back to XHP | CNR | radiation | rat haus | Index | Search | tree

ASCII text

X-Radiation and Gamma Radiation:
Comments on Their Nomination as Known Human Carcinogens
for the Eleventh Report on Carcinogens (RoC)

John W. Gofman, M.D., Ph.D.*
Professor Emeritus of Molecular and Cell Biology, U.C. Berkeley

September 11, 2001

Submitted by Priority Mail to:
Dr. C.W. Jameson
National Inst. of Environmental Health Sciences (NIEHS)
National Toxicology Program (NTP)
79 Alexander Drive, Building 4401, Room 3118
PO Box 12233
Research Triangle Park NC 27709

For the convenience of the NTP, this submission includes a copy of the author's 1999 peer-reviewed monograph, Radiation from Medical Procedures in the Pathogenesis of Cancer and Ischemic Heart Disease: Dose-Response Studies with Physicians per 100,000 Population.

* The author's CV is summarized on the two final pages of this submission.

Part 1: How Many Citizens Are Exposed to X-Radiation?
Part 2: How Big Are X-Ray Doses in Medical Imaging? Tables 1 & 2.
Part 3: A Major New Addition to Evidence that X-Rays Are a Known Human Carcinogen.
Part 4: The Absence of Any Risk-Free (Safe) Dose-Level of Low LET Radiation.
Part 5: Complex Mutations and Genomic Instability in X- and Gamma-Irradiated Cells.
Author's CV


Part 1:
How Many U.S. Citizens Are Exposed to X-Rrays and Gamma Rays?

These comments apply primarily to x-ray exposure received during medical imaging procedures (during diagnosis, during surgery, during placement of catheters, needles). These comments do not address x-rays and gamma rays used at very high doses for cancer therapy because the Report on Carcinogens (RoC) lists causes of cancer, not potential treatments.

The number of x-ray imaging procedures performed annually in the USA was and remains poorly documented. For instance, the annual number for 1985-1990 in the USA was estimated to be at least 800 diagnostic x-ray exams per thousand population, excluding dental x-rays (UNSCEAR 1993, Table 6, p.279). That estimate "could be an underestimate by up to 60%" (UNSCEAR 1993, p.229/46).

X-ray imaging procedures are hardly limited to older, presumably less radio-sensitive persons. According to NCRP 1989 (at p.19, citing the FDA in 1985), 47% of all x-ray imaging exams were administered to patients below age 45 years.

Since 1990, the number of procedures giving the highest doses --- namely fluoroscopy and CT scans --- have increased dramatically. Despite earlier expectations that MRI would replace CT, the use of CT has increased ceaselessly at an estimated rate of about 10% per year "and will continue to increase for the foreseeable future" (Ravenel 2001, p.279). The estimated number in 1998 was about 33 million CT exams (Nickoloff 2001, p.285).

The use of fluoroscopy during diagnostic cardiac catheterizations (estimated over 1 million times per year), during cardiac angioplasty (estimated over 700,000 times per year), and during other procedures, also has increased substantially in recent years (Shope 1997, p.i). Fluoroscopy creates the potential for very high, localized x-ray doses (hundreds of centi-Grays or rads), because the x-ray beam stays "on" either continuously or in a pulsed fashion.

There can be no doubt that significant numbers of U.S. citizens are exposed every year to medical x-rays for imaging purposes. Moreover, the evidence is overwhelming now that there is no threshold dose (no risk-free dose); please see Part 4. Therefore, every exposure counts, and the consequences (including carcinogenic mutations) accumulate.


Part 2:
How Big Are X-Ray Doses in Medical Imaging? Tables 1 and 2

The mistaken assumption, that x-ray exposure from medical imaging is negligible, has been very widely embraced. Although the NTP Reports on Carcinogens explicitly exclude any risk-assessments, the NTP has the responsibility to evaluate whether or not exposure to a nominated carcinogen is literally negligible.

Part 2 will show that x-ray exposure from medical imaging has been, and continues to be, far from negligible. It is much higher than appreciated.

In 1990, the BEIR-5 Committee embraced without critique the 1987 estimate by the National Council on Radiation Protection, that the average annual per capita "effective" dose equivalent from "x-ray diagnosis" was only 0.039 centi-Sievert (cSv) or rem (BEIR 1990, p.18).

We have challenged that estimate as non-credible (Gofman 1999, pp.33-38), and have shown the basis for estimating that annual average per capita dose from diagnostic medical x-rays in 1950 may well have been as high as 0.65 centi-Gray or rad (Gofman 1999, pp.609-616).

Since 1950, there may be only a little net change. Because x-ray doses were not and still are not measured, this is a question where uncertainty will be permanent. But the immense growth in CT and fluoroscopy since 1975 may well have offset the post-1950 dose-lowering effects of tuberculosis-eradication, the phase-out of pre-delivery pelvimetry, and introduction of faster films and better beam collimation.

Patients More Exposed than Nuclear Workers and Japanese A-Bomb Survivors

Tables 1 and 2 are located between pages 3 and 4.

Table 1 shows why medical patients can readily accumulate doses from x-rays considerably higher than the average annual dose accumulated occupationally by civilian "nuclear workers," who now accumulate less than 0.5 centi-sievert per year, on the average.

Table 2 indicates why it is easy for patients to accumulate higher doses to some internal organs than did most survivors of the Hiroshima-Nagasaki bombs --- a comparison which necessarily requires adjustment for the higher mutagenic potency of 90 kVp x-rays than the mutagenic potency of a-bomb gamma rays.  Gofman 1999 (pp.46-48) provides a detailed analysis of this point, with many references from the peer-reviewed literature.

Estimated Doses to Patients

Extra radiation during commercial airplane flights within USA. "Effective" dose/hr.  And
per 10 hours.
0.0003 cSv per hour.  And
0.003 cSv per 10 hours.
UNSCEAR 1993, p.38.  
CT scans, general. Tissue dose per scan. 1-3 cSv. Mettler 2000, p.352.  
CT head scan, adult. Surface dose. 3-7  cGy (rads). Nickoloff 2001, p.285.  
CT head scan, adult. "Effective" dose. 0.15 cSv. Mettler 2000, p.352. 50 to 1.
CT chest, typical. Surface dose. 2-5 cGy (rads). Nickoloff 2001, p.286.  
CT chest, typical. "Effective" dose. 0.54 cSv. Huda 2000, p.843. 180 to 1.
CT chest, unspecified. Breast: Mean glandular dose. Up to 5 cGy (rads). Gray 1998-a, p.63.  
CT multi-slice of heart    for calcium score. Surface dose. Up to 10-20 cGy. Nickoloff 2001, p.286.  
CT chest angiograph. Surface dose. 2-4 cGy (rads). Nickoloff 2001, p.286.  
CT chest, cancer screening. Surface dose. 0.2 - 0.4 cGy (rad). Nickoloff 2001, p.286.  
Electron Beam CT chest angiography or cardiac calcium score.  "EBCT." X-ray beam travels from back to front. Reduced dose to breasts and front chest wall. Nickoloff 2001, p.286.  
CT abdominal, adult. Surface dose. 2-5 cGy (rads). Nickoloff 2001, p.285.  
   Adult. "Effective" dose. 0.39 cSv. Ware 1999, p.64. 130 to 1.
   Young adult. "Effective" dose. 0.44 cSv. Ware 1999, p.64. 147 to 1.
   Child. "Effective" dose. 0.61 cSv. Ware 1999, p.64. 203 to 1.
CT-fluoroscopy, for imaging in biopsies, etc. Range of typical dose-rates. 20-60 cGy (rads) per minute. Nickoloff 2001, p.285.  

Dose-Rates per Minute, and One Estimate for Cardiac Balloon Angioplasty.

Fluoroscopy, general. Dose-rate per minute, on equipment made before 1995.  Upper limit can be restricted by choice to 20. 2 to 50 cGy (rads) per minute. FDA 1994, pp.2-3.
  Equipment made after 1995. 2 to 20 cGy (rads) per minute. Code of  Fed. Regulations:
21 CFR 1020.32 (e) Flu Equip.
CT-fluoroscopy, e.g. for complex needle biopsies. "Typical" dose-rates delivered per minute. 20 to 60 cGy (rads)  per minute. Nickoloff 2001, p.285.
Fluoroscopy during cardiac angioplasty. Surface dose, estimated per stenosis. 60 cGy (rads) total, per stenosis. NCRP 1989, p.31.


The Higher Mutagenic Power of Medical X-Rays than High-Energy Gamma

A reasonable estimate at this time is that the cancer-risk per rad or centi-gray is about three times higher from 90 kVp x-rays than from a-bomb gamma rays. Therefore, in risk-assessment, it would be a severe error to assume that results from the a-bomb study apply directly to medical patients.

Table 1: The Difference between Real Doses and "Effective" Doses

Doses are reported in units of grays or rads. A dose of ionizing radiation is a quantity of energy delivered per gram of tissue. A rad means 100 ergs of energy per gram of tissue. Since there are 100 rads per gray, 1 rad is exactly the same dose as one centi-gray (cGy). The roentgen is a dose-unit roughly equivalent to the rad or cGy.

By contrast, the term "effective" before a dose is a big flag which means, "This is not a dose at all; it is an artificial value which estimates relative detriment." The effective dose is an attempt to estimate what dose to the entire body would have caused the same amount of detriment (risk) as the actual exam which irradiated only specific parts of the body (McCullough 2000). Thus, "effective" doses are usually considerably lower than real doses --- as Table 1 shows. The dose-unit is the centi-Sievert (cSv), which is exactly the same as the rem. These units incorporate a crude adjustment for the different mutagenic potency of x and gamma (low LET) vs. alpha (high LET) radiation.

"Effective" doses are necessarily much less credible than real doses, because "effective" doses incorporate a long series of estimates and assumptions about "tissue weighting factors," which attempt (despite woefully inadequate evidence) to assess the attributable probability of fatal cancer in different organs, of the additional detriment from non-fatal cancer and hereditary disorders, and of the different latency periods for cancers of different kinds. By contrast, a real dose is an estimate or measurement of something objective: Energy delivered per gram of tissue.

McCullough emphasizes that "It is important to recall that these [effective doses] are estimates, based on many assumptions, and are not directly applicable to any one individual . . . These values, although not intended to describe the dose to an individual, can be used as a relative measure of stochastic injury (e.g., cancer induction or genetic effects)" (McCullough 2000, p.835).

Table 1: Why the Doses Are Merely "Ballpark" Estimates

With the rarest exceptions, actual surface doses are not measured during x-ray imaging procedures, even though small thermo-luminescent dosimeters (TLDs) have been shown not to interfere with images. Instead, some efforts have been made to measure doses on phantoms (dummies), but the limited exposure-circumstances during such tests provide unreliable dose information about exposures in real-world practice. After all, the very same equipment will deliver very different doses depending upon what settings and techniques the operator chooses (the first note in Table 1).


Part 3:
A Major New Addition to Evidence that X-Rays Are a Known Human Carcinogen

In November 1999, a major prospective dose-response study of unique design provided what is probably the most powerful confirmation anywhere, that virtually all types of human cancer are inducible by medical x-rays in both males and females. The study is entitled Radiation from Medical Procedures in the Pathogenesis of Cancer and Ischemic Heart Disease: Dose-Response Studies with Physicians per 100,000 Population, and I am its author (Gofman 1999).

That study has been independently peer-reviewed by a former chair of the BEIR Committee and former director of the National Cancer Institute (Arthur Upton, M.D.) and by a professor of physiology at Temple University School of Medicine (Prof. Howard S. Pitkow). Their comments are attached to this submission, after the Reference List.

Although prospective dose-response studies are the "gold standard" in epidemiology for establishing causation, they are inherently unable to prove that some "mystery agent" is not the real cause of a positive dose-response. However, many lines of evidence in Gofman 1999 do virtually rule out explanations other than medical radiation as the cause of its incontrovertible positive correlations (discussion in Chapters 68 and 69).

All-Cancers-Combined and Major Types, Males and Females Separately

Gofman 1999 reveals that by 1940, medical radiation in the United States had become a necessary co-actor in about 90% of the age-adjusted male cancer mortality rate, and about 58% of the age-adjusted female cancer mortality rate (Chapter 6 and 7). These percentages cannot be dismissed as irrelevant today, because average annual per capita exposure to medical x-rays may not be substantially lower now than it was in the years preceding 1940 (discussion in Part 2, above).

The prospective nature of the study is reflected by the fact that the 1940 age-adjusted national cancer mortality rates, for men and women separately, can be well predicted by analysis of the x-ray doses given in 1921 and 1931 (Gofman 1999, pp.213-214, p.222).

Three Meritorious Scientific Differences from Other Studies

The design of the 1999 study differs in three very positive ways from most other epidemiological analyses of low LET radiation.

First, such analyses are often based a) on highly unreliable dose-estimates (because individual doses are not measured --- they are estimated later, often decades later) and (b) on highly unreliable estimates of risk per dose-unit (because such values derive from unreliable dose-estimates and/or from unreliable assumptions about the relative carcinogenic potency of particular types of low LET radiation).

By contrast, the 1999 study avoids both of these pitfalls by using a sensible measure of relative accumulated x-ray exposure (Gofman 1999, Chapter 3).

Second, other radiation studies sometimes use databases where the opportunity has existed for subjective choices, particularly on dosimetry and on exclusions. This is particularly true of the A-Bomb Survivor Database, whose practices --- of changing input after the results are known --- are discussed in both Gofman 1990 (Chapters 4, 5, 6) and Gofman 1999 (pp.43-44, pp.54-55). Moreover, influential analysts of that database have sometimes chosen to discard selected pieces of it. For example, the BEIR-1990 analysis discarded the observations from the 1950-1955 period except for breast cancer (BEIR p.168), discarded the observations from the two highest dose-groups (BEIR p.165), discarded cancer deaths which occurred beyond age 75 (BEIR p.165), and made no use of its own finding that the dose-response was supra-linear (BEIR p.200).

By contrast, Gofman 1999 marries two databases which are utterly neutral with respect to radiation: The American Medical Association's database on physicians per 100,000 population, by the nation's nine Census Divisions, and the U.S. Vital Statistics on age-adjusted cancer mortality rates, by the nine Census Divisions. Moreover, the analysis does not discard data.

Third, other radiation studies suffer from the problem of small numbers. Even the Life Span Study of the A-Bomb Survivors has a (repeatedly revised) database of only about 100,000 participants. Subdivision of cancer by types in such a study often produces statistically marginal findings.

By contrast, the 1999 study of x-ray-induced cancer "enrolls" 130 million participants --- the entire 1940 population of the United States.  For age-adjusted mortality rates for all cancers combined, by Census Divisions, the male dose-response with accumulated x-ray exposure has an R-squared value of 0.95, and for females, the value is 0.86. Subdivision of the data by the major types of cancer, separately for males and females, still yields highly significant dose-response relationships (summary table in Gofman 1999, p.217). The R-squared values are as follows: 0.92, 0.91, 0.76, 0.92, 0.94, 0.78, 0.72, 0.87, 0.96. The only exception to a highly significant, positive dose-response in all these studies was found in female genital cancers, with an R-squared value of 0.07.

The Bottom Line: A Known Cause of Cancer, Lacking a Dose-Threshold

We submit these data (Gofman 1999) as a major new addition to the human epidemiological evidence that x-rays are a known cause of human cancer.


Part 4:
The Absence of Any Risk-Free (Safe) Dose-Level of Low LET Radiation

Gofman 1999 also presents (in its Appendix B) a nine-page summary of the overwhelming evidence that cancer risk from x-rays and gamma rays extends all the way down to zero dose (with excerpts from Gofman 1990, UNSCEAR 1993, and NRPB 1995). After publication of Gofman 1999, the finding of excess breast cancer in a study of scoliosis patients (Doody 2000) provided additional epidemiological evidence against any threshold dose. In the Doody study, the patients received x-radiation in serial doses which were estimated (long afterwards) at only 0.6 cGy (rad) per exam --- a dose which is the lowest conceivable dose (1 primary ionization track, on the average, per cell nucleus) with respect to DNA or chromosomal damage (Gofman 1999, p.522).

By any reasonable standard of biomedical proof, the evidence from human epidemiology and the physical evidence from track-analysis combine to demonstrate that cellular repair processes, for nuclear DNA and chromosome injuries, are unable to deliver a safe (risk-free) dose of low-LET radiation --- including x-rays and gamma-rays.


Part 5:
Complex Mutations and Genomic Instability in Irradiated Cells

There is a vast literature on human cell-studies which demonstrates that x-rays and gamma rays are a potent cause of structural chromosomal mutations of every sort, including re-arrangements, acentric fragments, and deletions ranging in size from multiple genes probably down to single nucleotides. (The deletion of a single nucleotide is no small matter, since it can scramble the genetic code by causing a frame-shift.)

The dose-response shape for the easily detectable aberrations appears to be linear down to the 2 cGy dose-level (Lloyd 1992). One of the best sources, for evidence about the complex types of damage inflicted by low-LET radiation upon the human genome, are the studies by Ward (1991, 1994, 1995) and Sutherland (2000).

Sutherland's observations include double-strand chromosome breaks (which are the basis for deletions, translocations and every other type of chromosomal re-arrangement) and other sorts of "clustered" DNA damage incorporating "two or more closely spaced damages" such as strand breaks, abasic sites and/or oxidized bases. Sutherland and colleagues are working experimentally with both gamma rays from cesium-137 and with 50 kVp x-rays (Sutherland 2000, p.107).

In the same paper, Sutherland and colleagues conclude that their work confirms that each cluster "results from a single radiation track" (Sutherland 2000, p.106). If so, this constitutes additional support for the conclusion that there is no risk-free dose-level of exposure to x-rays and gamma-rays.

The induction of single-strand and double-strand chromosome breaks by ionizing radiation is under study by Boudaiffa et al, who report on the role of the low-energy secondary electrons in strand breakage (Boudaiffa 2000).

Induction of Genomic Instability by X-Rays and Gamma Rays

Genomic instability refers to abnormally high rates (possibly accelerating rates) of genetic change occurring serially and spontaneously in cell-populations, as they descend from the same ancestral cell (Gofman 1999, p.533).

Many (not all) cancer biologists now believe that genomic instability "is one of the most important aspects of carcinogenesis" (Morgan 1996, p.247; additional references and discussion in Gofman 1999, Appendix D). In 1976, Peter C. Nowell published his classic paper proposing that "the biological events recognized in tumor progression represent (i) the effects of acquired genetic instability in the neoplastic cells, and (ii) the sequential selection of variant subpopulations produced as a result of that genetic instability" (Nowell 1976, p.25). In 1971, I saw in our lab the operation of selective advantage for certain gamma-ray-induced mutations in cultured human fibroblasts: "There is no question that the cells with profound structural re-arrangements of chromosomes became the dominant and finally, with adequate survival time, the only reproducing cells in the culture" (Minkler 1971, p.73).

Can exposure to x-rays and gamma-rays induce genomic instability?

Evidence appears to support an affirmative answer. The following references (which provide many additional references) can be recommended:

Holmberg 1993. Kronenberg 1994. Marder 1993. Mendonca 1993. Morgan 1996.


# # # # #


Reference List



The following links include the peer-reviewed contents included with this report:

XaHP: The X-rays and Health Project.
      An educational project of the
      Committee for Nuclear Responsibility.

The plan:  Lower x-ray doses per x-ray procedure.
The result:  Fewer cases of x-ray-induced cancer
and coronary heart disease.

back to XHP | CNR | radiation | rat haus | Index | Search | tree