies are somewhat higher for females than males, 80 the use 
of male life expectancies here leaJs to a slight undereatimateof the Joss of life 
expectancy for females with radiation-i nduced cancers. 
Note : The use of data from Table V (the Case 1 estimate) leads to the lowe,t estimate of Joss 
of life expectancy. For Case 2 (Table VI) and Case 3 frable VII), the radiation-induced excess 
cancer mortalities are more prominent at earlier ages. Hence, for either of these the life expect.ancy 
1038 would be appreciably Mgher than the 13 year estim~te for Case I. 

@ 
© Average ©X@ @ 
Number of Loss of (Man-Number of @X@l Age radiation-life years of radiation-á.\ Woman-years 
group induced expectancy 1088 of induced of loss of 
(in years) cancers (years) expectancy) cancers expectancy 

0-4 0 66.1 0 0 0

' 

5-9 60 62.0 3,100.0 36 2,232.0 
10-14 40 57.2 2,288.0 29 1,658.8 
15-19 77 52.5 4,042.5 51 2,677.5

' 

20-24 130 47.8 6,214.0 88 4,206.4 
25-29 186 43.2 8,035.2 . á171 7,387.2 
30-34 318 38.6 12,274.8 394 15,208.4 

C

35-39 641 34.0 21,794.0 881 29,954.0 
40-44 1,340 29.5 89,530.0 1,791 52,834.5 
45-49 2,372 25.3 60,011.6 2,914 73,724.2 
50-54 4,231 21.3 00,120.3 4,089 87,095.7 
56--59 6,123 17.7 108,377.1 4,007 88,446.9 
60-64 7,909 14.4 113,889.6 5,555 79,992.0 
65-69 8,968 11.5 103,132.0 6,358 73,117.0 
70-74 9,169 9.1 83,437.9 6,689 60,869.9 
75-79 7,589 6.9 52,364.1 5,954 41,082.6 
80-84 4,763 5.1 24,291.3 4,342 22,144.2ss+ 2,787 -3.0 8,361.0 3,217 9,651.0 
Total 56,703 741,263.4 47,556 652,282.3 

I 

Average 1088 in life expectancy (males) -
7!~:á4, 

or _!!_1 years. 

3

Average 1088 in life expectancy (females) -6!;!!!á

, or 13. 7 years. 

, 

behalf of the promoters of the technology responsible for the distribution of the 
poison. 

The ridiculous nature of this approach to calculation of Joss of life expectancy 
would be obvious to everyone if we considered an issue like the death of young 
Americans in Vietnam. After all, when those Americans who a.re at áhome a.re 
averaged in with those who are killed in Vietnam, the averageJoss of life expectancy 
is small, the deaths are not tragic, for, on the average, everyone is just losing 
days from their life. The public would not stand for such nonsense. Why they á 

262 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 
arc so readily brainwashed by pseudoscientific evaluation of loss of life expectancy 
for environmental poisons escapes understanding. 
á ¥ 4. Are there possible mitigating factors which could reduce 
the estimated hazard of population exposure? 
We have considered above the crucial parameters, such as latent period 1md 
duration of carcinogenic response plateau, which can determine in a major way 
the magnitude of expected population cost. We must address a few other concepts, 
since the uninitiated may hear that such concepts provide a reasonable 
basis for expecting a. lesser hazard. As will become evident, there is essentially no 
reason to expect any lessening of hazard. Among these concepts are : (a) a. possible 
threshold, (b) a possible ''practical" threshold, (c) protraction of radiation, ar:d 
(d) repair of radiation injury. 
¥ 4.1. Thresholds: absolute and "practical." In the discussions above it was 
demonstrated that abundant new data concerning the low dose region of radiation 
exposure indicate linearity of dose veram carcinogenic response over a wide 
range of doses. There really never has existed any acceptable evidence for an 
absolute threshold of exposure below which radiation carcinogenesis will not 
occur. It is to the credit of all radiation study groups that they have consistently 
rejected supposed evidence for radiation thresholds with respect to carcinogenesis. 
The linearit y of dose versus response, now demonstrated down to very low 
doses, indicates there is no reason to expect any evidence for an absolut e threshold 
ever to develop. á 
One total non sequiturhas often been introdu ced into discussions concerning a 
possible threshold. That concerns the development of signs and symptoms of 
acute radiation sickness following radiation exposure. Everyone cognizant with 
this . field has known for decades that acute radiation sickness is not linearly 
related to radiation dose, whereas carcinogenesis now appears definitely so 
related . The underlying mechanism in acute radiation sickness relates to whether 
or not cell replacementcan operate rapidly enough to prevent such phenomena as 
mucosa! ulceration or Jeukopenia. At radiation doses where cellular replacement 
is rapid enough, rajiation sickness just does not occur. For carcinogenesis, not a 
shred of evidence has ever been adduced that cellular replacement can avert ;.. 
cancerous change. 
'l'he modification of the threshold concept to the "practical" threshold we 
have dealt with above. There is no basis for expecting any help from this concept. r~ 
¥ 4.2. Protraction of radiation. It is very commonly stated, with appallingly 
little evidence, if any, that if radiation is delivered slowly, the carcinogenic effect 
is lessened. A little later this was modified io the statement that protraction 
protects against carcinogenesis from low LET radiation (such as beta rays, 
X-rays, or gamma rays), but not high LET radiation (such a.a neutrons or alpha 
particles). A variety of experiments have been cited as direct demonstrationa that 
protraction of radiation affords protection against carcinogenesis, [34). 


CARCINOGENESIS DY IONIZING RADIATION 263 264 SIXTH BERKELEY SYMPOSIUM : GOFMAN AND TAMPLIN 

Almost invariably spch experiments contrnst acute delivery of radiation earl11 
in life with protracted radiation extending from early in life through a significant 
part of the life span of the experimental animal. In some of the specific cases 
repor~, the author has himself demonstrated a marked diminution in carcinogenicity 
of radiation with increasing age at irradiation [33). In other studies, this 
point is entirely neglected . In the material presented throughout this communication 
the steep decline in carcinogenicity per rad with age in humans has been . 
docum ented . Thus, the most probable interpretation of experiments contrasting 
acute versus protracted irradiation is simply that protraction provid es part of 
the irradiation at olderages and, .hence, cancer induction is l~e ned. All that this 
re-emphasizes is the extreme seriousness of radiation as a carcinogen early in 
life. Whether there truly exists any residual mitigation from radiation protraction 
is uncertain within present evidence. Certainly such bodies as the 
á
áá1ntemational 
Commission on Radiological Protection have acted with wisdom, from. the 
public health viewpoint, in refusing to count upon protraction of radiation to 
lessen carc~og enic hazard . . 

We feel strongly that it would be appropriate to go further, for any environmental 
pollutant, and state the following principle : " If under any dosage rate 
schedule a pollutant shows a certa in magnitude of toxic effect, that toxic effect 
should be assumed to be at leaatCJ3high for any other dosage rate schedule, until 
and unless definit ively proven otherwise." 

Adh erence to such a public health principle might reduce the danger from 
those individuals all too ready to spew forth.clicMs, such ns, "Maybe the poison 
won't be so bad if we give it slowly." 

In the carcinogenesis field ther e is one special circumstance that deserves 
special consideration here . This is the case, eith er in humans or experimental 
animals, of a cancer whose incidence does not increase spontaneously in a á 
monotonic fashion with increasing age. While most of the familiar cancers of 
adult life do show monotonically increasing incidence rates with increasing age, 
this is not true for several human cancers that occur in childhood (for example, 
neuroblastoma, Wilma' tumor) . Some of these childhood cancers show a peak 
incidence in the first decade of Jife and a declining incidence thereafter . There is 
every reason to suspect that certain cancers of experimental animals may have 
a similar age related incidence pattern . 

Earlier in this communication we presented a generalization (Generalization l) 
which stated "the correct way to describe the phenomenon (cancer induction by 
ionizing radiat ion) is either in terms of the dose required to double the spontaneous 
mortality rate for.each cancer, or, altem 'atively, of the increase in mortal ity 
rate of such cancers per rad of exposure ." Let us consider what might occur if one 
happened to do dose protract ion lleTBU8 acute radiation studies on a cancer 
having a peak incidence at one age period . If Generalization l is correct, then 
the results obtained by dose protraction could appear to be a lesser incidence of 
the cancer simply becau.,eof its spontaneous age incidence pattern, and be wholly 
unrelated to any "protection" resulting from slow delivery of the radiation. We 

suspect that in time such an experiment will be done, and the results misint.crpreted, 
to society's detrime nt . 

1 8 4.3. Repair of radiation injury. Lostly, we must consider the phenomenon 
known as "repair." We hear commonly stated that DNA repair mechanisms 
exist and, hence, low dose radiation may not be as harmful as a carcinogen as 
had been suspected. No serious student of biology doubt s th e existence of DNA 
excision-repair or of such phenomena as lightr-stimulated thymine dim er rep air. 
However, the existenceof such phenomena by no means argues in any way for 
mitigation of radiation carcinogenesis. There is no evidence whatever that has 
been adduced relati ng such repair to ionizing radiation carcinogenesis. 
When we observe the induction of cancer by ionizing radiation, we are, as yet, 
totally in th e dark concerning the mechanism operativ e in produ ction of the 
cancer . Whatever such mechanism may be it is entirely conceivable that a large 
part of the carcinogenic damage of radiation may get repaired. What we are 
observingis th e net, .unrepaired carcinogenic damage . The o~ly conceivable way 
that any such hypoth etical carcinogenic repair could help at low dose would be 
for moreejficientrepair to exist at low doses or slow delivery of dose than for high 
doses or rapid delivery of dose. If the fraction of unrepaired carcinogenic damage 
by radiation were ind ependent of total dose and/or dose rate, th en the very 
existence of any such repair mechanism would be wholly irrelevant as a possible 
mitigating fact or for population consequences of low dose rate exposure. And 
since (a) we know of no such carcinogenio repair mechanism, and (b) nothing 
whatever is known áabout variation in efficiency of an unknown repair mechanism 
as a function of dose and dose rate , it should be clear that all this represents the 
sheerest of speculative fancy. The linearity of dose response in carcinogenesis by 
radiation argu es strongly against repair of carcinogenic damag e at low doses 
with decreasing repair at successively higher doses. 

Inj ection of speculative fancy into a serious matter of pubJie health protection 
is irresponsible. Relating DNA repair phenomena to mitigation of carcinogenic 
injury by radiation, in the absence of any demonstration that these phenomena 
are in any way related to each other, seems equally irresponsible. 

\

:a 6. A re-look at the purposes of this symposium after consideration of the 
.! potential population consequences of low dose radiation exposure 

Do we really want to design epidemiologic studies to evaluate the population 
effects o{pollutants, or potential pollutants, past, present, or future? Radiation, 
to paraphrase many nuclear enthusiasts, is one o{ the most intensiv ely studied 
environmental poisons. Yet, for those who have had the patience to read through 
this communication, certain points, we hope, will stand out. Twenty-five years 
into the atomic era, and 75 years after Roentgen 's discovery of the X-ray, we 
realize that, while the risk of cancer is high, certain parameters, still not pOBBihle 
to evaluate within present epidemiologic data, may make the cancer risk more 


I 

Iá 

! 
i 

" 

SIXTH BERKELEY SYMPOSIUM: OOFMAN AND TAMPLIN

CARCINOOENESJS BY IONJZINO RADIATION 265 266 

than three times higher than our pessimistic estimates of 1969. Are there rational 
humans who will be able to understand setting an allowable radiation guide for 
population exposure which may provoke a public health hazard one-third the 
magnitude of the entire cancer problem? We can only hope that the lessons of 
the radiation story will lead to a radical change in human approach to the 
questions of environ mental poUutants. 

Statisticians and epidemiologists, of course, are inclined to look forward to 
doing what statisticians and epidemiologists are professionaUy prepared to do. 
Unfortunately, this is true also about pbyaiciata, chemists, and. engineers. 

The purpose of this Symposium implies that, for the host olpotential pollutants 
now being introduced into our environment, enough epidemiologio evidence 
will, in the course of time, accu.mlate so that the statisticians ~d epidemiologists 
can do their thing. Thia means that the statisticians and epidemiologists 
have capitulated in toto to the dictum that progreaa means we must expose 
humans to by product poisons of industry in the future as we have in the past. 
And then the effects will be studied. If our radiation experience is any guide at 
all concerning the time scale over which we wiU learn the effect of our folly, and 
there is every reaeon to believe for carcinogenesis or genetic injury that the time 
scales will be similar, then the chances for humane surviv4tg this approach are 
slim indeed. 

We think it might have been more important if this gathering of statisticians 
and epidemiologists had met instead to lend their talents and wiedom to a concerted 
human effort to work toward to total recycling economy, in which 
easentiaUy zero polJution ie the objective instead of the building up of a reservoir 
of epidemiologic evidence of the effects of pollutants on humans. Indeed, such a 
thrust might even lead to the revolutionary idea of "Why do some of these 
nonsensical activities labelled 'Progrese' at all?" 

a 
6. Summary 

Ionizing radiation ie a potent leukemogen and carcinogen. The demand for 
epidemiologic evidence of human injury has resulted in a belated appreciation of 
the true magnitude of the serious carcinogenic hazard of population exposure to 
radiation. Even now, a quarter of century into the evaluation of the epidemiologic 
evidence, certain parameters of crucial chara~ter remain indeterminate. 
Should these parameters turn out to have unfavorable values, the seriousness of 
the hazard may truly be even larger than recent pessimistic estimates. We 
question, therefore, the wisdom of epidemiologic studies of human exposure for 
new potential carcinogens being introduced iinto our environment. 

Refined estimates presented here suggest that our earlier estimate of 32,000 
extra cancer deaths per year for exposure to the still permissible 0.17 rads per 
year (average for U.S. population from the "peaceful" atom) are not at an 
conservative. The true cancer risk may be closer to 100,000 extra deaths per 

year, representing a 30 per cent increnBe over the current spontaneous ca11cer 
mortality. Fol'tunately, atomic energy programs have not yet progressed to a 
point where suchá allowable exposure are being experienced. 

The National Council on Radiation Protection has recently stated that the 
current standards for radiation exposure are satisfactory (1971). We would not 
for one moment challenge the fact that the exposure standards are satisfactory 
to the membership of the National Council on Radiation Protection any more 
than we would challenge the concept that possession of 10,000 nuclear missiles ie 
satisfactory for the Department of Defense. What escapes our understanding, 
however, is how ~ne might go about evaluating the quantitative nature of the 
nebulous relationship between the interests of the membership of the NCRP and 
the public's interest in good health. 

Medical uses of X-rays presently are a major source of population exposure 
and a.re undoubtedly responsible for a significant pa.rt of our currently experienced 
cancer mortality rate . Morgan's suggestions for feasible reduction in 
medical X-ray exposure, without loss of medical diagnostic information, deserve 
immediate action [25). 

Natural radiation, while in large part not directly witl1in our control, is 
comparable in responsibility to medical X-rays in the quantitative fraction of 
cancer mortality rate currently being experienced. No rational ~Mis exists for 
the frequently heard suggestion that natural radiation can be used as a benchmark 
for estimation of "safe" exposures. Natural radiation must be estimated as 
poaaibly responsible'for -t11:kinga toll of several tens of thousands of lives annually 
by premature cancer and leukemia in the USA alone. Here again we must agree á 
with Morgan, that man may decide to look carefully at the radioactivity of 
certain "natural" building materials before using them for home construction. 

Life expectancy loaa experienced by those who will become the victims of 
allowablepopulation radiation exposure will average more than 13 years. The 
assertions of "only a few days of loss of life" are arrived at by the absurd and 
dangerous practice of distribution of the man-years lost in life expectancy into 
the larger group of nonvictims of radiation carcinogenesis. 

Epidemiologic investigations are extremely interesting and carry, for the 
investigators, the thrill experienced in solving murder mysteries and other 

; ..,

challenging problems. We have extreme doubt that the planning of appropriate 

epidemiologic investigations for future environmental poHutants is likely to be 

any real contribution to the public health. There has to be a more rational 

approach to the question of potential environmental carcinogens-like not 

introduci11g them into the environment at all. 

REFERENCES 

[I] 
G. W. BEEDE, H. KATO, and C. E. LAND, "Mortality and radiation dose, atomic bomb 
survivors, 1950-1060," prosentation at the IVth l~!,en111ii1>1~lConsrest1 pf Jl11(li11tion 
Jlcs!l~ri:h,~vi~n, fr,mi:c, J.11!l~July 4, 1\170, 

CARCINOGENESIS DY IONIZING RADIATION 267 

{2) R . E . CARROLL, W . HADDON, J11., V. H . HANDY, and E . E . WEEDEN, S11., "Thyroid 
cancer: cohort analysis of increasing incidence in New York State , 104.1-1002," J. Nat. 
Cancer Imt., Vol. 33 (1964), pp. 277-283. 

(3) 
W. M. CouRT-BaowN and R. Dou, "Mortality from cancer and oth er causes after radiotherapy 
for ankylosiug epondylitis," Brit. Med. J., Vol. 2 (1005), pp. 1327-1332. 
[4] 
0. W. DOLPHIN and I. 8. Ev.11, "Some aspects of the radiological protection and dosimetry 
of the gastrointestinal tract," Ga,troinleatinal RadiationInjurJI (edited by M . F. Sullivan), 
Ameterdam, Exoerpta Medica Foundation, 1068, pp. 465-474. 
(6) 
FEDERAL RADIA'l10N CouNcn ., "Staff report no. 1. Background material for tl1e development 
of radiation protection standards," Washington, D.C. , 1060, Part V, pp. 20-30 . 
[6) 
M. P. F1NU:L, D. 0. D1sK1s, and P . D. J1NKINB, ''Toxicity of .. ro.dium-226 in mioo," 
Radiation-Induced Cancer (Proceedings of a Symposium, Athene, Gre'ece, 28 April-2 May, 
1009, Organized by International Atomic Energy Agency in Collaboration with the World 
H ealth Organization), Vienna, Austria: al,o International Atomic Energy Agency, 1000, 
pp. 300-301 . 

[7] J. W. GorMAN and A. R. TAMPLIN, "Low dose radiation and cancer," IEEE Tran,. Nw:. 
Sci., Vol. NS-17 (1070), pp . 1-9. 
(8) J 
W . GOFMAN and A. R . TAMPLIN, á"Studies of radium exposed humans II : further 
refutation of the R . D. Evane' claim that the linear, non threshold model of human radi&tion 
carcinogenesis Is incorrect," testimony (on Bill S3042) presented before the Subcommittee 
on Air and Water Pollution, U.S. Senate, 01st Congress, 1970, pp. 320-350 . 
[9) J . W. Goll'MAN and A. R. TAMPLIN, "Federal 
radiation council guidelines for radiation 
exposure of tl18population-at-large-prote ction or die11Bter7/' Underground Uau of NuclMr 
Energy, Part 1 (Hearings before the Subcommittee on Air and Water Pollution, 

U.S. Senate, 91st Congress, áNovember 18, 1000), WMli.ington, D.C., U.S. Government 
Printi ng Office, 1970, pp . 68-73. 
(10] J. W. GOFMAN and A. R. TAMPLIN, "A proposal for at least a ten-fold reduction in FRC 
guidelines for tadiation oxpoeure to the population-at-large: supportive evidence," ibid., 
1070, pp . 319-326 . 

[11) J. W. Gol'MAN and A. R. TAlll'LIN, "Nuclear energy and the publio health," Ntvad.a 
Engin. , Vol. 0 (1970), pp. 1-16. 

[12) J. W. Goli'MAN and A. R. TAMPLIN, "The question of safe radiation thresholds for alpha 
emitting bone seekers in man," I/Mith Ph11a.,Vol. 21 (1971), p. 47. 

{13] J. W. GOFMAN, J. D. GOFMAN, A. R. TAMPLIN, and E. Kov1cn, "Radiation as an environmental 
hazard," present.ation at the 1971 Symposium on Fundam ental Cancer Research, 
The University of Texas, M. D. Anderson Hospital and T umor Institute, Houston, 
Texas, March 3, 1971, in pre911. 

[14] A. GaENDON, "Radiation 
protection standards," in Enuiromrnmtal Ejfecll of Producing 
Electric ápower, hearings before the Joint Committee on Atomic Energy, 01st Congress, 
2nd Session, January 27-February 20, 1970, Part 2, Vol. II, p. 2371. 
[15] 
L. D. HAMILTON, "Biological elgnificanoo of environm ental ro.diation: calculation o( the 
risk," presentation at the 1971 Spring Meeting or the American Physical Socioty, Washington, 
D.C., April 29, 1971. 
[16) L. H. HEMPBLIIANN, "Riek of thyroid neoplllBms after irradiati on in childhood," Science, 
Vol. 160 (1008), pp. 159-16 3. . 
{17) INTERNATIONAL ColtlllSSION ON RADIOLOOICAL PROTECTION, Publication No . 8, lladialion 
Protection: TM Evaluation of Riska from [ladiation, Oxford, Pergamon Press, 1006, 
Table 15, p. 56. 

[18) INTERNATIONAL ON RADIOLOGICAL roN, Publication No. 14, RadioCoMM18SION 
Pn<>TECTaemitiuity and Spatial Diatribution of Dose, Oxford, Pergamon Prcs.<1

, 11169, Appendix III, 
pp. 66-100 . 
[IO] S. JABLON and J. L. DELBKY, "Iladiati on-induccd canL'Cr in atomic bomb survivorA," 
present ation at tho Xth International Cancer Congres.~, Houston, Tcxna, May 1970. 

268 SIXTH BERKELEY SYMPOSIUM : GOFMAN AND TAMPLIN 

[20) 
8. JABLON, J. L. BEIHY, K. TAOHlltAWA, and A. BTuia, "Cancer in Japanese expoeed aa 
children to atomic bombs," Lancet,No. 7700 (1971), pp. 927-031. 

[21) E . B . LEwm, "Ionizing radiation and tumor," Gemlic Concept, and Neopla,ia (23rd 
Annual Symposium on Fundamental Cancer Research, 1060, University of Texas, M. D. 
Anderson Hospital and Tumor Institute, Houston , Texas) , Baltimore, Williama and 
Wilkine Co., 1970, pp. 67-73. 

[22] 
B. MAcMABON, "Pre-natal x-ray exposure and childhood cancer, " J. Nat. Conca ln,l., 
Vol. 28 (1002), pp. 1173-1191. 
(23] 
I. MAcK:sNzu:, "Breast cancer following multiple fluoroeooples," Brit . J . Conur, Vol. 19 
(1066), pp. HI. 


[24] R. B. Mot.11, "Radiation elfecta In man : current views and prospects," IleallA Ph111., 
Vol. 20 (1971), pp . 485-400. 
[25) K. Z. MoaoAN, "Never do harm," Emiironmmt, Vol. 13 (1971), pp. 28-38; alao NCRP 
Report 30, "Basic radiation protecLion criteria," published by NaLional Council on 
Radiation Protection and Measurements, Washington, D.C ., 1971, p. 97. 

(26] E . E. PocmN, "Somat ic riaka-thyroid 
carcinoma," Internal. Comm. Rad. Prolu., Publication 
8, Oxford, Pergamon Prees, 1906, p. 9. 
{27] L. SAGAN, "A positive word for nuclear power," Thia World, section of tll8 8¨ Framiaco 
Examiner and C/aronic~,January 10, 1971. 
[28) C. J. SrmLLAilABOJ:11,

V. P. BoND, E . P . CaoNJUTII, and 0 . E. APONTE, "RelatioD11hip of 
doee of total-body MCoradiation to incidence of mammary neoplaeia in female rate," 
Radiation-InducedCancer(Proceedings of a Symposium, Athens, Greece, 28 April -2 May , 
1960. Organised by International Atomio Energy Agency in Collaboration with the World 
Health Organization), Vienna, International Atomio Energy Agency, 1069, pp. 161-172. 
[29] 
H. SnJ:88 and C. W. Mna, "Bone cancen induced by "'Ra (ThX) in children and 
adults," IIMIIAPh111.,Vol. 19 (1970), pp. 713-729 . 
[30) A. Sn:WAIIT and 0. W. KNIIALII, "Changes in tll8 cancer risk associated with obstetric 
radi ography," Laiitd,No. 7632 (1968), pp . l(M-107. 
[31) A. Sn:wABT and 
0. w.' KNJ:ALII, "Radiation dose effects in relation to obstetric x-raye 
and childhood cancers," Lcmcel,No. 7668 (1970), pp. 1185-1188. á 

(32) A. R . TAMl'UM and J. W . GonaAN, "Biological effects or radiation," 'Populalion Control' 
TllroughNuckar Pollution, Chicago, Nelson-Hall Co.,1970, pp. 7-27. 
(33) A. C. UPTON, T . T . OD:sLI., Jn ., and E . P . 8N1rnN, 
"Inftuence of age at time of irradiaiion 
on induction of leukemia and ovarian tumors in RF mice," Proc. Boe. Bzper. Biol. Mt.d., 
Vol. 104 (1960), pp . 769-772 . 
(34) A. C . UPTON1 "Comparative obsenratio1111on radiation carcinog enesis in man and animal," 
Carcinogene,i,, a BrtXlll Critiqu, (20th Annual Symposium on Fundamental Cancer Research, 
1066, University of Texu, M. D. Anden,on Hospital and Tumor Institute , Houeton, 
Texas), Baltimore , Williams and Wilkins Co., 1007, pp . 63H}76. 
[36] A. C. UPTON, R. G, ALLEN, R. C. DnoWN, N. K. CLAPP, J. W. CoN1tLIN, G. E. Cosoaov., 
E. B. DARDIIN, J11., M.A. KABTIINBAUM, 
T . T . ODBLL, Ja., L. J . BanaANo, R. L. TYNDALL, 
and B. E . W ALBUJl<l, Ja ., "Quantitative experimental study of low-level radiaLion 
carcinogenesis," Radiation-Induced Cancer, Vienna, International Atomic Energy Agency, 
1009, pp. 425-438 . 
[36] O. J. B1zzozaao, 
Ja ., K. G. JoBNBON, and A. Crocco, "Radiation-related leukemia in 
Hiroshima and Nagasaki, 1946--64. I," New Engl.and Journal of Mt.dicina,Vol. 274 (1006), 
pp. 1006-1101. 
[37) R. D. EvANII, A. T. Kun, R. J. KOLIINltOW, W. R. NaAL, and M. M. SHANAHAN, 
"Radiogenic tumors in radium and mSBOtborium cases studied at M.I .T .," Ihl.a11edEjfecll 
of Bone-Sukiflf Rodionw:lidu (edited by C. W. Maye), Salt Lake City, University of Utah 
Press, 1969, pp . 167-194. 


I

(. . 

,
. 


1 
CARCINOGENESIS BY IONIZING RADIATION 269 

Discussion 

Question: David L . Levin, National Cancer lnsWute 

In regard to the cancer/leukemia ratio and the extrapolation of risk down to 
low levels of irradiation, we have also looked at the Court-Brown and Doll data 
on spondylitics and the Hiroshima-Nagasaki data . We have found that the 
cancer/leukemia ratio, when adjusted for difference in dose to the target organs 
is indeed higher than often quoted. Aleo, we have made calculations on the 
lifetime risk of developing cancer from irradiation levels of 170 mrade based 
upon extrapolations from high dose situations (X-ray therapJC and atom bomb) 
to low dose situations. Using different methodology than tha~á described today, 
we computed risks to be of the same general level as those shown by Dr. Gofman. 
W c arc now continuing our calculations trying to improve our (lstimate based 
upon demographic studies. 

Reply: J. Gofman 

We are, of course, pleased to hear that Dr. Levin and his colleagues arrive at 
the same general conclusions as we do. This confirms our statement that the so 
called "radiation controversy" concerning cancer risk is over. Dr~ Levin is 
certainly correct that there are other calculation meth~ologies . We have tried 
and presented several approaches. It turns out that, since áwe .11.llhave the same 
epidemiologic data at our disposal, the rules of arithmetic and algebra require 
that the same general conclusions will be reached. 

II

Often the issue is confused by those who don't really disagree with the calcuI. 
lation of the cancer hazard. It is just that some people don't worry about adding '.I 

, I 

100,000 extra deaths per year from cancer. Thie is more a problem of central 
nervous system physiology and pathology than it is one of epidemiology. i. 

Question: E. B. Hook, Birth Defects lmtitute, Albany Medical College 

How do the calculations of a presumed increase in tumor incidence following 
an increase of 0.17 rads per year compare with the known "background" dose of 
radiation and known tumor incidence? 

Reply: J. Gofman 

As discussed in the presentation, we would calculate the tumor incidence due 
to natural background radiation would stand to that for 0.17 rads/year as does ' 

0.125 rads per year to 0.17 rads/year. No one has experimentally or by an 
epidemiological study segregated the cancer cases due to natuml background 
radiation from so-called "spontaneous" cancer cases. There is every reason to ! 
believe that background radiation accounts for its expected share of the total 
observed cancer mortality rate. 
Question: E. Tompkins, Human Studies Branch, Environmental Protection Agency 

Would you please explain your statement that there is no reason to believe 
that the number of films (as reported by Stewart and Kneale) is associated with 
disease of the mother. 

270 
SIXTH .BERKELEY SYMPOSIUM : GOFMAN A"ND TAMPLIN 

How do you explain the difference in distribution of number of films by 
trimester? 

Reply: J. Gofman 

Let us consider first those women who received X-rays specifica.lly for diagnostic 
obstetric radiography. This group constituted the bulk of the Stewa.rtKneal 
sample. Within this group of women selected to receive diagnostic obstetric 
radiography, I know of no reason to believe the radiographer would take a 
number of films, to achieve pelvic dimension measurements, that would be 
correlated with pre-existing disease in the mother. 

With respect to the difference in distribution of film number by trimester, this 
is not at all surprising. Films taken before the third trimester almost certainly 
were taken for an indication other than diagnostic pelvimetry. It is commonplace 
to find that the number of films required for one particular diagnostic purpose á 
may differ from that required for another. 

Question: 
Alexander Grendon, Donner Laboratory, University of California, 
Berke'ley 

You answered a previous question by saying that the component of the 
natural cancer incidence due to background radiation is a.lso multiplied by added 
radiation. Doesn't this constitute a square law relation rather than the linear 
relation postulated by you? 

Reply: J. Gofman á. 

I have said nothing at all that suggests a square law relationship. I believe the 
difficulty, Mr. Grendon, resides in your interpretation of what I said. I said that 
when the cancer mortality increment per rad is expressed as a percentage of the 
"epontaneouri" cancer mortality rate, the "spontaneous" rate incorporates the .. 
contribution of natural plus medical radiation. Such expreseion of the increment 
is simply a statement of observationin a particular population sample. For 
purposes of d~termining the rea.l per cent increment per rad over nonradiation 
"spontaneous" cancer mortality rate, the appropriate procedure would be to 
subtract out the medical plus natural radiation contribution to the observed 
"spontaneous" ra~ . I believe this is adequately clarified in the text. I had no 
intention of suggesting that additional radiation multiplies the effect of background 
radiation. 

Question: Unnamed discussant 

..

Could you comment on the quality of the Hiroshima data on leukemia as 
compared to that of Dr. Stewart's data and that of Dr. MacMahon? Is it not 
true that the effects of trauma, dietary and medical care effects on the population 
exposed to the bomb radiation as compared to the controls and the subsequent 
heavy mortality by abortion and congenital defects among tlie exposed fetuses 
and infants make these results far more questionable than the studies of infants 
exposed in peace time to diagnostic X-rays? 


CARCINOGENESIS BY IONIZING RADIATION 271 

Reply: J. Gof111an 

Yes, I prefer the Stewart and MacMahon data to the Hiroshima-Nagasaki 
data on leukemia in childhood following in utero radiation both for the reasons 
you cited and for other reasons. 

Perhaps we should clarify for the audience what the iBBuc is here. Recently 
Jablon [48] pub1ished data on subjects irradiated in ulero in the Japanese atom 
bombings. Jablon indicated that for the exposure received by this population 
sample, the observed occurrence of childhood cancer and leukemia was far less 
than that to be expected from the Stewart or MacMahon data. I have several 
very serious reservations about the Japanese data. "<


(1) Tbe bulk of the man-rads of exposure in the Japanese sample arises from 
those cases with very high exposures (100-200 rads) . We are all familiar with the 
observations in experimental studies that , while the carcinogenic dc;>Be 
response 
relationship is linear over a large range, one does arrive at high doses where. the 
carcinogenic response levels out and then drops drastica1ly. Many refer to this 
as "the other side of the dose response curve." From the Stewart evidence, it 
appears that the doubling dose for in utero induction of leukemia and cancer is 
of the order of one rad . This would mean that the bulk of the man-rad exposure 
of the Japanese sample occurred in infants receiving 60.to 200 doubling doses. 
There is every reason to suspect that this would place them . :well over on the 
"other side of the dose response curve." Hence this factor alone should lead to a 

Jess.er carcinogenic/leukogenic response in the Japanese sample than anticipated 
based upon the Stewart evidence. 

(2) The Japanese sample of infant~ exposed in utero was characterized by an 
enormous mortality during the first year of life. No data are provided concerning 
the nature of these mortalities . If the risk of subsequent mortality from cancer or 
leukemia is correlated with risk of mortality in the first year of life (and we know 
nothing of the existence or nature of such relationships), it is conceivable that the 
Japanese sample was depleted, by enormous first year mortality, of the most 
likely candidates for subsequent cancer or leukemia. The discussant, for example, 
pointed out the iBBues of dietary and medical care effects in the Japanese 
sample. This is certainly appropriate, and it is entirely poBBible that those with 
enough radiation injury to develop leukemia later may be especially susceptible 
to earlier death from malnutrition, for example. This effect would not have 
occurred in the Stewart or MacMahon population samples. 
(3) The Jablon data on children exposed between O and 9 years of age at the 
,
time of bombing show a marked increase in carcinogenic and leukemogenic risk. 
It would be very surprising that the carcinogenic/Jeukemogenic risk in the 
Japanese in utero cases would be absent. 

(4) I would take serious iBBue with Dr. Jolin Totter's statement (in diecUBBing 
Dr. Sternglass' paper this morning) that the Japanese data represent an unbiassed 
sample compared with the Stewart data. For the obvious reasons listed 
above, I would draw the opposite conclusion to that of Dr. Totter. In so doing, 
I agree with the discussant who raised this question. 
272 SIXTH BERKELEY SYMPOSIUM: OOFMAN AND TAMPLIN 

Question: Prem S. Puri, Department of Statistics, Purdue University 

You presented a table showing the mortality figures by age for individuals 
who were exposed to radiation at Hiroshima, Japan. Do you have some information 
on the possible radiation effects on the mortality of the descendants in 
the next generation of these individuals? 

Reply: J. Gofman 

I believe it is far too early to be able to comment on morta1ity effects among the 
descendants of the survivors of the atom bombing in Japan. 
Question: J. Martin Brown, Stanford Medical Center 

Is it not true that the data of Stewart, I{neale and of MacMahon show that 
the radiation induced excess over the spontaneous rate has disappeared eight to 
10 years after the pelvic X-rays of the fetus? Won't this make a big difference to 
your calculation of the radiation induced incidence of cancer in view of its 
marked dependence on the length of the plateau period? 

Reply: J. Gofman 

I do not believe that either the Stewart-Kncale or the MacMahon data really 
allow for one to draw the conclusion that the radiation induced excess disappears 
at eight to ten years or at any other time, for that matter. I do understand how 
that impreBBion can have arisen, however. The overall mortality curves (spontaneous) 
for cancer show a peak at about the middle of the first decade of life 
and then a decline to a relativelylow level until the early 20's of age when the 
upturn begins, due-.t.Q the appearance of the various malignancies of adult life. 
In the Stewart-Kneale and MacMahon children the radiation induces the same 
kinds of cancers that occur spontaneously in the children. Since there is about a 

' 
50 per cent increase due to obstetric radiography in such cancers and leukemias 
over their spontaneous occurrence, it is not surprising that the radiation induced 
cases would show a peaking just as do the spontaneous cases. The decline from 
the á peak among the radiation induced cases gives the impressum that the á ¥ 
radiation induced excess is disappearing. This, I would consider, is totally 

illusory. á
. 
If one could study this pbenonmenon into later years (adulthood), radiation 
induced cases might rise in incidence as the spontaneous incidence rises, and, for 
all we know, this effect might persist throughout the lifetime of the exposed 
population. A very different study from those of Stewart-Kneale or MacMahon 
would be required to address this issue. Incidentally, for children irradiated 
between O and 9 years of age, the radiation induced cancers are rising in incidence 
20 years post irradiation in the Japanese survivors, as would be expected from 
the rising spontaneous incidence with age increase. 
But Jet us presume the in ulero effect did decrease after 8 or 10 years. From 
Table V, the average increase in cancer mortality rate is calculated to be approximately 
34 per cent of the spontaneous mortality rate. If the O per cent due to in 
utero irradiation is subtracted, the radiation induced rate would be approximately 
28 per cent of the spontaneous mortality rate. 



CARCINOGENESIS DY IONIZING RADIATION 273 

Question: B. G. Greenberg, School of Public Health, U1iiversity of North. Carolina, 
Chapel II ill 

I wonder, Dr. Gofman, if the ordinate in your graph, mortality from, say, 
leukemia, hos been adjusted for other deaths on a competitive risk basis . If not, 
the graph seems to imply that the best protection for a person who has been 
irradiated is to expose him to an additional overwhelming dose in order to bring 
his mortality risk down. 

MORTALITY 

FROM 

LEUKEMIA 

DOSAGE ( Rods) 

Reply : J. Gofman 

Of course there would be an increasing mortality with increasing radiation 
dose for a whole variety of 1áadiation induced causes of death. ,\mong the survivors 
of the increased dose, the leukemia risk would still be a lower fraction of the 
total irradiated population than at lower doses. The competitive risk of other 
mortalities must be considered . If we wish to focus on keeping just leukemia 
mortality down, I would suggest that execution of the entire irradiated population 
sample by a firing squad would be even more effective than the supplemental 
radiation suggested by Dr. Greenberg. (See Figure 2, Curve A.) 

Question: Thomas F. Budinger, Donner Laboratory, University of California, 
Berkeley 

Dose rate effect has been shown for carcinogenesis (45), survival ((42], (41]), 
and genetic mutations (44). Exposure in the range of 170 mrem distributed 
through one year is more than 101 times less dose rate than the dose rate of 
exposures from which Dr . Gofman draws his conclusions. Therefore, from a 
biophysical standpoint it is difficult to accept his figures as applicable to any 
anticipated population exposure. These studies as well as curvilinear dose 
response relationships in radiation carcinogenesis (for example, see (46],) suggest 
some repair mechanism is present. A reasonable assumption is that DNA and 
chromosomes are involved in somatic and genetic mutation. We now have 
ample biophysical data to show DNA breaks are repaired very efficiently by at 
least two cellular mechanisms (for example, see (43)). In fact, the histol'y of 
radiation damage and successful or unsuccessful repair is readily seen by follow


274 SIXTH BERKELEY SYMPOSIUM: GOFMAN AND TAMPLIN 

ing chromosome aberrations [40], [38]. Radiation insult repair is dependent on 
linear energy transfer [39] and for high LET exposures we would expect, and 
indeed find, linear .dose effect response and little or no dose rate effect on DNA 

repair, chromosome aberrations, or carcinogenesis. Thus, only for high LET 
irradiation is a linear extrapolation valid . If we all were being exposed to neutrons 
or penetrating á high Z charged particles, I would agree with Dr . Gofman's 
figures. Anticipated radiation exposures are low LET at low dose rates; thus, I 
do not see a public health threat as great as the threat of trace elements and 
other man-made contaminants. 

REFERENCES 

(38) W. C. DBWllY and R. M. RoMPBBl!IY, .''Restitution of radiation -induced chromoeomal 
damage in Chinese hamster cellsrelated to the cell's life cycle, Exper. Cell Ru., Vol. 35 
(1964), p. 262. 

(39) 
M. M. ELl(JND, Cu". Top. Rad. Ru. Quart.,Vol. 7 (1970}, p. I. 
(40) H.J. 
Ev.A.NB, "Repair &nd recovery from chromosome damage after fractionated X-ray 
dosage," OmeticA,pect, of Radioaenntivit11: M edlaniama of Repair, IAEA, 1006, pp. 31-48. 
[41) D. GaABN, "Biological effects of protracted low dOBe radiation exposure of man and 
animals," Lale EJTedl of Radiation, London, Taylor & Francis, LTD., 1970, p. 101. 
[42) E. J. HALL, R. J. BBRRT, J. 8. Bsoroan, in Dou Rau in Mammalian Biolog11,AEC 
Conference Report Conf.-680410 (1968), pp. 15.1-15.20. 
[43) R. B. PAINTJ:R, Curr. Top. Rad. Re,. Quart., Vol. 7 (1970), p. 45. 
[44) W. L . RoBBBU., Nucle<>nicl,Vol. 23, p. 63. 

(45) 
A. C. UPTON, Radiation lnjuf'JI, Chicago, University of Chicago Prel!8, 1960, p. 79. á 
(46) --
, 
''The ~reJ>Onae relation in radiation-induced cancer," Canur Rea., Vol. 21 
(1961}, p. 717. 
Reply: J. Gofman 

I believe I have disposed of Dr. Budinger's major concerns in the body of this 
conimunication. A few of the specifics raised by Dr. Budinger are eitlier non 
aequitura or reflect such unsatisfactory' public health principles that they deserve 
comment here. 

(a) Dr. Budinger states that dose rate effect for carcinogenesis has been 
shown by Upton, [45]. This is simply not true. No studies of Upton are free of 
the criticism amply. discussed in the text that the chronic exposures extend into 
b.ter life of the exp~rimental animal where sensitivity to carcinogenesis drops . 
Indeed, the studies of Upton, which Dr. Budinger quotes, are provably suspect 
on these grounds based upon Upton's own data. (This issue is thoroughly 
discussed . in one of our earlier papers, Gofman and Tamplin [47] .) 
(b) Dr. Budinger comments that "exposure in tbe range of 170 mrem distributed 
throughout one year is more than 101 times less dose rate than the dose 
rate of exposures from which Dr. Gofman draws his conclusions. Therefore, from 
a biophysical standpoint it is difficult to accept his figures as applicable to any 
anticipated population exposure." 
There may be BOUnd grounds upon which our figures are unacceptable, but I 
truly am surprised that Dr. Budinger states "from a biophysical standpoint" it 

.:' 

, II 

I. 
j 

CARCINOGENESIS BY IONIZING RADIATION 275 

is difficult to accept them. It becomes necessary to point out a few biophysical 
principles to Dr. Budinger. 

First, the existence of a linear relationship between carcinogenic response and 
dose (see references in text) implies, biophysically,a single event phenomenon 
for radiation induction of cancer . This biophysicalpoint should remove most of 

Dr . Dudinger's concern. 

Second, let us consider Dr. Budinger's factor of 106 in dose ra~e. Especially, let 
us consider the biophysics involved. Dr. Dudinger would appear to be considering 
that 170 mrem delivered over the course of a year by environ .mental pollutants 
delivering low LET radiation to be oozed into tiSBue at a sfow, smooth rate. 
Unfortunately, the biophysical reality is considerably different from the picture 
of Dr. Budinger. 

Let us consider, biophysically, the delivery of 170 mrem to cells over one year 

versus delivery in a fraction of a second, as from an X-ray machine . We shall 
find that Dr. Dudinger's factor of 106 melts away with great speed. . 

At 1 MEV or less, X-rays and gamma rays deliver energy to tiBBues through 
their photoelectric or Compton conversion to electrons. Therefore, the entire 
group of low LET radiations (X-rays, gamma rays, fl particles) can be covered 
by consideration of fl particle (electron) interaction with ácel~, 

j 

1 rad= 100 ergs/gram= 6.25 X 107 MEV /gram. 

Let us consider 1 MEV fl particles as representative. 
This means 1 rad represents 6.25 X 107 fJparticles delivering their energy per 

gram of tissue. 
The range in tissue for 1 MEV fl particles is approximately 4000 microns. 
For a.cell of 20 micron diameter, a 1 MEV fJparticle traverses 200 cells, onthe 

average. Therefore, 6.25 X 107 fl particles traverse 1.25 X 1010 cells. 
For cells of approximately 20 micron diameter, volume is approximately 

4 X 101.1, and 1 gram of tissue represents approximately 1011.1¥ 
So there arc 1012/(4 X 101) '.::::'.2.5 X 109 cells per gram of tissue cells. 
If one rad represents traversal of 1.25 X 1010 cells, then each cell is traversed 

(1.25 X 1010)/(2.5 X 108) = 50 times. 
Therefore for 170 mrads, each cell is traversed (0.17)(50) = 8.5 times on the 
average. 
Each traversal of a.cell by a single beta particle occurs in a time frame of a 
fraction of a.second. Now, if we are to compare equivalent doses, 170 mrads, 
delivered instantaneously versus over the span of one year, for the same kind of 
radiation, for example, 1 MEV fJparticle (the same would be true for X-rays or 
gamma rays), it follows, obviously that the exact same number of fJparticles, on 
the average, must traverse the cells whether instantaneously or over the span of 
a year . For the instantaneous delivery, let us assume all the fJparticles (between 
8 and 9 of them, average 8.5) all are delivered together exciting effects over a 
short interval, t (time frame, seconds or less). For the one year delivery, there 
will still be 8.5 fJparticles delivered per cell, ,:i.nd eachfJparticle will exert effects 

' 

I

'! 

'. 
' 
I 276 SIXTH BERKELEY SYMPOSIUM: OOFMAN AND TAMPLIN 

in precisely the same short time interval, t, except that the individual events will 
be separated from each other by a little over a month, on the average. Therefore, 
the "slow" delivery, so far as cellular events are concerned occupies (8.5 t) 
instead of t. Therefore, the maximum realdifference in rate of delivery of energy 
to the cell is 8.5 fold. Thus, Dr. Budinger's "factor of 101," drops to a factor of 
8.5, a drop of more than 100,000 fold. The interjection of the iBBue of about a 
month between events is a red herring. I do not believe that, for carcinogenic 
events occurring 5 to 25 years later, Dr . Budinger would argue that irradiation 
in September of a particular year would be much different from irradiation in 
May. 

In all likelihood, even the factor of 8.5 fold is musory. For a single cell, not all 
of the 8 or 9 events occur in the same region of the cell's volume. For a particular 
region within a.cell receiving irradiation, 170 mrads may, therefore, represent 
the effect of only one(Jparticle traversal, notthe traversal of 8 or 9. So the actual 
time frame of events for the cellular level may be precisely the same at a particular 
sensitive site whether all the 170 mrads is delivered instantaneously or 
spread over one year. This would melt Dr. Budinger's factor of 101 down to a 
factor of unity, for the two regimes of irradiation . But we needn't quibble as to 
whether Dr. Budinger's concern is irrelevant by a factor of 100,000 or a factor of 
1,000,000. 

There are additional data, indeed available from Upton's work (A. C. Upton 
and G. E. Cosgrove, Jr. "Radiation-induced leukemia," ExperimentalLeukemia, 
(edited by M.A. Ricli) New York Appleton-Century Crofts, 1~8, pp. 131-158) 
which show conclusively, at least for thymic lymphoma in the mouse, that even 
at doses like 20 rads, there is no difference between "instantaneous" and "slow" 
delivery of radiation with respect to thymic lymphoma development. One of the 
good features of this experiment of Upton's is that the two regimes of irradiation, 
instantaneous and slow, were delivered at comparable age periods in the life 
span á of the mouse, a feature not controlled in the vast majority of acute versus 
chronio irradi~tions. What Upton and Cosgrove showed was that there was no 
difference for thymic lymphoma incidence whether 200 rads total was delivered 
as 10 exposures, each of 20 rads (7 to 25 rads/min), spaced 30 days apart or 
whether delivered á !t 0.5 millirad,s/minute; which consumed the same approximate 
overall time period (approximately 300 days). 

Using the same type of calculations as above, where 20 millirads represents 

i 

one ionizing event (50 events per rad), the Upton data would indicate that one 

event per 40 minutes gave no different result, for thymic lymphoma induction, 
from 1000 events in om minute (20 rads in one minute~ 1000 events). 

(c) Dr. Budinger brings up DNA repair and chromosome aberrations. These 
are very interesting phenomena that every knowledgeable scientist realillCS do 
exist. But neither Dr. Budinger nor anyone else has suggested any relevance of 
these phenomena for the question of radiation dose rate and carcinogenesis. If 
Dr. Budinger knows the events in radiation carcinogenesis well enough to assert. 
that DNA repair has anything to do with dose rate and cancer production, 

CARCINOGENESIS BY IONIZING RADIATION 277 

. (

urge him strongly to publish these findings. At present, without such evidence, 
the mere mention of phrases like "DNA repair," or like "DNA and chromosomes 
are involved in somatic mutation" is about as useful in asseB8ing tbe question of 
dose rate and carcinogenesis as are yesterday's stock market quotations. We 
wouldn't deny that the stock market quotations are interesting. ~ 

\á

(d) Lastly, we must respond to Dr. Budinger's statement, "Anticipated 
radiation exposures a.re low LET at low dose rates; thus, I do not see a public 
health threat as great as the threat of trace elements and other man-made 
contaminants." 
We have seen above th~t Dr. Budinger's factor or 10¥ in dose rate for low LET 
radiations m~lts away at least by 100,000 fold, when the biophysics is considered. 
But, whatever the magnitude or carcinogenic effect of ionizing radiation may be, 
it is difficult to understand the philosophy implied in Pr . Budinger's t1tatement 
that other poisons may be worse. They may very well be. Jf one unneceB8ary 
poison kills 10,000 people per year while another kills 30,000 ápeople per year, 
shall we exonerate the first unneceseary poison because it unhappily didn't 
reach the top of the heat seller list? 

REFERENC)j',S 

(47) J. W. GorMAN and A. R. TAMPLIN, ''The mechanism of radiation carcinogenesis (a) An 
explanation for the illusory effect of protraction of radiaLi~n in reducing carcinogenesis 
for low LET radiation, " Brwironmffllal Bffut. of Producing Blutric Power, hearinp of 
the Joint Committee on Atomic Energy, 91st CongreBB, 2nd Session 1970, Part 2, Vol. 
II, 1970, pp. 2008-2098. . 
(48) S. JABLON and H. KATO, "Childhood cancer in relation to prenatal ~posure to A-bomb 
radiation," Lance', Vol. 2 (1970), pp. 1000-1003. ¥1 
')