Sternglass on reactors and breast cancer

[John Moulder <jmoulder@post.its.mcw.edu>]

EJ Sternglass and JM Gould: Breast cancer: evidence for a relationship to 
fission products in the diet.  International Journal of Health Services 
23:783-804, 1993.


**Authors abstract:
"To establish the possible relationship between breast cancer mortality and 
low doses of radiation due to fission products in the environment, the 
mortality rates in the 9 census regions of the United States for the years 
1984-1988 were correlated with the cumulative airborne releases from all 
nuclear plants in each region for the period 1970-1987.  A high correlation 
coefficient of 0.91 was obtained for a logarithmic dependence on the total 
releases, consistent with an indirect action via free-radical oxygen at very 
low dose rates, in contrast to a direct action on DNA at high dose rates, 
explaining the wide differences in risk per unit dose obtained in the earlier 
studies.  The recent temporal changes of breast cancer rates in the New York 
metropolitan area including nearby Connecticut, Westchester, and Long Island 
were examined in relationship to the releases from nearby nuclear power 
plants and found to be consistent with a dominant role of short-lived fission 
products in drinking water and fresh milk.  The results support a major role 
for nuclear plant releases in industrial countries in the recent rise of 
breast and other forms of cancers not related to smoking, especially among 
older persons, and strongly support the need to replace nuclear reactors with 
more benign ways to generate electricity."


**Source:  
This was published in the Environmental Health Policy section of a non-
science journal.  It is not clear whether the article was peer-reviewed.  


**Authors Methodology:  

*In the first part of the paper, age-adjusted breast cancer rates (1984-1988) 
in the nine census regions of the United States were compared to cumulative 
(1970-1987) per capita releases of airborne particulate radio-isotopes from 
nuclear power plants in the same states.  The three regions with the highest 
releases (New England, Middle Atlantic and East North Central) had age-
adjusted breast cancer mortality rates of 29-31 cases per 100,000 persons per 
year.  In the remainder of the US the rates are 23-27 cases per 100,000 per 
year.

*The per capita cumulative releases in the NE portions of the US were stated 
to be about 3.5 Curies per million persons.  That comes to about 0.2 
microcuries per person per year.  In the remainder of the US the cumulative 
releases were less than 0.9 Curies per million persons.

*Based on their number, I calculate the relative risk for the "high emission" 
regions compared to the "low emission regions" to be 1.16 (95% c.l. of 1.11 - 
1.22).

*Next, the authors compare trends in breast cancer death rates (not age-
adjusted??) in the NE United States with the dates at which nuclear power 
plants were started up in the area.  


**What to we know about breast cancer?

*The incidence of breast cancer has been rising recently in the US, from a 
low of about 80 cases per 100,000 per year in the late 70's to a high of 
about 110 cases per 100,000 per year in the late 80's.  There is some 
argument about whether the recent rise in breast cancer rates in the US is 
real, or whether it reflects advances in diagnosis.  The fact that the 
incidence seems to have leveled-off or even dropped slightly since 1987 may 
indicate that the rise was a diagnostic artifact.  The mortality rate from 
breast cancer has not risen over this period of time.

*The known major risk factors for breast cancer:
  Risk increases with age (that why you need age-adjusted rates)
  Family history of breast cancer increases risk
  "Whites" are at higher risks than "blacks"
  Upper socioeconomic classes have higher risks than lower
  Age at first pregnancy (older is worse)

*Breast cancer incidence rates are known to be higher in the Northeast and 
Northcentral United States than elsewhere in the US.  There has been 

speculation that this is due to geographical differences in risk factors. 

*Internationally, breast cancer rates are highest in Northern Europe 
(England, Denmark, Scotland, Ireland, Belgium, Netherlands hold the top 
positions).  Outside of northern Europe only New Zealand breaks into the top 
ten.  The lowest rates are found in Southeast Asia (China, Japan, Korea, 
Thailand, Hong Kong) and Central and South America (Ecuador, Chile, Mexico, 
Panama, Venezuela).

*Breast radiation is a known risk factor for breast cancer.  Best current 
estimates are for a life-time mortality risk of 700 cases per 100,000 persons 
per Sv (7 per 100,000 per rem).  Excess breast cancer has been seen in the 
Japanese atomic bomb follow-up studies, for women irradiated for post-partum 
mastitis and for women who received multiple fluoroscopic chest exams.


**Review

*There is relatively little dispute that breast cancer rates are higher is 
the Northeast and Northcentral portions of the US than elsewhere.  The 
authors, main argument is that since this is the region where the density of 
power plants is highest, and where most releases have occurred, then these 
releases must be the cause of the excess cancer.  Since the authors take no 
possible differences in risk factors (other than age) into account this is 
not a very impressive argument. 

*Next the authors argue that the breast cancer rates in Connecticut rose in 
the mid-60's because that was when the Haddam and Millstone plants went on-
line.  In fact, the authors' own graph shows that breast cancer incidence 
rates in Connecticut have been increasing steadily since 1945. This is 
epidemiology by anecdote. 

*The authors further argue that the recent differences in breast cancer rates 
between Westchester County (an upper-class NY city suburb with a high breast 
cancer rate) and New York City are due to the fact that the city draws its 
water from "uncontaminated" reservoirs in upstate NY.  Such a comparison is 
meaningless until controlled for risk factors such as race and socioeconomic 
status which are known to be different in the two geographical areas.

*The author's then turn to the international data and argue that breast 
cancer rates are increasing in most industrialized countries that rely on 
large commercial nuclear reactors. To make their argument the authors analyze 
changes in rates rather than actual rates, since as discussed above, the 
actual breast cancer rate data shows a dependence on regions/cultures rather 
than on levels of nuclear power use.  The authors argue that the low breast 
cancer rates in Japan, despite their nuclear program, are due to the fact 
that Japanese women do not consume dairy products, red meat or freshwater 
fish (where they argue that the "contamination" is highest).  No actual 
analysis of the correlation between nuclear power use (or radionuclide 
emissions) and breast cancer rates are presented, so this argument is 
unevaluable.

*The authors do not appear to have made any estimate of dose, but they claim 
that there data shows that reactor isotopes "are 1000 times more biologically 
serious" than other forms of radiation.  

*They explain this huge difference in risk factors between their analysis and 
all other studies of radiation-induced breast cancer by arguing that at the 
very low dose rates associated with these ingested isotopes: "free-radical 
oxygen molecules are much more efficient at producing biological damage to 
cell membranes than direct hits on the nuclear DNA, which is the dominant 
mechanism involved in short, high dose rate exposures such as occur in chest 
X-rays or mammography".  This explanation is at odds with current 
understanding of the biology of ionizing radiation.  First, the damage done 
to DNA by X-ray exposure at high dose rates is due predominantly to hydroxyl 
free-radicals produced by the radio-ionization of water.  It is only at very 
low dose rates that direct DNA damage is though to occur (for medical 

physicists, consider the Theory of Dual Radiation Action, e.g., Rossi & 
Zaidler, Med Phys, 1991).  Second, oxygen radicals occur naturally in cells; 
they are toxic, but the cell has biological systems for dealing with small 
quantities of such radicals.  Furthermore, the dose rates involved will not 
produce any significant increase in oxygen radicals.  Third, there is no 
evidence that oxidative damage to membranes is involved in carcinogenesis. 

* The authors claim that there has been "a sharp rise in the incidence and 
mortality rates for most types of cancer not related to smoking, particularly 
among older age groups, in the industrialized countries".   This is certainly 
not true for women in the US.  In women age-adjusted mortality rates for 
uterine, colorectal, stomach and liver cancer decreased significantly between 
1945 and 1990.  Ovary and breast cancer age-adjusted mortality rates have 
been steady over the same period.  Only lung and pancreatic mortality rates 
in women are up. 

*The authors state that the cumulative total of 370 Curies released is 
"biologically very significant, considering that one curie is one trillion 
picocuries". Does this mean that it will be even more biologically 
significant when converted to Becquerels??  Again, there is no attempt to 
calculate a dose or to compare the dose from reactor-produced radionuclides 
to the dose from naturally-occurring internalized radioisotopes.

*The authors state that the "The average yearly dose from natural background 
in the United states from all external and internal sources...[is] 87 
millirads."  According to the latest NCRP estimate the effective dose 
equivalent is 360 millirem (3.6 mSv).


**Questions:
1)  Can anyone give me a even a rough estimate of the dose that would be 
produced by the type of body burden described (even if you don't believe the 
data)?
2)  Does anyone know how the body burden of reactor-produced isotopes 
compares the body burden of natural isotopes.

John Moulder (jmoulder@its.mcw.edu)          Voice: 414-266-4670
Radiation Biology Group                      FAX: 414-257-2466
Medical College of Wisconsin, Milwaukee