Good science gives the same answer to questions posed by both male and female investigators. But good science normally only answers questions as they are posed. In the case of breast cancer, ordinarily only females experience the disease, with its attendant costs to themselves—and opportunities for others. The same is true for current breast cancer detection practices, which rely primarily on screening x-rays (mammography).
This reality is illustrated in the commentary published in the British Medical Journal on the 2014 article summarizing the 25 year follow-up of the Canadian National Breast Cancer study. The comments—all from men—focused, with one exception,[1] on the methodological and numerical minutiae of various randomized control trials and population-based investigations of the relationship of mammography screening and mortality from breast cancer. As for risks–overdiagnosis and radiation-induced cancer in later years–virtually nothing was said.
By assessing methods of breast cancer detection or treatment solely within the limited epistemic confines of quantitative measures of breast cancer detection and mortality, we neglect the single most compelling constant of all: the irradiation of women’s breasts and mid-sections for screening, diagnosis, and treatment.
The American Cancer Society (ACS) assures screening participants that mammography’s radiation dose is 0.4 millisieverts for “the typical mammogram with 2 views of each breast.” It adds that Food and Drug Administration regulations ensure that the “radiation dose is required to be very low.”[2] [See Baconspromise.org post IV, “Hunting Cancer with X-Rays.”] The society also compares a woman’s x-ray exposure from a mammogram to “about the same amount of radiation she would average from her natural surroundings over about 7 weeks.” This comparison obscures the fact that at any given time mammography exposures represent net increments to exposures from a woman’s natural surroundings.
When the ACS and the American College of Radiology further assure women that, despite its “risks,” screening mammography has been shown to have “benefit,” most women will infer that the “benefit” being referred to is a reduced risk of death from breast cancer. After all, “mammography saves lives.” Alas, it is not so simple.
For regulatory purposes the Food and Drug Administration defines mammography’s “benefit”as “the quality of the resulting radiograph[s].”[3] Meanwhile, since x-ray image quality is directly proportional to the radiation dose, improvement in mammography “benefit” has occurred along with a gradual increase in mean glandular dose per image to over four times the amount of exposure claimed by the American Cancer Society.[4]
The FDA’s own surveys of trends in mammography dose and image quality have found that the typical four image mammogram entails a mean x-radiation exposure of over 7 milligrays, or almost 18 times the amount advertised by the ACS. (This does not take into account repeated images that may be taken during the same mammography session for the benefit of the radiologist attempting the most accurate reading.) Meanwhile, the FDA’s regulatory upper dosage limit for standard mammography equipment is 3.0 milligrays per exposure, or a total of 12 milligrays for the four exposures of the typical screening mammogram.[6]
Radiology technicians naturally make every effort to minimize radiation exposures. But they must also make every effort to ensure the best possible image for the waiting radiologist, and the reality is that standard FDA-compliant equipment is able to direct much more radiation during a mammography than the limits the ACS suggests.
These higher levels of radiation are delivered with repeat or diagnostic mammograms, which are involved in breast biopsies performed whenever a woman’s screening mammogram produces “suspicious” results. According to one recent estimate, such biopsies are done “about 1.6 million” times a year.[7]
A recent analysis of the insurance records of 700,000 women between the ages of 40 and 59 found that all had had routine screening mammographies, 11% of which were regarded as “suspicious” and thus resulted in repeat mammograms, ultrasounds and needle biopsies. Except for the ultrasounds, the additional testing resulted in considerable additional x-radiation exposure. Less than 2% of these follow-on procedures confirmed cancer.[8] Repeat screening mammography, often within days of the initial exposures, has reportedly increased with the transition to digital mammography.[9]
Precise numbers for the frequency with which women are brought back for repeat mammograms are probably unobtainable. Combing through proprietary records of radiology departments and insurance claims would be a daunting task. That being so, assurances that the x-radiation exposure from mammography is too small to worry about—or “worth the risk”—are meaningless.
Perhaps the most insidious consequence of efforts to assure women that the radiation risk from mammography is “low,” is that such statements, and any comparative numbers offered to support them, invite women to infer that there is a safe threshold for x-radiation exposure, and that “low” is meaningful with reference to that threshold. But there is no such thing as a safe threshold.
In 1955 the National Academy of Sciences created a committee on the biological effects of atomic radiation to review what had been learned from laboratory science and the findings of the Atomic Bomb Casualty Commission, formed in late 1945 to assess the radiation consequences of the bombings of Nagasaki and Hiroshima.
The committee issued an initial assessment in 1956, which was updated for its 1960 report.[10] The sub-group examining genetic effects confirmed that irradiation of female mice produced genetic damage. (That x-rays damage chromosomes, and are thus capable of producing tumors, had been shown by Theofore Heinrich Boveri before World War I.) It also confirmed genetic consequences for the “children of survivors of the atomic bombings at Hiroshima and Nagasaki,” as well as “children elsewhere whose parents received radiation for medical or other reasons.”[11] However, cautioned the report,
. . . most of the man-made radiation to which the population of the United States is exposed involves dose rates not yet adequately investigated experimentally. For example, we do not know whether the effects of low doses given at high dose rates, as in medical exposures, will be more like the response from acute irradiation or more like that from chronic irradiation.”
Pathologies such as skin cancer and leukemia may result from “relatively low level” exposure “from time to time over a period of years. . . .” and it is “characteristic” of radiation that its “effects may manifest themselves not only immediately, but perhaps only after a long period of intermittent . . .exposure.”[12]
Delayed manifestation of biological effects results from the fact that . . .all types of induced and spontaneous tumors appear not to arise at once . . . . There is much evidence indicating that malignant change ordinarily develops only after a series of ‘precancerous’ changes or a state of tissue disorder has taken place.
The committee cautioned against predicting “human tumor incidences from small radiation doses based on extrapolation from the observed incidences following high dosage,” such as those recorded after the atomic bombings of Japan. Furthermore, prediction—
. . .requires evaluation of the possibility that there is a threshold dose below which there is no probability of inducing leukemia, a concept which implies a factor of safety that would be most reassuring to those who are exposed to radiation in excess of the natural background . . . . However, no member of the Subcommittee feels that he can estimate the size of the threshold or, for that matter, even prove its existence. . . . The detailed information available . . . argues Strongly against a DNA repair-mediated low-dose threshold for cancer initiation. [13]
Subsequent reports from the National Academy of Sciences’ Committee to Assess Health Risks from Exposure to Low Level Ionizing Radiation have affirmed this conclusion, viz. “it is prudent to assume that there is no threshold.”[14]
It is a truism that we usually get answers only to the questions we ask. So let us enlarge the limited frame of reference for most randomized control trials or population-based studies of screening mammography’s merits, in three ways:
First, begin a population based inquiry in 1950, prior to the adoption of screening mammography [See Table 3], to detect larger “before” and “after” patterns requiring explanation. This, as we have seen, the National Cancer Institute has already begun to do in order to examine whether the earlier diagnosis of breast cancer has, in fact, made a difference in the overall rate of breast cancer incidence. (See Blog Post VI, “The National Cancer Institute Weighs In.”)
Secondly, if we are serious about learning the risks of mammography screening, we will examine data reflecting changes in the incidence of conditions that are known to be susceptible to x-radiation, for example, (a) other forms of cancer, especially originating in the chest area, such as the thyroid glands, trachea, bronchus and lungs, and (2) heart and cardiovascular disease.
Thirdly, we can focus on those ages in which mortality from cancer and heart disease are most pertinent, viz., over the age 45. Nature has provided us with a perfect comparison group for women confronted with these diseases: men.
The National Center for Health Statistics collects and maintains the largest and most comprehensive data on health in the United States, data which goes as far back as 1950. And it does so using analogous definitions, collection procedures, and statistical methods across all measures of health and disease.
If one compares deaths from malignant neoplasms of the trachea, bronchus and lungs for both men and women 45 years and older, over the six decades between 1950 and 2010, one cannot but be struck by the continued increase of female mortality from those diseases beginning in 1980, and especially after 1990, when corresponding male deaths began to decline precipitously. [See Table 1 ] (The same dramatic disparity is evident between men and women after the age of 65.) This disparity cannot be blamed on smoking. [See Table 2]
Meanwhile, the incidence of thyroid cancer has increased more than twice the rate in women than in men since 1991, with increases being larger among white women, who are more likely to have received screening mammograms. In the early 1970s, rates of thyroid cancer among women had begun to hold steady (black women) or decline (white women). An increase in the rates for both white and black women becomes evident in the 1980s and 1990s and does not reverse before 2011, year of the most recent SEER data.[15]
Then there is cardiovascular disease. During the three decades after 1950 the death rate from heart disease (or cardiovascular disease, CVD) in the United States for women 65 years and older remained the same, at about 300 per 100,000. Substantially more men 65 and older died from heart disease during that period, though their death rate improved significantly (from 425 to 377 per 100,000) in 1977. At the end of the 1970s appreciably fewer women of all ages were dying from cardiovascular disease than men.
By 1980 something changed. Deaths from cardiovascular diseases among women began to approach CVD deaths among men, surpassing the rate for men in four years. Indeed the numbers of men dying from cardiovascular disease began to decline after 1980, while the death rate for women continued to increase.
For a brief period (1990-1995) the CVD death rate for both men and women of comparable ages increased slightly, but in 1995 the rate resumed its decline for men. Women’s CVD death rate not only increased more sharply during 1990-1995, but continued to increase until 2000. Then their CVD death rate also began to decline, even more sharply than the rate for men. By 2010 the disparity in CVD death rates for men and women was the smallest since 1985.[16] [See Chart – CVD Mortality Trends]
Nearly half (47%) of the decrease in age-adjusted mortality from cardiovascular disease among men and women by 2000 has been attributed to improved treatments, with the remaining improvement attributable to reductions in risk factors, such as smoking and physical inactivity.[17] Nonetheless, in spite of improvements in the treatment of both cardiovascular disease (CVD) and cancer, in 2010 a woman over the age of 45 was twelve times as likely to die from CVD as from breast cancer.[18]
Common explanations for the post-1980 disparity in CVD mortality between men and women are that “women have smaller hearts and smaller arteries,” their hearts beat faster and 300 million more times in a lifetime than men’s hearts, while women’s heart rate dynamics are more variable and complex. What’s more, tobacco use is supposedly more dangerous in women.[19] However, such reasoning does not explain why fewer women than men died from heart failure for the three decades between 1950 and 1984.
Finally, let’s return to the possibility that some breast cancers that occurred, recurred, or metastasized among women in their 50s through 70s did so as a result of cumulative x-ray exposures from diagnostic or therapeutic mammography. A rereading of the National Cancer Institute’s attempt to quantify “overdiagnosis” in its 2015 guidance for physicians inadvertently raises this question.
For any given number of women over the age of 40, a certain number of breast cancers is likely to occur. Thus increases in early detection should be accompanied by corresponding decreases in breast cancer detection in later years. [20] Accordingly, the NCI examined “several observational population-based studies” that compared “breast cancer incidence before and after adoption of screening.”
If mammography screening were as effective as its proponents promise,
. . . there would be a rise in incidence followed by a decrease to below the prescreening level and the cumulative incidence would be similar. Such results have not been observed. Breast cancer incidence rates increase at the initiation of screening without a compensatory drop in later years. . . . A population-based study showed increases in invasive breast cancer incidence of 54% in Norway and 45% in Sweden in women aged 50 to 69 years, following the introduction of nationwide screening programs. No corresponding decline in incidence in women older than 69 years was ever seen. Similar findings . . . have been reported from the United Kingdom and the United States.
The NCI attributes these findings to “overdiagnosis.” But they could also reflect the induction of breast cancer by the cumulative exposure of women’s breasts to x-radiation. We cannot know how many of those later incidences of breast cancer are radiation-induced, but the fact that they do occur is a serious and troubling outcome, especially for older women.
* * *
If numbers provide an inadequate foundation for establishing medical protocols, it is because statistical correlation is too readily confused with causation, and because those aspects of biological phenomena that can be captured by quantities do not replicate biological processes over time—which carcinogenesis certainly is.
For some the “gold standard” of evidence in medicine is the randomized control trial. For others, scientific proof entails not only quantification, but replicability. But numbers cannot replicate biological processes, nor can we completely replicate in the laboratory the etiology of disease in a given group of persons. Still, numbers—especially in epidemiology—do have the power to capture evidence of trends, and thus to pose questions that might not otherwise be raised as a result of cultural bias, or economic interest.
So it is that we must question the ethics of continuing to subject the upper torsos of women asymptomatic and at low-to moderate risk to breast cancer to cumulative and repeated exposure of even low-levels of ionizing radiation without compelling and justifying reasons, applicable and acceptable to each individual patient. Compounded by the continuing and larger radiation exposures involved in over-diagnosis and over-treatment, the ethics of policies that promote widespread screening mammography beg to be challenged.
Screening mammography remains an experiment. We are learning the amount and costs of over-diagnosis and over-treatment. Given the complexity of possible additional health risks from the practice, the likelihood of providing sufficient reliable information to obtain genuinely informed consent to mammography is remote. Therefore, before continuing to consider routine screening mammography as appropriate for all women of any age we must also eliminate it as a contributor to disparate disease trends among adult men and women, as well as a source of recurring or new breast cancers.
(This is the concluding post in a seven part series on the unknowns of routine screening mammography for asymptomatic and low-to moderate risk women. As a result, screening mammography remains a medical experiment. The complexity of the unresolved questions mean that genuinely informed consent to the procedure is highly unlikely. Posted June 29, 2015.)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[1] Donald J. Benjamin, “Comment re. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: Randomised screening trial,” British Medical Journal (11 February 2014), http://www.bmj.com/content/348/bmj.g366, Downloaded March 2, 2015.
[2] American Cancer Society, “Radiation exposure from mammography,” http://www.cancer.org/treatment/understandingyourdiagnosis/ examsandtestdescriptions/mammogramsandotherbreastimagingprocedures/ mammograms-and-other-breast-imaging-procedures-mamm-radiation. Downloaded March 3, 2015. The sievert is a derived unit in the International System of Units which, when applied to human radiation exposure, incorporates an estimate of the probability of cancer induction or genetic damage. 1 milligray is conventionally considered equivalent to 1 millisievert.
[3] David C. Spelic, “Dose and Image Quality in Mammography: Trends during the First Decade of MQSA,” http://www.fda.gov/Radiation-EmittingProducts/MammographyQualityStandardsActandProgram/ FacilityScorecard/ucm113606.htm. Downloaded March 3, 2015.
[4] The specific FDA amounts are 1.5 milligrays in 1995 and 1.76 milligrays in 2003. The milligray is the metric measurement unit of absorbed radiation dose of ionizing radiation, e.g., x-rays. For FDA findings, see David C. Spelic, “Dose and Image Quality in Mammography: Trends during the First Decade of MQSA,”http://www.fda.gov/RadiationEmittingProducts/Mammography QualityStandardsActandProgram/FacilityScorecard/ucm113606.htm. Downloaded March 3, 2015.
[5] 1990 Recommendations of the International Commission on Radiological Protection (ICRP Publication 60). (Oxford: Pergamon Press, 1991).
[6] Food and Drug Administration, Mammography Quality and Standards Act Regulations, (67 FR 5446), Subpart B, Section 900.11 (e)(5)(iv)-(x), http://www.fda.gov/Radiation-EmittingProducts/ MammographyQualityStandardsActandProgram/ Regulations/ucm110906.htm. Downloaded March 3, 2015.
[7] Joann G. Elmore, et al., “Diagnostic Concordance Among Pathologists Interpreting Breast Biopsy Specimens,” Journal of the American Medical Association, Vol. 313, No. 11 (March 17, 2015), pp.1122-1132.
[8] Mei-Sing Ong and Kenneth D. Mandl, “National Expenditure for False-Positive Mammograms and Breast Cancer Overdiagnoses Estimated at $4 Billion A Year,” Health Affairs, Vol. 34, No. 4 (April 2015), pp. 576-583.
[9] Denise Grady, “In Shift to Digital, More Repeat Mammograms,” The New York Times (April 20, 2008).
[10] National Academy of Sciences-National Research Council, The Biological Effects of Radiation: Summary Reports (Washington, DC: 1960).
[11] National Academy of Sciences-National Research Council, The Biological Effects of Radiation: Summary Reports (Washington, DC: 1960), p. 3.
[12] National Academy of Sciences-National Research Council, The Biological Effects of Radiation: Summary Reports (Washington, DC: 1960), pp. 27-29.
[13] National Academy of Sciences-National Research Council, The Biological Effects of Radiation: Summary Reports (Washington, DC: 1960), pp. 32-35.
[14] For example, “at the level of cancer-associated gene or chromosomal mutation, the presence of a true dose threshold demands totally error-free DNA damage response and repair.” National Academy of Sciences, Biological Effects of Ionizing Radiation, V (National Research Council, 1990), National Academy of Sciences, Biological Effects of Ionizing Radiation, VII, Phase 2 (National Research Council, 2006), p.245.
[15] National Cancer Institute, “A Snapshot of Thyroid Cancer.” http://www.cancer.gov/researchandfunding/progress/snapshots/thyroid; National Cancer Institute, Surveillance, Epidemiology, and End Results (SEER) Program, FastStats, Table 26.3. Downloaded April 14, 2015.
[16] Alan S. Go, MD, et al., “AHA Statistical Update: Heart Disease and Stroke Statistics—2014 Update: A Report From the American Heart Association,” Circulation, Journal of the American Heart Association (December 18, 2013); Source: National Center for Health Statistics: “Health, United States, 2012,” Hyattsvile, MD (2013); National Center for Health Statistics, “Health, United States, 2010, With Special Feature on Death and Dying” (Hyattsville, MD., 2010).
[17] Earl S. Ford, MD., et al., “Explaining the Decrease in U.S. Deaths from Coronary Disease, 1980-2000,” New England Journal of Medicine, Vol. 356 (June 7, 2007), pp. 2388-2398.
[18] National Center for Health Statistics: “Health, United States, 2012,” Hyattsvile, MD (2013); National Center for Health Statistics, “Health, United States, 2010, With Special Feature on Death and Dying” (Hyattsville, MD., 2010).
[19] See, for example, Hope Ricciotti, MD, “Heart Disease – Differences Between Men and Women,” Beth Israel Deaconess Medical Center. http://www.bidmc.org/Centers-and-Departments/Departments/Medicine/ Divisions/Cardiovascular-Medicine/For-Patients/Your-Heart-Health/Tips-for-Heart-Health/Heart-Disease—Differences-Between-Men-and-Women.aspx. Downloaded March 17, 2017, and Mike Zimmerman, “Heart Disease: Lifesaving News,” AARP: The Magazine (April-May, 2014).
[20] National Cancer Institute, “Breast Cancer Screening (PDQ®): Harms of Screening Mammography: Overdiagnosis,” Update: February 6, 2015. http://www.cancer.gov/cancertopics/pdq/screening/breast/healthprofessional/page8. Downloaded February 25, 2015.
