The Evidence for Excess Risk of Cancer and Non-Cancer Disease at Low Doses and Dose Rates

Abstract

The question of whether there are excess radiation-associated health risks at low dose is controversial. We present evidence of excess cancer risks in a number of (largely pediatrically or in utero exposed) groups exposed to low doses of radiation (<0.1 Gy). Moreover, the available data on biological mechanisms do not provide support for the idea of a low-dose threshold or hormesis for any of these endpoints. There are emerging data suggesting risks of cardiovascular disease and cataract at low doses, but this is less well established. This large body of evidence does not suggest and, indeed, is not statistically compatible with any very large threshold in dose (>10 mGy), or with possible beneficial effects from exposures. The presented data suggest that exposure to low-dose radiation causes excess cancer risks and quite possibly also excess risks of various non-cancer endpoints.

INTRODUCTION

The detrimental tissue-reaction (deterministic) and stochastic effects associated with moderate- and high-dose ionizing low-linear energy transfer (LET) radiation (e.g., X rays, γ rays) exposure are well known (1). Much more controversial are the health effects at low doses (<0.1 Gy) or low dose rates (<5 mGy/h) (2, 3). In contrast to tissue-reaction effects, for stochastic effects scientific committees generally assume that at low doses there is a positive linear component to the dose response and that there is no threshold, or beneficial effect (1). However, as we review below there is also accumulating direct evidence of excess risk² of cancer and various other health endpoints in a number of populations exposed at moderate and low doses.

The health risks of low-level exposure to ionizing radiation have been assumed to be associated primarily with cancer (1). However, evidence has recently emerged of an association between lower doses (<0.5 Gy) and late-presenting cardiovascular disease (CVD) (all circulatory disease) (4-6). There is also accumulating evidence from various occupational groups exposed at low dose rate of excess risks of cataract (7, 8). The possible associated mechanisms are necessarily somewhat uncertain for both endpoints, although some plausible hypotheses have been advanced (9-11).

Nevertheless, the issue of low-dose radiation risk is controversial, and there have been claims that low dose risks are markedly overestimated by the use of linear extrapolation from moderate dose exposed groups (12) and there are also those claiming that linear extrapolation substantially underestimates low dose risk (13-16).

A related question to that of the existence or non-existence of low-dose risk is whether the risk at low doses is approximately linear with dose, an assumption which underlies the linear-no-threshold (LNT) model commonly assumed by expert advisory bodies (2). LNT is recognized to be an approximation, made for practicality in the context of radiological protection, although one for which there is some radiobiological basis, based on DNA damage considerations, as we demonstrate below; as we argue there is also a considerable body of evidence that it is not excessively conservative, indeed that there is considerable evidence of cancer risk at low dose (<0.1 Gy), and emerging evidence of certain types of non-cancer risk at somewhat higher levels of dose (<0.5 Gy). The present paper briefly summarizes a large number of comprehensive reviews of the low-dose epidemiologic literature (17, 18) as well as more specialist and mostly systematic reviews (5, 19-23); there have been similar reviews of radiobiologic data (24), albeit not so narrowly focused on low doses. This commentary does not address the question of possible genetic risks.

Radio-Epidemiological Findings

Detrimental tissue-reaction effects (deterministic effects) and cancer initiation and development (which, along with assumed hereditary effects constitute stochastic radiation effects) associated with moderate- and high-dose low-LET ionizing radiation (e.g., X ray) exposure are well known (1). There is abundant evidence that moderate doses (0.1–1 Gy) and high doses (>1 Gy) (19) of sparsely ionizing low-LET radiation (e.g., X rays, γ rays), particularly when received at a high dose-rate, are associated with elevated cancer risks (1, 2, 25, 26). Reduced statistical power means that less is known about the risks arising from exposures at low doses (<0.1 Gy) and low dose rates (<5 mGy/h). Many regulatory bodies assume that at sufficiently low doses there is an increasing linear component to the dose response for stochastic effects, i.e., that there is a positive correlation of risk with dose, with no threshold, or beneficial effect of radiation exposure (2). However, there is accumulating direct evidence of excess risk of cancer in a number of populations exposed to low doses. Some of these data are summarized in Table 1. We review some of this evidence below. A more comprehensive review of the findings from various radio-epidemiological studies of cancer is provided by Rühm et al (27).

TABLE 1.

Cancer Risks of Childhood or In Utero Exposure in Selected Higher Quality Studies of Medical Diagnostic Radiation Exposure [taken in part from Little et al. (19)]

Description of study data	Ref.	Mean dose (range) (Gy)	Persons (person years of follow-up)	Notes	Endpoint [incidence unless otherwise stated]	Number of cases/deaths	ERR/Gy (95% CI)
Exposure in childhood
Pooled analysis of 9 datasets	Lubin et al. (67)	0.02991 (0−<0.2)	107,594 (4,454,516)	Dose < 0.2 Gy	Thyroid cancer	252	11.1 (6.6,19.7)
		0.01730 (0−<0.1)	96,318 (3,980,642)	Dose < 0.1 Gy	Thyroid cancer	184	9.6 (3.7,17.0)
Nine cohort pooled moderate dose medical+LSS analysis - dose < 0.1 Gy	Little et al. (30)	0.0196 (0−<0.1)	262,573 (5,154,464)		Acute myeloid leukemia + MDS	87	20.9 (4.1,49.2)
					Acute lymphoblastic leukemia	40	46.6 (3.5,187.1)
					Chronic myeloid leukemia	36	−6.4 (<−10,13.6)
					Leukemia excluding CLL	221	8.4 (−0.3,20.8)
Finnish Cancer Registry based case-control study of computed tomography (CT) 1990–2011	Nikkilä et al. (39)	0.00629 (0−<0.0332)	1093 cases, 3279 controls	Median dose for controls, using NCICT software	Leukemia	1093	130 (20,260)
Dutch CT study of children (age <18 years) at first CT, 1979–2012	Meulepas et al. (38)	0.0385 (0−>0.22)	168,394 (1,201,357)	Exclusion and lag 5 years, brain dose	Brain/CNS	84	8.6 (2.0,22.2)
		0.0095 (0−>0.017)		Exclusion and lag 2 years, ABM dose	Leukemia	44	2.1 (−1.2,24.0)
		0.0095 (0−>0.017)		Exclusion and lag 2 years, ABM dose	Leukemia + MDS	63	0.4 (−1.2,16.1)
French CT study of children (age <10) at first CT in years 2000–2011, followed up 2012–2016	Foucault et al. (40)	0.0103 (SD 0.0128)	100,560 (NA)	Lag and exclusion 2 years, ABM dose	Leukemia	39	16 (7, 26)
		0.0277 (SD 0.0392)		Lag 5 y, exclusion 2 years, brain dose	Brain/CNS	75	6 (2, 9)
		0.0103 (SD 0.0128)		Lag and exclusion 2 years, ABM dose	Lymphoma	41	−11 (−39, 30)
Meta-analysis of childhood CT studies	Berrington de Gonzalez et al. (68)	0.006–0.012 (0−>0.1)	NA	5 studies in relation to ABM dose	Leukemia	1259	10.5 (−5.8, 26.9)
		0.018–0.043 (0−>0.1)	NA	5 studies in relation to brain dose	Brain/CNS (including meningioma)	344	7.9 (4.7, 11.1)
Exposure in utero
Oxford Survey of Childhood Cancers, case-control pairs born 1953–1979	Bithell (69)	NA (0−<0.03)	14,759 cases and matched controlsa	Estimated via log linear-quadratic model of OR fitted to data of Gilman et al (70) by year for 1959, using dose estimate of 6.1 mGy per obstetric radiograph of Mole (71) for that year	All cancer mortality	14,759	51 (28, 76)

Abbreviations: CI, confidence interval; MDS, myelodysplastic syndrome; CLL, chronic lymphocytic leukemia; NCICT, National Cancer Institute Dosimetry System for Computed Tomography; CNS, central nervous system; ABM, active bone marrow; OR, odds ratio.

^aNumbers taken from Gilman et al. (70).

One of the most important sources of information on radiation risks is a study of the survivors of the atomic bombings of Hiroshima and Nagasaki, a cohort of about 120,000 persons identified via information collected from the 1950 Japanese national census and assembled in the early- to mid-1950s, i.e., 5–10 years after the bombings. Despite what is often thought, the mean dose in the Japanese atomic bomb survivor Life Span Study (LSS) cohort is quite low, about 0.1 Gy, with many analyses restricted to 4 Gy or less (28, 29). The most recent analyses of the Japanese atomic bomb survivor LSS incidence data suggest that there is significant excess risk of all solid cancers for assessed doses of less than 0.1 Gy (29). A combined analysis of data for leukemia and myeloid neoplasms among groups exposed in childhood in the LSS and elsewhere found evidence of significant excess risk of all myeloid malignant neoplasms under 100 mSv,³ and for acute lymphoblastic leukemia under 20 mSv (30) (see also Table 1).

Another important source of information on radiation risks is studies of radiation workers, i.e., of those exposed to radiation in the course of their work in the nuclear industry or elsewhere. One of the most important such studies is the International Nuclear Workers Study (INWORKS), which included over 300,000 workers with a mean cumulative exposure of 20.9 mGy (31). Although not a low-dose study (the maximum cumulative dose is about 1.3 Gy), the exposures are all at low dose rate and yield significant excess risks of solid cancer and leukemia (31, 32).

Many of the low-dose studies cited in Table 1 yield significant excess risk for various cancer endpoints, strongly suggesting that risk at low doses is not zero. It is also clear from comparison of the excess relative risks per Gy (ERR/Gy) given in Table 1 that they are consistent with each other and with ERR/Gy that can be derived from the LSS. They would not be consistent with risks several orders of magnitude higher than those derived from the LSS, as has been suggested by various researchers (13-16).

In particular, there is evidence of excess risk of most types of childhood cancer associated with radiation exposures of the order of 10–20 mGy from diagnostic X-ray exposure in the Oxford Survey of Childhood Cancers and in various other groups exposed in utero (20, 33, 34) (see Table 1). While these data are not yet universally accepted, Wakeford and Little note “the consistency of the childhood cancer risk coefficients derived from the Oxford Survey and from the Japanese cohort irradiated in utero supports a causal explanation of the association between childhood cancer and an antenatal X-ray examination found in case-control studies. This implies that doses to the foetus in utero of the order of 10 mSv discernibly increase the risk of childhood cancer” (33). There are also a number of studies of childhood cancer and natural background radiation exposure, at doses of the order of 10–20 mGy, suggesting excess risk for leukemia and brain cancer (35, 36). At slightly higher doses, increased risks of leukemia and brain cancer have been observed in pediatrically-exposed groups given multiple computed tomography (CT) examinations, at doses of about 60 mGy to the respective tissues (active bone marrow, brain) (37-40). Again, the excess risks in all these studies are consistent with each other and with those observed among the Japanese atomic bomb survivors (33, 35-39).

The health risks of low-level exposure to ionizing radiation are most commonly assumed to be associated primarily with cancer (1). However, there is evidence of excess CVD risk in a number of moderate dose (<5 Gy) exposed groups, including the Japanese atomic bomb survivors (41, 42). Evidence has recently emerged of an association between lower doses (<0.5 Gy) and CVD, in particular in a number of groups of nuclear workers (43, 44). This has been reinforced by conclusions of a number of recent (systematic and non-systematic) reviews, all suggesting an excess radiation-associated CVD risk at occupational and environmental dose levels (<0.5 Gy) (4, 5, 45) (see also Table 2). However, the presence and magnitude of the excess CVD risk at low doses is still relatively controversial, largely due to the difficulties in accurately assessing the role of confounding exposures and other contributory risk factors for CVD. Interstudy heterogeneity complicates a causal interpretation of the observed risks, so that much remains unknown as to the shape of the dose response (4, 5, 46), if indeed the observed trends represent causal relationships.

TABLE 2.

Cardiovascular Disease Risks in Selected Higher Quality Diagnostically Exposed Groups, in the Japanese Atomic Bomb Survivors and in Certain Higher Quality Occupationally and Environmentally Exposed Groups, taken from Little et al (5). All Data are In Relation to Underlying Cause of Death, Unless Otherwise Indicated

Description of study data	Ref.	Mean (range) heart/brain dose (Gy)	Persons (person years of follow-up)	Endpoint (mortality unless otherwise indicated)	Cases/deaths	Excess relative risk Gy−1 (95% CI)
Diagnostically exposed groups
Canadian and Massachusetts TB fluoroscopy cohorts	Tran et al. (72)	0.18 (0–0.50) [<0.5 Gy]/1.16 (0–27.77) [total]	77,275 (1,945,041)	All cardiovascular disease ICD9 390–459: <0.5 Gy	10,209	0.246 (0.036, 0.469)a
				Ischemic heart disease ICD9 410–414: < 0.5 Gy	6410	0.268 (0.003, 0.552)
				Cerebrovascular disease ICD9 430–438: < 0.5 Gy	1561	0.441 (−0.119, 1.090)a
Japanese atomic bomb survivors
Japanese atomic bomb survivors 1950–2003	Shimizu et al. (41)	0.1 (0–4)b	86,611 (NA)	Cerebrovascular disease total (ICD9 430–438)	12,139	0.12 (0.05, 0.19)c
				Heart disease total (ICD9 393-429 excluding 401, 403, 405)	14,018	0.18 (0.11, 0.25)c
				Circulatory disease apart from heart disease and stroke (ICD9 390-392, 401, 403, 405, 439–459)	5846	0.58 (0.45, 0.72)c
				All cardiovascular disease (ICD9 390–459)	25,113	0.15 (0.10, 0.20)c
Japanese atomic bomb survivors 1950–2008	Takahashi et al. (42)	0.1 (0–4)b	86,600 (3,462,847)	Heart disease (ICD10 I05–I08, I09.1, I11, I13, I20–25, I34–I39, I50) overall	9303	0.140 (0.060, 0.220)
				Ischemic heart disease (ICD10 I20–I25)	3556	0.030 (−0.080, 0.150)
				Myocardial infarction (ICD10 I21–I23)	1883	0.020 (−0.130, 0.200)
				Other ischemic heart disease (ICD10 I20, I24–I25)	1673	0.040 (−0.120, 0.220)
				Valvular heart disease (ICD10 I05–I08, I09.1, I34–I39)	744	0.450 (0.130, 0.850)
				Rheumatic valvular heart disease (ICD10 I05–I08, I09.1)	223	0.960 (0.280, 1.920)
				Non-rheumatic valvular heart disease (I34–I39)	521	0.240 (−0.080, 0.680)
				Hypertensive organ damage (ICD10 I11–I13)	1122	0.360 (0.100, 0.680)
				Heart failure (ICD10 I50)	3334	0.210 (0.070, 0.370)
Occupational studies
International Nuclear Workers Study (INWORKS)	Gillies et al. (43)	0.0252 (0–1.932)	308,297 (8.2 × 106)	Cardiovascular disease (ICD10 I00–I99)	27,848	0.22 (0.08, 0.37)d
				Ischemic heart disease (I20–I25)	17,463	0.18 (0.004, 0.36)d
				Acute myocardial infarction (I21)	11,076	0.26 (0.03, 0.51)d
				Chronic ischemic heart disease (I25)	6238	0.07 (−0.19, 0.36)d
				Cerebrovascular disease (I60–I69)	4444	0.50 (0.12, 0.94)d
Mayak workers	Moseeva et al. (73); Azizova et al. (74)	0.62 ± 0.80 (males)e	22,377 (447,281)	Ischemic heart disease morbidity (ICD9 410–414)	7225	0.14 (0.08, 0.21)a
		0.51 ± 0.68 (females)e	22,377 (836,048)	Ischemic heart disease mortality (ICD9 410–414)	2848	0.05 (−0.01, 0.13)a
			18,856 (341,663)	Cerebrovascular disease morbidity (ICD9 430–438)	7440	0.497 (0.393, 0.601)a
			18,856 (272,525)	Cerebrovascular disease mortality (ICD9 430–438)	1382	0.057 (−0.046, 0.161)a
Mayak workers stroke subtypes	Azizova et al. (6)	NA	22,377 (459,520)	Cerebrovascular disease morbidity (ICD10 I60–I69)	9469	0.39 (0.31, 0.48)f
			22,377 (573,781)	Stroke morbidity (ICD10 I60–I64)	2078	0.00 (−0.08, 0.10)f
			22,377 (593,051)	Hemorrhagic stroke morbidity (ICD10 I61)	262	0.10 (−0.16, 0.58)f
			22,377 (585,806)	Ischemic stroke morbidity (ICD10 I63)	1611	0.06 (−0.04, 0.20)f
UK NRRW heart disease	Zhang et al. (75)	0.0232 (0−>0.4)	174,541 (NA)	All heart disease (ICD9 393–398, 402, 404, 410–429)	11,014	0.37 (0.11, 0.65)
UK NRRW stroke	Hinksman et al. (44)	0.0031g (0–1.9)	166,812 (3,665,413)	Cerebrovascular disease mortality (ICD9 430–438)	3219	0.57 (0.00, 1.31)
				Ischemic stroke mortality (ICD9 433-435)	422	1.03 (−0.28, 3.48)
				Hemorrhagic stroke mortality (ICD9 430–432)	666	1.06 (−0.52, 2.01)
				Ill-defined and other cerebrovascular disease mortality (ICD9 436–438)	2131	0.54 (−0.10, 1.42)
Environmental studies
Techa River study	Krestinina et al. (76)	0.035 (0–0.51)h	29,735 (901,563)	All cardiovascular disease mortality (ICD9 390–459)	7595	0.18 (−0.13, 0.52)a, h
				Ischemic heart disease mortality (ICD9 410–414)	3194	0.26 (−0.22, 0.81)a, h

Abbreviations: CI, Confidence Interval; ICD, International Classification of Diseases; NRRW, National Registry for Radiation Workers; TB, tuberculosis.

^aAssuming a lag period of 5 years.

^bAnalysis based on colon dose.

^cAnalysis using underlying or contributing cause of death.

^d90% CI.

^eRisk estimates in relation to cumulative whole body external gamma dose; doses given here are from Moseeva et al. (73).

^fAssuming a lag period of 10 years.

^gMedian dose.

^hAnalysis based on dose to muscle.

Although there are long-established risks of cataract at high doses (47), there is now a considerable body of evidence of excess risk of cataract at moderate levels of dose (<5 Gy) (7, 48), and some large and well powered occupational studies suggesting excess risk at <0.1 Gy (8) (see also Table 3). The cataract risks derived from various studies are reasonably consistent with each other (Table 3). However, most of the studies [all except Little et al. (8)] are not at exposure levels that can truly be defined as low dose (<0.1 Gy), although many are at low dose rate (7, 8, 49-51).

TABLE 3.

Risks for Cataract in Selected Higher Quality Radiation-Exposed Cohorts, Taken from Little et al. (5)

Description of study data	Ref.	Dose (Gy), mean (range)	Persons (person years of follow-up)	Notes (method of ascertainment)	Endpoint	Cases	Excess hazard ratio Gy−1 or excess odds ratio Gy−1 (95% CI)
Japanese atomic bomb survivor AHS	Nakashima et al. (77)	0.522 (0–4.94)	730 (NA)	LOCS II	Cortical	618a	0.30 (0.10, 0.53)
					Posterior subcapsular	214a	0.44 (0.19, 0.73)
					Nuclear opacity	415a	0.07 (−0.11, 0.30)
					Nuclear colour	358a	0.01 (−0.17, 0.24)
Chernobyl recovery worker	Worgul et al. (49)	NA (0−>1.095)	8607 (NA)	Merriam-Focht	Non-nuclear stage 1–5	3369b,c/274d	0.65 (0.18, 1.30)
					Posterior subcapsular stage 1	2781b,c/252d	0.42 (0.01, 1.00)
					Nuclear	382b,c/113d	0.07 (−0.44, 1.04)
					All cataract stage 1–5	3751b,c/384d	0.70 (0.22, 1.38)
Japanese atomic bomb survivor AHS cataract surgery	Neriishi et al. (48)	0.500 (0–5.14)	6066 (84,209)	Surgical removal	All cataract removal	1028	0.32 (0.17, 0.52)d, e
US Radiologic Technologists	Little et al. (8)	0.056 (0–1.514)	67,246 (832,479)	Questionnaire	Cataract history	12,366	0.69 (0.27, 1.16)
			67,709 (888,420)		Cataract surgery	5509	0.34 (−0.19, 0.97)
Mayak nuclear workers	Azizova et al. (7)	0.526 (0−>2.0)	22,377 (486,245)	Slit lamp exam	Cortical	3132	0.62 (0.50, 0.75)f
			22,377 (489,162)		Posterior subcapsular	1239	0.90 (0.67, 1.19)f
			22,377 (492,004)		Nuclear	2033	0.47 (0.35, 0.60)f
Mayak nuclear workers	Azizova et al. (50)	0.515 (0 - >2.0)	22,377 (495,868)	Slit lamp exam	Cataract surgery	701	0.09 (−0.02, 0.22)f
Chinese high natural background area	Su et al. (51)	NA (0.0221 - 0.3104)	941 (NA)	LOCS III	Cortical	101	2.6 (0.0, 6.0)
					Posterior subcapsular	23	7.3 (0.5, 18.5)
					Nuclear	245	−1.9 (−3.6, 0.1)

Abbreviations: CI, confidence interval; AHS, Adult Health Study; LOCS, Lens Opacities Classification System

^aAll cases with LOCS II grade I and above.

^bSummed over cataracts in left and right eyes.

^cPrevalent cataract.

^dIncident cataract.

^eAdjusted to persons in Hiroshima, aged 70, exposed at age 20 years.

^fLag period of 5 years.

Radiobiological Considerations

There are data, reviewed elsewhere (52), suggesting an increase in stable chromosome aberrations and other markers of biological damage in the peripheral blood lymphocytes of nuclear workers and other groups with protracted radiation exposures. Chromosome changes play a major role in carcinogenesis (the process by which normal cells are transformed into cancer cells) and there is mounting evidence that the presence of increased frequencies of chromosome aberrations in peripheral blood lymphocytes in healthy individuals could be a surrogate for the specific changes associated with carcinogenesis and, therefore, indicative of cancer risk (53-57).

Cancer is thought to result from mutagenic damage to a single cell, specifically to its nuclear DNA, which in principle could be caused by clustered single-strand breaks (SSB) which result in a double-strand break (DSB) of the DNA, as well as DNA-replication processing of SSBs that lead to DNA DSB (58); this argues against the existence of a threshold of dose below which cancer risk is not elevated, as discussed elsewhere (52). A more recent evaluation of the biological mechanisms relevant for low dose radiation cancer risk inference concluded that “There remains good justification for the use of a non-threshold model for risk inference for radiation protection purposes, given the present robust knowledge on the role of mutation and chromosomal aberrations in carcinogenesis” and, in relation to the potential targets in addition to nuclear DNA, “The potential contributions of phenomena such as transmissible genomic instability, bystander phenomena, induction of abscopal effects and adaptive response remain unclear.” (24). As shown in Table 4, for orthovoltage (250 kV) X rays with various degrees of standard filtration irradiating cells having a mean 4–10 μm diameter nucleus (a reasonable range), a radiation dose of 1 mGy (0.001 Gy) corresponds to between 0.051 to 0.53 electron tracks traversing the cell nucleus (59). Table 4 also demonstrates that the number of electron tracks per cell nucleus are slightly lower, with a range of 0.046 to 0.39, for the lower radiation energy (65 kV) X rays that were likely used obstetrically in the 1950s (59). This suggests that at low doses (0.01 Gy or less spread over a year), it is unlikely that temporally and spatially separate electron tracks could cooperatively produce DNA damage (60), so that in this very low-dose region, DNA damage at a cellular level would be proportional to dose.

TABLE 4.

Mean Number of Electron Tracks Traversing a Cell Nucleus for Given Mean Dose, as Function of X-Ray Energy/Filtration and Nuclear Diameter, as Evaluated by Goodhead (59) from Measurements of Frequency Mean Lineal Energy of Braby and Ellett as Summarized by Booz (78)

Cell nuclear diameter (μm)	Radiation energy, filtration	Mean nuclear electron tracks per 1 mGy	Mean nuclear electron tracks per 10 mGy
4	65 kV 1.9 mm Al HVL	0.046	0.46
4.5	65 kV 1.9 mm Al HVL	0.061	0.61
5	65 kV 1.9 mm Al HVL	0.078	0.78
6	65 kV 1.9 mm Al HVL	0.12	1.2
8	65 kV 1.9 mm Al HVL	0.23	2.3
10	65 kV 1.9 mm Al HVL	0.39	3.9
4	250 kV 0.44 mm Cu HVL	0.051	0.51
4.5	250 kV 0.44 mm Cu HVL	0.067	0.67
5	250 kV 0.44 mm Cu HVL	0.086	0.86
6	250 kV 0.44 mm Cu HVL	0.13	1.3
8	250 kV 0.44 mm Cu HVL	0.26	2.6
10	250 kV 0.44 mm Cu HVL	0.43	4.3
4	250 kV 1.77 mm Cu HVL	0.073	0.73
4.5	250 kV 1.77 mm Cu HVL	0.095	0.95
5	250 kV 1.77 mm Cu HVL	0.12	1.2
6	250 kV 1.77 mm Cu HVL	0.17	1.7
8	250 kV 1.77 mm Cu HVL	0.33	3.3
10	250 kV 1.77 mm Cu HVL	0.53	5.3

Abbreviations: HVL, half value layer; Al, aluminum; Cu, copper.

Cells have substantial repair mechanisms. It is known that the efficiency of cellular repair processes varies with dose and dose rate (61, 62), and this may be the reason for the curvature that is observed in the cancer dose response at higher levels of dose [e.g., for leukemia (63) and some solid cancers (28)] and dose rate effects observed in epidemiological (1) and animal (61, 64, 65) data. However, none of these repair processes are 100% efficient, so after mutagenic damage there is a non-zero probability of a damaged cell surviving with unrepaired damage, that may manifest later as cancer. Here, we point out that not all radiation protection theory is based on a simple linear relationship, indeed the idea of non-linearity in biological response is clearly implied by use of concepts such as the dose and dose rate effectiveness factor (DDREF) (2) and, thus, is actually more complex than implied by some (12).

Some Considerations on Interpretation of Epidemiologic Studies

Not all epidemiological studies have equal degrees of validity or generalizability and, for that very reason, academic, research, and other expert institutions like the National Academy of Sciences (NAS), the International Agency for Research on Cancer (IARC/WHO), the United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) routinely examine the evidence on radiation-associated health risk and produce a group consensus opinion that weighs the strengths and limitations of the many published studies that contribute to the total knowledge base on radiation health risks. Conclusions about the nature of radiation-associated health risk should not be drawn from single studies, but by the overall weight of evidence. In this kind of evaluation, study findings are weighted by specific criteria including type of study (cohort, case-control, randomized trial, correlational), population (sample) size, degree of control of bias and confounding, statistical methods used for analysis, use of pooled- or meta-analyses, uncertainties in diagnoses and estimated exposures, and the degree to which specific and well-known criteria for causality have been satisfied. Readers are referred to, for example, discussions on these issues by NAS (25), IARC (26) and UNSCEAR (1). Focusing on just a few studies can easily lead to unreliable conclusions, whether in the direction of underestimating risk (12) or substantially overestimating risk (13-16); that neither extreme position is tenable is strongly suggested by review of the totality of epidemiological data, as for example shown in part in Table 1. The requirement to understand the theoretical bases as well as the limitations of epidemiology cannot be over-emphasized for those attempting to derive conclusions about the existence as well as the magnitude of radiation health risks.

CONCLUSIONS

Based on the data and explanations we have provided, we believe that the arguments proposed by some that LNT overestimates low-dose cancer risk (12, 66) are likely to be grossly invalid. Likewise, the overall body of epidemiologic data are clearly inconsistent with cancer risks substantially higher than those implied by LNT, as has been suggested by others (13-16).

We have presented evidence that excess cancer risks have been noted in a number of (largely pediatrically or in utero exposed) groups exposed to low radiation doses (<0.1 Gy) (19, 20). The available data on biological mechanisms do not provide general support for the idea of a low-dose threshold or hormesis for any of these endpoints (24, 61, 62). This large body of evidence does not suggest and, indeed, is not statistically compatible with any very large threshold in dose (>10 mGy), or with possible beneficial effects from exposures.