In1 J Rudrarwn Ondo~> B,ol Phv.\ Vol. I I. pp. 142 I- 1423 0360-3016/U $03.00 + .

OO
Pnnted 3” the U.S.A. All rights reserved. Copyright 0 1985 Pergamon Press Ltd.

??Editorial

BREAST CARCINOGENESIS: RISK OF RADIATION

SEYMOUR H. LEVITT, M.D.* AND JACK S. MANDEL, PH.D.?

University of Minnesota Hospitals, 420 Delaware St. Southeast, Box 197, Minneapolis, MN 55455

INTRODUCIION dose exposure. lo Finally, there are technical differences

The risk of radiation carcinogenesis in the opposite between the irradiation received by individuals involved
breast is a major concern for physicians and breast in most epidemiologic studies and the therapeutic irra-
cancer patients who choose to preserve the involved diation received by breast cancer patients. In the carci-
breast through conservation treatment, i.e., conservation nogenesis study, the quality of the irradiation, fraction-
surgery and radiation therapy. Although there are other ation of the radiation, overall time, fraction dose and
areas of criticism concerning this approach to the treat- total dose, are all different from the modern techniques,
ment of breast cancer, most of which are unfounded, equipment and fractionation of therapeutic radiation.
radiation carcinogenesis remains a major concern to All of these factors make it difficult, if not impossible,
patients, physicians, and epidemiologists. It is obvious to directly correlate the irradiation risk ascertained from
that a risk exists, but what the risk is, and how great, is the studies and modern radiotherapy. However, the
the question and the subject of this paper. major issue in evaluation of breast radiation carcinogen-
Much of the data relating to radiation carcinogenesis esis is the accuracy and objectivity with which the
is derived from: epidemiologic studies of atomic bomb studies have been done and the actual risk involved.
victims.2,7,8,9,’‘,‘3,‘4,16reports on patients with tuberculosis,
treated’ by hospitalization and repeatedly exposed to
METHODS AND MATERIALS
fluoroscopic procedures;3*4*‘2 a study of Swedish patients
treated with irradiation for cystic disease;’ and a recently Atomic bomb data
published long term follow-up study of a group of Among the problems in the atomic bomb studies are
patients with breast cancer, who did or did not receive the evaluation of dose, the epidemiologic methods and
irradiation as part of their cancer treatment.6 the analytic strategies. lo For example, the estimated
In analyzing the carcinogenic effect of irradiation on standard error of individual dose estimates for the LSS
the breast, the radiobiologic risks assumed from the (Life Span Study) sample were plus or minus 30%.’
studies must be evaluated first in order to determine the Inconsistent methods were used in rounding estimated
accuracy of the epidemiologic data and radiation dosage. exposure doses so that in lower dose exposure groups,
It is generally assumed from the carcinogenic studies the error rates were estimated to be fairly substantial.8
that radiation is carcinogenic at any dose rate. However, There is still difficulty in determining the quantity and
it is well-known that repair processes exist at the cellular quality of the irradiation to which the population was
level, and low dose rates are less effective at producing exposed.
cancer in animal species than higher dose rates. Despite Gathering of the data for atomic bomb studies was
this knowledge, it has been assumed and recommended mostly done by death certificates. In the study of
that the same linear relationship for all dose rates be McGregor et al. I3 in which biopsies were evaluated, only
used. Thus, in most epidemiologic studies no apparent 138 of the 23 1 cases were evaluated by the investigators,
account is taken of dose rate. In addition, there is 54 cases were diagnosed by other physicians, 16 were
evidence that the risk from high radiation dose, i.e., clinical descriptions and 23 were evaluated from death
over 500 rad, may be less per rad than that from low certificates. Dose estimates were available for only 80%

* Professor and Head, Department of Therapeutic Radiology. Acknowledgment--This work was funded in part by Public
t Associate Professor, Department of Epidemiology, School Health Service grant #CA-15548, awarded by the National
of Public Health. Cancer Institute, NIH, DHHS.
Reprint requests to: Seymour H. Levitt. Accepted for publication 5 December 1984.

1421
1422 Radiation Oncology 0 Biology 0 Physics July 1985. Volume I I. Number 7

(187 of 23 1) of the cases. Among the potential biases This study reviews a series of patients treated for post-
were the possibility that high dose cases were evaluated partum mastitis with orthovoltage irradiation, matched
more completely than low dose cases, loss of information to a non-irradiated group of patients.
through migration, and lack of all pertinent information Patients were treated with 175 to 200 KV X rays (no
concerning the patient. Also, a number of patients who filter). Theoretically, the radiation exposures were noted.
developed breast cancer following atomic bomb exposure However, the question of machine calibration-how
had exposures to radiation from other sources.‘4 Fur- often or how accurately this was done-was not ad-
thermore. it was difficult to judge an appropriate com- dressed. The radiation doses were reported in roentgens
parison group for the disease using all the known in air, and dose estimates in rad were determined by
predisposing factors, including child bearing data, family calculation.
data, etc. There were four groups of patients in the study: A
= treated patients; B = sisters of group A; C = non-
Flwroscopy dutu irradiated patients with mastitis; D = sisters of group C.
Two studies of fluoroscopically exposed patients with The three non-treated groups (B, C, D) were combined
tuberculosis treated by pneumothorax were retrospec- to produce a control group. There are a number of
tively done. For our purposes, they will be referred to methodological problems with this study. First, there
as the Newfoundland study and the Massachusetts was a loss of 30% of the control patients and a failure
study.3,4.‘2 to contact or determine information about sisters in 148
In the Newfoundland study, dosimetry evaluation was of the 380 control patients. There was no mention of
a major problem since there was no possibility of review of death certificates for the control patients or
accurately assessing the total x-irradiation received by their sisters (if an increased incidence of breast cancer
an individual patient. Although an attempt was made in this group was found, it would be helpful).
to measure the dose rates produced by the fluoroscopic There was insufficient information provided about
units used in the 1940s this was not completely satis- differences between groups A and C. For example, why
factory. There was variation in line voltage as well as were some patients treated and others not? Was this
other uncontrollable factors, so that accurate kilo-voltage because of social class or some other demographic
figures could not be obtained. There was no information factor(s)? It should be noted that the treated group was
about the filter type or whether one was used. Further- from Rochester, New York, and the control group came
more, there was no statement made as to the number from New York City. The demographic factors could
of patients who did not receive an artificial pneumo- bear on the incidence of breast cancer. There were
thorax, how many were alive, how long they lived, and major epidemiologic differences between the treated and
how many developed breast cancer. control group. One of the most important differences
The Massachusetts study, carried out by questionnaire was the significantly increased incidence of chronic
in 1974, evaluated patients subjected to repeated fluo- cystic mastitis in the treated group. This is a known
roscopy after pneumothorax in Massachusetts between predisposing factor to increased risk of breast cancer. A
1930 and 1954. final question is whether the women receiving irradiation
Analysis of the dosimetry in this study was compli- were followed more closely.
cated. Estimates of radiation exposure were derived by
a Monte Carlo radiation transport technique using data Patients treated with irradiation jiw q>stic disease
from medical records, interviews of physicians and A Swedish study’ reported on I, 115 women treated
patients, and exposure measurements on representative between 1927 and 1957 with ionizing irradiation for
fluoroscopes. Overall, there was much imprecision as- different non-neoplastic conditions of the breast (fi-
sociated with the retrospective determination of fluoro- broadenomatosis, acute and chronic mastitis). A critique
scopic procedures and associated radiation doses. of the design of this study is the lack of controls and
Also, there was a delay of 20 to 40 years in interviewing the fact that the patients were treated for diseases which
the patients. Of the 1,764 patients hospitalized, responses predispose to breast cancer.
were obtained from 917 (80%) of the patients still alive.
There was no data obtained on the 35% of the patients Incidence o/‘cancer in the contralateral breast
who had died. A recent report of one of the largest follow-up studies
It was noted that the interpretations could not be on patients treated with radiation and surgery for breast
made with any degree of certainty and that the results cancer reported a rate of second breast cancer three
must be regarded as generating hypotheses that required times greater than expected.6 However, a close exami-
confirmation by a larger series. nation of this data fails to substantiate the claims. When
the data are stratified by age, stage, and length of follow-
Putients treated.ftir postpartum mastitis with up, there is not a consistent pattern of elevated risk.
orthovoltage irradiation Evaluation of the majority of risk estimates, the 90%
One of the most important studies relating radiation confidence interval includes 1.0, thus indicating a non-
to breast carcinogenesis is the study by Shore et al.” significant result (see Tables 3 and 4 of ref. 6).
 Breast carcinogenesis 0 S. H. LEVITTAND J.S.MANDEL 1423

DISCUSSION that the highest excess risk is 6-fold from 1 to 400 rad,
approximately 36 cases of breast cancer per one million
In a recent paper, Feinstein and HorwitzS comment
women would result from exposure to radiation following
on some of the problems inherent in epidemiologic
a 10 year latent period. Thus, among the 20,000 cases
research. Many experimental principles were noted that
exposed to radiation therapy, less than one additional
could be employed but are often overlooked. These
case of breast cancer would result from exposure to
include: “verification of quality and accuracy in raw
radiation following a 10 year latent period, assuming
data, avoidance of major biases in comparison, vigilance
the worst risk. However, as we have demonstrated, this
in checking for methodologic errors, and maintenance
risk is almost certainly overestimated because of the
of a careful distinction between research data that gen-
epidemiologic and dosimetric problems inherent in the
erate hypotheses and the new data needed to test these
studies of radiation and carcinogenesis.
hypotheses”. The paper is extremely informative and
Numerous clinical studies fail to substantiate the
useful and sets up standards not necessarily met by the
worst case-risk for patients treated therapeutically with
major studies used to determine the radiation breast
carcinogenesis risk. modern techniques, fractionation, and equipment. We
can assume that the risk of radiation carcinogenesis to
We have analyzed the major studies and noted that
they suffer from one or more major methodological these patients is not nearly as great as might be assumed.
flaw: the absence of an appropriate comparison group, Obviously, if there is no benefit from a therapeutic
the greater diagnostic surveillance in the irradiated group, procedure, then any risk is excessive, no matter how
the over-interpretation of relative risks, and the inability negligible. Both the conventional method for the treat-
to separate the risk of breast cancer associated with the ment of breast cancer, the radical mastectomy, and the
original disease from the additional risk of irradiation. more conservative procedure, lumpectomy with irradia-
We note that in none of these studies has it been tion, have adverse consequences. The benefit, as mea-
possible to evaluate accurately the effect of the other sured by survival in numerous clinical studies, is ap-
known risk factors in breast cancer, including geographic proximately the same. The cosmetic consequences ob-
areas of residence, social class, age at first birth, history viously favor the conservative approach. The relative
of benign breast disease, family history of cancer, men- balance of the risks, consequences and benefits of these
strual history, etc. treatments must be considered by the patient and her
We conclude that the risk per rad is overestimated. physician. Our review leads us to believe that the
We have determined that the total dose in the opposite obvious benefits of the conservative approach far out-
breast in patients treated with radiation following con- weigh the relatively minimal radiation carcinogenesis
servation surgery is from 100 to 400 rad. If we assume risk.

REFERENCES

1 Baral, E., Larsson, L.-E., Mattson, B.: Breast cancer fol- weapon: A reanalysis of available data with recommended
lowing irradiation of the breast. Cancer 40: 2905-2910. values. Report ORNL-7M-5139. TN, Oak Ridge National
1977. Laboratory, 1976.
2. Beebe, Cl., Kato, H., Land, C.: Studies of mortality of A- 10. Levitt, S.H., Mandel, J.: Breast irradiation and future risk
bomb survivors. Radiat. Re.s. 75: 138-201, 1978. of carcinogenesis. Proceedings of the 17th Annual San
3. Boice, J., Monson, R.: Breast cancer in women after Francisco Symposium (February 27-28, 1982). Fron. Ra-
repeated fluoroscopic examinations of the chest. JNCI 59: diat. Ther. Oncol. 17: 134-142, 1982.
823-832, 1977. 11. Loewe, W., Mendelsohn, E.: Revised dose estimates at
4. Boice, J., Rosenstein, M., Trout, E.: Estimation of breast Hiroshima and Nagasaki. Health Phlx 41: 663-666, 198 I.
doses and breast cancer risk with repeated fluoroscopic 12. MacKenzie, I.: Breast cancer following multiple fluoros-
chest examinations of women with tuberculosis. Radiat. copies. Br. J. Cancer 19: l-8, 1965.
Rex 73: 373-390, 1978. 13. McGregor, D.H., Land, C.D., Choi, K., Tokuoda, S., Klu,
5. Feinstein, A.R., Horwitz, R.I.: Double standards, scientific P.L., Wakabayashi, T., Beebe, B.W.: Breast cancer inci-
methods, and epidemiologic research. N. l?ng/. J. Med. dence among atomic bomb survivors, Hiroshima and
307: 161 I-1617, 1982. Nagasaki, 1950-1969. JNCI 59: 799-811, 1977.
6. Hankey, B.F., Curtis, R.E., Naughton, M.D., Boice, J.D., 14. Pinkston, J.A., Antoku, S., Russell, W.J.: Malignant neo-
Flannery, J.T.: A retrospective cohort analysis of second plasms among atomic bomb survivors following radiation
breast cancer risk for primary breast cancer patients with therapy. Acta Radial. Oncol. 20: 267-27 1. 198 1.
an assessment of the effect of radiation therapy. JNCZ 70: 15. Shore, R.E., Hempelmann, L.H., Kowaluk, E., Mansur,
797-804, 1983. P., Pasternack, B., Albert, R., Hanghie, G.: Breast neo-
7. Jablon, S., Kato, H.: Studies of mortality of A-bomb plasms in women treated with x-rays for acute postpartum
survivors. 5. Radiation dose and mortality, 1950-1970. astitis. JNCI 59: 8 13-822, 1977.
Radiat. Rex 50: 649-698, 1972. 16. Tokunaga, M., Norman, J., Asano, M., Toluoka, S., Ezaki,
8. Kato, H., Schull, W.: Studies of mortality of A-bomb H., Nishimorei, I., Tsuji, Y.: Malignant breast tumors
survivors. Radiat. Res. 90: 395-432, 1982. among atomic bomb survivors, Hiroshima and Nagasaki,
9. Kerr, G., Solomon, K.: The epicenter of the Nagasaki 1950-1974. JNCI 62: 1347-1359. 1979.