Low- and moderate-dose non-cancer effects of ionizing radiation in directly exposed individuals, especially circulatory and ocular diseases: a review of the epidemiology

Abstract

Purpose:

There are well-known correlations between high and moderate doses (>0.5 Gy) of ionizing radiation exposure and circulatory system damage, also between radiation and posterior subcapsular cataract. At lower doses correlations with circulatory disease are emerging in the Japanese atomic bomb survivors and in some occupationally exposed groups, and are still to some extent controversial. Heterogeneity in excess relative risks per unit dose in epidemiological studies at low (<0.1 Gy) at low-moderate (>0.1 Gy, <0.5 Gy) doses may result from confounding and other types of bias, and effect modification by established risk factors. There is also accumulating evidence of excess cataract risks at lower dose and low dose rate in various cohorts. Other ocular endpoints, specifically glaucoma and macular degeneration have been little studied. In this paper we review recent epidemiological findings, and also discuss some of the underlying radiobiology of these conditions. We briefly review some other types of mainly neurological non-malignant disease in relation to radiation exposure.

Conclusions:

We document statistically significant excess risk of the major types of circulatory disease, specifically ischemic heart disease and stroke, in moderate- or low-dose exposed groups, with some not altogether consistent evidence suggesting dose response non-linearity, particularly for stroke. However, the patterns of risk reported are not straightforward. We also document evidence of excess risks at lower doses/dose-rates of posterior subcapsular and cortical cataract in the Chernobyl liquidators, US Radiologic Technologists and Russian Mayak nuclear workers, with fundamentally linear dose response. Nuclear cataracts are less radiogenic. For other ocular endpoints, specifically glaucoma and macular degeneration there is very little evidence of effects at low doses; radiation-associated glaucoma has been documented only for doses >5 Gy, and so has the characteristics of a tissue reaction. There is some evidence of neurological detriment following low-moderate dose (~0.1-0.2 Gy) radiation exposure in utero or in early childhood.

1. Introduction

Cancer risk has been thought to be the main health risk of low-level ionizing radiation exposure (Committee to Assess Health Risks from Exposure to Low Levels of Ionizing Radiation 2006; United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008a). Leukemia was observed within 5 years of the Hiroshima and Nagasaki atomic bombings (Folley et al. 1952), but excesses of most other cancer types have also been observed in the Life Span Study (LSS) cohort of Japanese atomic bomb survivors (Ozasa et al. 2012; Grant et al. 2017) and elsewhere (United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008a). At very high radiation doses (> 5 Gy) damage to many organs and tissues occurs, leading to a variety of tissue reaction (formerly deterministic) effects (Edwards and Lloyd 1998; International Commission on Radiological Protection 2012).¹ The damage to parts of the heart and large arteries occurs within months of exposure (Adams et al. 2003; Hamada et al. 2020). Various recent systematic reviews document excess risk of circulatory disease associated with much lower doses (< 0.5 Gy) (McGale and Darby 2005; Little et al. 2008; McGale and Darby 2008; United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008b; McMillan et al. 2010; Little, Azizova, et al. 2012; Bernstein et al. 2020) although this remains controversial, with some suggesting that risk becomes negligible below doses of 0.5 Gy (International Commission on Radiological Protection 2012; National Council on Radiation Protection and Measurements (NCRP) 2018) or more (Schöllnberger et al. 2012).

High radiation doses (>1 Gy) are associated with posterior subcapsular cataract (PSC) (Edwards and Lloyd 1998). In the atomic bomb survivors (Minamoto et al. 2004; Neriishi et al. 2012), Chernobyl-exposed groups (Worgul et al. 2007), US astronauts (Chylack et al. 2009; Chylack et al. 2012), Mayak nuclear workers (Azizova, Hamada, et al. 2018) and some other exposed populations [reviewed by (Ainsbury et al. 2009; Hammer et al. 2013; Little 2013; Shore 2016)] cortical cataracts (CC) are also associated with radiation exposure; nuclear cataracts (NC) appear to be more weakly radiogenic (Rafnsson et al. 2005; Azizova, Hamada, et al. 2018). There is very little evidence that other types of ocular disease, in particular glaucoma and macular degeneration are radiation inducible (Little M. P. et al. 2018; Hamada et al. 2019).

In this paper we present evidence for associations between moderate-/low-level radiation exposure and certain types of non-malignant diseases in the directly exposed individuals, with particular emphasis on diseases of the circulatory system. This paper is not a systematic review, unlike some others (McGale and Darby 2005; Little et al. 2008; Little, Azizova, et al. 2012), although circulatory disease is reviewed systematically. We aim to update previous circulatory disease reviews and those of cataract (Ainsbury et al. 2009; Hammer et al. 2013; Little 2013). We shall concentrate attention in the main text on studies published since the last such review of non-cancer effects conducted by the first author (Little 2013); however the relevant Tables 1-3 include all studies. The systematic review of circulatory disease given here is preparatory to a more wide-ranging review that will look at all studies irrespective of dose range, will use more than one type of literature database search, and will also incorporate various types of meta-analysis.

Table 1.

Estimated excess relative risks of circulatory disease in diagnostically exposed groups, in the Japanese atomic bomb survivors and in other groups with other than very high dose radiation exposure (< 5 Gy), and with mean dose generally low-moderate or low (< 0.5 Gy), or with hyperfractionated delivery of dose. All data are in relation to underlying cause of death, unless otherwise indicated.

Cohort/Study	Reference	Mean (range) heart/brain dose, Gy	Persons (person years of follow-up)	Endpoint (mortality unless otherwise indicated)	Excess relative risk Gy−1 (95% CI)
Diagnostically exposed groups
Canadian and Massachusetts TB fluoroscopy cohorts	Tran et al (Tran et al. 2017)	0.18 (0 - 0.50) [<0.5 Gy] / 1.16 (0 - 27.77) [total]	77,275 (1,945,041)	All circulatory disease ICD9 390-459	−0.024 (−0.042, −0.005)a
				All circulatory disease ICD9 390-459: <0.5 Gy	0.246 (0.036, 0.469)a
				Ischemic heart disease ICD9 410-414	−0.037 (−0.060, −0.013)a
				Ischemic heart disease ICD9 410-414: < 0.5 Gy	0.268 (0.003, 0.552)
				Cerebrovascular disease ICD9 430-438	−0.014 (−0.067, 0.044)a
				Cerebrovascular disease ICD9 430-438: < 0.5 Gy	0.441 (−0.119, 1.090)a
				Hypertensive heart disease ICD9 401-405	−0.035 (−0.152, 0.153)a
				Hypertensive heart disease ICD9 401-405: < 0.5 Gy	1.121 (−0.351, 3.228)a
				Heart disease apart from hypertensive and IHD ICD9 390-400, 406-410	−0.010 (−0.064, 0.043)a
				Heart disease apart from hypertensive and IHD ICD9 390-400, 406-410: < 0.5 Gy	−0.226 (−0.679, 0.307)a
				All circulatory disease apart from heart and cerebrovascular ICD9 439-459	0.055 (−0.028, 0.164)a
				All circulatory disease apart from heart and cerebrovascular ICD9 439-459: < 0.5 Gy	0.507 (−0.322, 1.541)a
Japanese atomic bomb survivors
Japanese atomic bomb survivors	Shimizu et al (Shimizu et al. 2010)	0.1 (0 - 4)b	86,611 (n.a.)	Ischemic heart disease (ICD9 410-414)	0.02 (−0.10, 0.15)
				Myocardial infarction (ICD9 410)	0.00 (−0.15, 0.18)
				Hypertensive heart disease (ICD9 402, 404)	0.37 (0.08, 0.72)
				Rheumatic heart disease (ICD9 393-398)	0.86 (0.25, 1.72)
				Heart failure (ICD9 428)	0.22 (0.07, 0.39)
				Other heart disease (ICD9 390-392, 415-427, 429)	−0.01 (−0.21, 0.24)
				Hypertensive disease without heart disease (ICD9 401, 403, 405)	0.07 (−0.22, 0.55)
				Heart disease total (ICD9 393-429 excluding 401, 403, 405)	0.18 (0.11, 0.25)c
				Cerebral infarction (ICD9 433,434)	0.04 (−0.10, 0.20)
				Cerebral hemorrhage (ICD9 431)	0.05 (−0.06, 0.17)
				Subarachnoid hemorrhage (ICD9 430)	0.30 (−0.04, 0.76)
				Other or unspecified cerebrovascular disease	0.16 (0.01, 0.34)
				Cerebrovascular disease total (ICD9 430-438)	0.12 (0.05, 0.19)c
				Circulatory disease apart from heart disease and stroke (ICD9 390-392, 401, 403, 405, 439-459)	0.58 (0.45, 0.72)c
				Other circulatory disease (ICD9 399-400, 406-409, 439-459)	−0.01 (<−0.01, 0.34)
				All circulatory disease (ICD9 390-459)	0.15 (0.10, 0.20)c
Japanese atomic bomb survivors	Yamada et al (Yamada et al. 2004)	0.1 (0 - 4)d	10,339 (n.a.)	Hypertension incidence, 1958-1998 (ICD9 401)	0.05 (−0.01, 0.10)d
				Hypertensive heart disease incidence, 1958-1998 (ICD9 402, 404)	−0.01 (−0.09, 0.09)d
				Ischemic heart disease incidence, 1958-1998 (ICD9 410-414)	0.05 (−0.05, 0.16)d
				Myocardial infarction incidence, 1964-1998 (ICD9 410)	0.12 (−0.16, 0.60)d
				Occlusion incidence, 1958-1998 (ICD9 433, 434)	0.06 (−0.11, 0.30)d
				Aortic aneurysm incidence, 1958-1998 (ICD9 441, 442)	0.02 (−0.22, 0.41)d
				Stroke incidence, 1958-1998 (ICD9 430, 431, 433, 434, 436)	0.07 (−0.08, 0.24)d
Japanese atomic bomb survivors in utero	Tatsukawa et al (Tatsukawa et al. 2008)	0.001 (0 - 1.79)	506 (9,265)	Morbidity in utero: hypertension	0.20 (−0.39, 1.38)
				Morbidity in utero: nonfatal stroke or myocardial infarction	−0.91 (−1.00, 79.3)
Japanese atomic bomb survivors in childhood	Tatsukawa et al (Tatsukawa et al. 2008)	0.13 (0 - 3.53)	1,053 (20,216)	Morbidity: hypertension	0.15 (−0.01, 0.34)
				Morbidity: stroke or myocardial infarction	0.72 (0.24, 1.40)
Japanese atomic bomb survivors 1950-2008	Takahashi et al (Takahashi et al. 2017)	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	0.140 (0.060, 0.220)
				Ischemic heart disease (ICD10 I20-I25)	0.030 (−0.080, 0.150)
				Myocardial infarction (ICD10 I21-I23)	0.020 (−0.130, 0.200)
				Other ischemic heart disease (ICD10 I20, I24-I25)	0.040 (−0.120, 0.220)
				Valvular heart disease (ICD10 I05–I08, I09.1, I34–I39)	0.450 (0.130, 0.850)
				Rheumatic valvular heart disease (ICD10 I05–I08, I09.1)	0.960 (0.280, 1.920)
				Non-rheumatic valvular heart disease (I34–I39)	0.240 (−0.080, 0.680)
				Hypertensive organ damage (ICD10 I11-I13)	0.360 (0.100, 0.680)
				Heart failure (ICD10 I50)	0.210 (0.070, 0.370)
Occupational studies
International Agency for Research on Cancer15-country nuclear worker study	Vrijheid et al (Vrijheid et al. 2007)	0.0207 (0.0 - >0.5)	275,312 (4,067,861)	Circulatory disease (ICD10 I00-I99, J60-J69, O88.2, R00-R02, R57)	0.09 (−0.43, 0.70)e
				Ischemic heart disease (ICD10 I20-I25)	−0.01 (−0.59, 0.69)e
				Heart failure (ICD10 I50)	−0.03 (<0, 4.91)e
				Deep vein thrombosis and pulmonary embolism (ICD10 I26, I80, I82, O88.2)	−0.95 (−1.00, 9.09)e, f
				Cerebrovascular disease (ICD10 I60-I69)	0.88 (−0.67, 3.16)e
				All other circulatory disease (ICD10 R00-R02, R57, I00-I99 excluding I20-26, I50, I60-69, I80, I82)	0.29 (<0, 2.40)e
International Nuclear Workers Study (INWORKS)	Gillies et al (Gillies et al. 2017)	0.0252 (0 - 1.932)	308,297 (8.2 x 106)	Circulatory disease (ICD10 I00-I99)	0.22 (0.08, 0.37)g
				Ischemic heart disease (I20-I25)	0.18 (0.004, 0.36)g
				Acute myocardial infarction (I21)	0.26 (0.03, 0.51)g
				Chronic ischemic heart disease (I25)	0.07 (−0.19, 0.36)g
				Cerebrovascular disease (I60-I69)	0.50 (0.12, 0.94)g
Mayak workers	Moseeva et al (Moseeva et al. 2014) Azizova et al (Azizova, Grigoryeva, et al. 2015)	0.62 ± 0.80 (males)h 0.51 ± 0.68 (females)g	22,377 (447,281)	Ischemic heart disease morbidity (ICD9 410-414)	0.14 (0.08, 0.21)a
				Ischemic heart disease morbidity (ICD9 410-414)	0.14 (0.08, 0.21)e
				Ischemic heart disease morbidity (ICD9 410-414)	0.16 (0.10, 0.24)i
			22,377 (836,048)	Ischemic heart disease mortality (ICD9 410-414)	0.05 (−0.01, 0.13)a
				Ischemic heart disease mortality (ICD9 410-414)	0.05 (−0.01, 0.13)e
				Ischemic heart disease mortality (ICD9 410-414)	0.05 (−0.01, 0.13)i
			18,856 (341,663)	Cerebrovascular disease morbidity (ICD9 430-438)	0.497 (0.393, 0.601)a
				Cerebrovascular disease morbidity (ICD9 430-438)	0.529 (0.415, 0.642)e
				Cerebrovascular disease morbidity (ICD9 430-438)	0.572 (0.450, 0.695)i
			18,856 (272,525)	Cerebrovascular disease mortality (ICD9 430-438)	0.057 (−0.046, 0.161)a
				Cerebrovascular disease mortality (ICD9 430-438)	0.064 (−0.042, 0.170)e
				Cerebrovascular disease mortality (ICD9 430-438)	0.076 (−0.033, 0.186)i
Mayak nuclear workers lower extremity arterial disease	Azizova et al (Azizova et al. 2016)	0.51 ± 0.72	22,377 (512,801)	Lower extremity arterial disease morbidity (ICD9 440.2)	0.30 (0.13, 0.53)a
				Lower extremity arterial disease morbidity (ICD9 440.2)	0.28 (0.12, 0.50)e
				Lower extremity arterial disease morbidity (ICD9 440.2)	0.32 (0.14, 0.5)i
Mayak part of combined nuclear worker study	Azizova et al (Azizova, Batistatou, et al. 2018)	0.52 (0 - 8.4)	22,734 (842,538)	Circulatory disease (ICD10 I00-I99)	0.04 (−0.00, 0.09)
				Ischemic heart disease (ICD10 I20-I25)	0.06 (0.01, 0.13)
				Cerebrovascular disease (ICD10 I60-I69)	0.00 (−0.06, 0.08)
Sellafield part of combined nuclear worker study	Azizova et al (Azizova, Batistatou, et al. 2018)	0.07 (0 - 1.88)	23,443 (602,311)	Circulatory disease (ICD10 I00-I99)	0.42 (0.12, 0.78)
				Ischemic heart disease (ICD10 I20-I25)	0.53 (0.14, 1.00)
				Cerebrovascular disease (ICD10 I60-I69)	0.05 (−0.46, 0.79)
UK NRRW heart disease	Zhang et al (Zhang et al. 2019a)	0.0232 (0 - >0.4)	174,541 (NA)	All heart disease (ICD9 393-398, 402, 404, 410-429)	0.37 (0.11, 0.65)
				Ischemic heart disease (ICD9 410-414)	0.32 (0.04, 0.61)
				Myocardial infarction (ICD9 410)	0.54 (0.16, 0.95)
				Other types of IHD (ICD9 411-414)	0.01 (−0.36, 0.45)
				Rheumatic heart disease	−0.59 (−1.89, 4.6)
				Heart failure (ICD9 428)	0.72 (−0.77, 3.21)
				Hypertensive heart disease (ICD9 402, 404)	0.06 (−1.57, 4.00)
				Other heart disease (ICD9 415-427, 429)	1.08 (0.03, 2.45)
French nuclear fuel cycle workers	Zhivin et al (Zhivin et al. 2018)	0.002 (0 - 0.072)	102 cases and 416 controls	Circulatory disease (ICD10 I00-I99)	10 (−20, 40)
			44 cases	Ischemic heart disease (ICD10 I20-I25)	0 (−50, 40)
			31 cases	Cerebrovascular disease (ICD10 I60-I69)	−10 (−50, 30)
French uranium miners case-control study	Drubay et al (Drubay et al. 2015)	0.0662 (0 - 0.4701)	76 cases, 237 counter-matched controls	All circulatory disease (ICD10 I00-I99)	0.4 (−1.6, 2.9)j
				Ischemic heart disease (ICD10 I20-I25)	−1.0 (−3.9, 3.3)j
				Cerebrovascular disease (ICD10 I60-I69)	2.4 (−0.6, 11.4)j
German uranium miner study	Kreuzer et al (Kreuzer et al. 2013)	0.047 (0.0002 – 0.909)	58,982 (2,180,639)	All circulatory disease (ICD10 I00-I99)	−0.13 (−0.38, 0.12)e
				Ischemic heart disease (ICD10 I20-I25)	−0.03 (−0.38, 0.32)e
				Cerebrovascular disease (ICD10 I60-I69)	0.44 (−0.16, 1.04)e
Eldorado uranium miners and processing (male) workers	Lane et al (Lane et al. 2010)	0.0522 (<0.0234 – >0.1215)	16,236 (508,673)	Ischemic heart disease	0.15 (−0.14, 0.58)
				Stroke	−0.29 (<−0.29, 0.27)
				All other circulatory disease	0.07 (<−0.33, 0.77)
Mound workers	Boice et al (Boice JD, Jr. et al. 2014)	0.0150 (0 - 0.9391)	7269 (293,462)	Heart disease (ICD9 390-398, 404, 410-429)	−0.974 (−3.359, 1.902)k
Chernobyl emergency workers	Ivanov et al (Ivanov et al. 2006)	0.109 (0 - >0.5)	61,017 (n.a.)	Hypertension (ICD10 I10-I15) morbidity	0.26 (−0.04, 0.56)
				Essential hypertension (ICD10 I10) morbidity	0.36 (0.005, 0.71)
				Hypertensive heart disease (ICD10 I11) morbidity	0.04 (−0.36, 0.44)
				Ischemic heart disease (ICD10 I20-I25) morbidity	0.41 (0.05, 0.78)
				Acute myocardial infarction (ICD10 I21) morbidity	0.19 (−0.99, 1.37)
				Other acute ischemic heart disease (ICD10 I24) morbidity	0.82 (−0.62, 2.26)
				Angina pectoris (ICD10 I20) morbidity	0.26 (−0.19, 0.71)
				Chronic ischemic heart disease (ICD10 I25) morbidity	0.20 (−0.23, 0.63)
				Other heart disease (ICD10 I30-I52) morbidity	−0.26 (−0.81, 0.28)
				Cerebrovascular disease (ICD10 I60-I69) morbidity	0.45 (0.11, 0.80)
				Morbidity from diseases of arteries, arterioles and capillaries (ICD10 I70-I79)	0.47 (−0.15, 1.09)
				Morbidity from diseases of veins, lymphatic vessels and lymph nodes (ICD10 I80-I89)	−0.26 (−0.70, 0.18)
				All circulatory disease (ICD10 I00-I99) morbidity	0.18 (−0.03, 0.39)
Chernobyl emergency workers	Kashcheev et al (Kashcheev et al. 2016)	0.161 (0.0001 - 1.24)	53,772 (958,540.5)	Cerebrovascular disease (ICD10 I60-I69) morbidity after no diabetes	0.35 (0.18, 0.53)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after diabetes	1.29 (0.63, 1.94)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after no atherosclerosis	0.43 (0.25, 0.62)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after atherosclerosis	0.50 (0.09, 0.90)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after no hypertensive disease	0.38 (0.08, 0.68)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after hypertensive disease	0.48 (0.27, 0.68)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after no IHD	0.41 (0.14, 0.68)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after IHD	0.47 (0.25, 0.69)
				Cerebrovascular disease (ICD10 I60-I69) morbidity after no concomitant disease	0.19 (−0.99, 1.37)
				Cerebrovascular disease (ICD10 I60-I69) morbidity	0.45 (0.28, 0.62)
Chernobyl emergency workers	Kashcheev et al (Kashcheev et al. 2017)	0.161 (0.0001 - 1.42)	53,772 (940,204.5)	Circulatory disease (ICD10 I00-I99) morbidity	0.44 (0.16, 0.74)
Seversk male cohort	Karpov et al (Karpov et al. 2012)	0.168 (0 - 1.0)	49,067 (NA)	Acute myocardial infarction (ICD10 I21)	0.480 (0.280, 0.679)l
Environmental studies
Techa River study	Krestinina et al (Krestinina et al. 2013)	0.035 (0 - 0.51)m	29,735 (901,563)	All circulatory disease mortality (ICD9 390-459)	0.18 (−0.13, 0.52)a, m
				All circulatory disease mortality (ICD9 390-459)	0.24 (−0.08, 0.59)e, m
				All circulatory disease mortality (ICD9 390-459)	0.36 (0.02, 0.75)i, m
				Ischemic heart disease mortality (ICD9 410-414)	0.26 (−0.22, 0.81)a, m
				Ischemic heart disease mortality (ICD9 410-414)	0.40 (−0.11, 0.99)e, m
				Ischemic heart disease mortality (ICD9 410-414)	0.56 (0.01, 1.19)i, m
Semipalatinsk nuclear test study	Grosche et al (Grosche et al. 2011)	0.09 (0 - 0.63)	19,545 (582,656)	Heart disease (ICD9 410-429): all settlements	3.22 (2.33, 4.10)e
				Heart disease (ICD9 410-429): exposed settlements	0.06 (−0.39, 0.52)e
				Stroke (ICD9 430-438): all settlements	2.96 (1.77, 4.14)e
				Stroke (ICD9 430-438): exposed settlements	−0.06 (−0.65, 0.54)e
				Cardiovascular disease (ICD9 390-459): all settlements	3.15 (2.48, 3.81)e
				Cardiovascular disease (ICD9 390-459): exposed settlements	0.02 (−0.32, 0.37)e
Semipalatinsk nuclear test study	Markabayeva et al (Markabayeva et al. 2018)	0.059 (0 - 1.0)	2000 (NA)	Essential hypertension	3.528 (−3.188, 10.244)n
Yangjiang high background area study	Tao et al (Tao et al. 2012)	(<0.025 - > 0.125)	31,604 (736,942)	All circulatory disease mortality (ICD9 390-460)	0.14 (−0.84, 1.29)e
				Ischemic heart disease mortality (ICD9 410-414)	0.54 (−2.65, 6.13)e
				Cerebrovascular disease mortality (ICD9 430-438)	0.44 (−0.88, 2.08)e

CI, Confidence Interval; ICD, International Classification of Diseases

^aAssuming a lag period of 5 years.

^bAnalysis based on colon dose.

^cAnalysis using underlying or contributing cause of death.

^dAnalysis based on stomach dose, derived from Table 3 of Yamada et al (Yamada et al. 2004) with smoking and drinking in the stratification.

^eAssuming a lag period of 10 years.

^fEstimate derived from log-linear model, evaluated at 1 Sv.

^g90% CI

^hRisk estimates in relation to cumulative whole body external gamma dose; doses given here are from Moseeva et al (Moseeva et al. 2014).

ⁱAssuming a lag period of 15 years.

^jAssuming a lag period of 0 years.

^kAnalysis based on linear binomial odds model fitted to data in Table 7 of Boice et al (Boice JD, Jr. et al. 2014) assuming mean doses of 2.5 mSv, 27.5 mSv and 75 mSv in <5, 5-49, 50+ mSv dose groups.

^lestimate derived by fitting a linear model by (inverse-variance) weighted least squares, applied to the adjusted SRR provided in Table 2 of Karpov et al (Karpov et al. 2012), and using the delta method to derive the variance of the ERR: see Appendix A. Average cardiac doses of 0.01, 0.04, 0.1225, and 0.25 Sv were assumed for the respective groups with the following specified ranges of effective doses: <20, 20-59, 60-185, >185 mSv.

^mAnalysis based on dose to muscle.

ⁿestimate derived by fitting a linear model by (inverse-variance) weighted least squares, applied to the adjusted OR provided in Table 2 of Markabayeva et al (Markabayeva et al. 2018). Median cardiac doses of 0.009, 0.041, 0.070, and 0.326 Sv were assumed for the respective groups with the following specified ranges of effective doses: <20, 20-59, 60-185, >185 mSv, as given by Markabayeva et al (Markabayeva et al. 2018).

Table 3.

Risks for glaucoma and macular degeneration in various radiation-exposed cohorts. IOP = intraocular pressure.

Model	Dose (Gy), mean (range)	Cases	Excess relative risk (hazard ratio) /Gy	95% CI
Glaucoma
US radiologic technologists (unadjusted for covariates)(Little M. P. et al. 2018)	0.058 (0 – 1.51)	1631	−1.39	(−1.94, −0.69)
US radiologic technologists (adjusted for covariatesa)(Little M. P. et al. 2018)			−0.57	(−1.46, 0.60)
Japanese atomic-bomb survivors (Yamada et al. 2004)	0.57 (0 – >4.14)	211	−0.18	(−0.20, −0.03)
Japanese atomic-bomb survivors (Kiuchi et al. 2013) – primary open-angle normal tension glaucoma (IOP ≤21 mmHg)	0.47 (0 – >3.01)	226	0.31	(0.11, 0.53)
Japanese atomic-bomb survivors (Kiuchi et al. 2013) – primary open-angle hypertensive glaucoma (IOP > 21 mmHg)		36	−0.21	(−0.48, 0.21)
Japanese atomic-bomb survivors (Kiuchi et al. 2013) – primary angle-closure glaucoma		25	−0.46	(−0.71, 0.02)
Mayak nuclear workers primary all primary glaucoma (Bragin et al. 2019)	0.44 (0 - >2.0)b	476	0.01b	(−0.12, 0.19)
Mayak nuclear workers primary open-angle glaucoma (Bragin et al. 2019)		461	−0.01b	(−0.14, 0.17)
Macular degeneration
US radiologic technologists (unadjusted for covariates)(Little M. P. et al. 2018)	0.058 (0 – 1.51)	1331	0.73	(−0.01, 1.74)
US radiologic technologists (adjusted for covariatesa)(Little M. P. et al. 2018)			0.32	(−0.32, 1.27)
Japanese atomic-bomb survivors: retinal degeneration (Minamoto et al. 2004)	NA (0 – >3.0)	55	0.42	(0.07, 0.88)
Japanese atomic-bomb survivors early age-related macular degeneration (Itakura et al. 2015)	0.45 (0 – > 2.0)	191	−0.07	(−0.25, 0.15)
Japanese atomic-bomb survivors late age-related macular degeneration (Itakura et al. 2015)		6	−0.21	(−0.79, 1.94)

^aadjusted by stratification for sex, race and birth year (by decade <1900, 1900-1909, …, 1950-1959, ≥1960), and with adjustment to the baseline hazard for diabetes, body mass index (BMI), smoking status (current smoker/ex-smoker/never smoker, numbers of cigarettes/day, age stopped smoking), each ascertained at the baseline survey

^busing brain dose, including unweighted neutron dose

2. Epidemiological data for circulatory disease

Circulatory disease is the dominant cause of death in the US and in most developed countries (https://www.who.int/health-topics/cardiovascular-diseases/#tab=tab_1). Appendix A Table A1 lists some of main types of circulatory disease. The major and independent environmental and lifestyle risk factors for circulatory disease include diabetes, smoking, hypertension, high total cholesterol or high LDL cholesterol, and obesity (Wilson et al. 1998; Hajifathalian et al. 2015). Many types of circulatory disease are thought to be largely inflammatory, and in particular this is the case for atherosclerosis, the process underlying ischemic heart disease (IHD) and cerebrovascular disease (CeVD), the two major groups of circulatory disease (Ross 1999; Libby 2002). At a tissue level both macrovascular and microvascular disease are important contributors to circulatory disease (Bairey Merz et al. 2006; Taqueti and Di Carli 2018). Some patients treated with radiotherapy (RT), e.g., those treated for breast cancer, Hodgkin lymphoma, brain cancer and leukemia, can get very high radiation doses (> 5 Gy) to the heart or the head. In such patients and in similarly treated experimental animals various adverse circulatory system effects are observed, including damage to the parts of the heart and large arteries (e.g., aorta, carotid). These include marked fibrotic damage, especially of the pericardium and myocardium, stenosis of the valves, microvascular damage and pericardial adhesions (Adams et al. 2003; Little 2016). The not fully understood mechanisms of such high dose damage to the circulatory system may include a number of inflammatory effects on vascular endothelial cells and capillaries (Schultz-Hector and Trott 2007; Hamada et al. 2020).

2.1. Data selection

We used PubMed to search the literature, using the search term (((radiation AND cerebrovascular) AND disease) OR (radiation AND stroke) OR ((radiation AND myocardial) AND infarction) OR ((radiation AND heart) AND disease) OR ((radiation AND circulatory) AND disease) OR ((radiation AND artery) AND disease) OR (radiation AND hypertension)). We excluded animal studies, and any study without an abstract. It was conducted on 16^th October 2019, and yielded 10,348 studies. Based on perusal of the PubMed abstracts a total of 1644 associated articles were then fully read by the principal authors (MPL, NH) to ascertain if they had high quality organ dosimetry and to determine if they were potentially informative on the desired outcome measure (excess relative risk per unit dose). In general, studies were excluded if heart or brain dose could not be reliably estimated; in occupational studies it was assumed that whole-body dose should approximate heart (or brain) dose. We excluded studies in which a sizable proportion of the cohort had very high doses (>5 Gy). Two studies of North American nuclear (Howe et al. 2004) and radiation workers (Sont et al. 2001; Zielinski et al. 2009) were excluded because of the very high (and uniform) risks seen for a number of cancer and non-cancer endpoints, suggestive of bias (Gilbert E.S. 2001). We also excluded the study of persons exposed as a result of the Three Mile Island accident (Talbott et al. 2003), in which the doses were extremely low, so that the resulting risk estimates are very likely to be biased (Gilbert E. S. et al. 2020), as well as highly uncertain. The most up-to-date follow-up was used for each cohort. Cohorts could overlap, but each cohort must contribute an underlying study population or an extra year or more of follow-up not contained in other cohorts. Therefore, we omit from further consideration the US (Schubauer-Berigan et al. 2015) and French (Leuraud et al. 2017) nuclear worker studies, which are entirely subsumed within the International Nuclear Workers Study (INWORKS) (Gillies et al. 2017); however, the study of French nuclear fuel cycle workers (Zhivin et al. 2018), the French uranium miners study (Drubay et al. 2015) and the UK nuclear workers study (Zhang et al. 2019a) had an extra two, three and ten years of follow-up, respectively, so were judged eligible for inclusion. Although the follow-up of the UK, US and French nuclear workers in INWORKS (Gillies et al. 2017) substantially exceeds that for these three workforces in the previous International Agency for Research on Cancer (IARC) 15-country analysis (Vrijheid et al. 2007), there are many cohorts in the 15-country analysis not included in INWORKS, and for that reason we judge that both cohorts should be included; however, as noted in the Introduction, most attention will be paid in the text to the INWORKS study (Gillies et al. 2017), not least because of the possibility of bias in the IARC 15-country analysis resulting from the Canadian data (Ashmore et al. 2010).

We use as our fundamental radiation risk metric the excess relative risk (ERR) per unit of absorbed dose of radiation exposure (ERR per Gy), wherever possible. Mostly unweighted radiation dose (Gy) is employed, but a few studies (e.g., LSS) use weighted dose, for example to account for the higher biological effectiveness of neutrons compared with photons in the LSS (Takahashi et al. 2017). Mostly the ERR are taken directly from the relevant manuscripts, and are shown in Table 1.

2.2. Diagnostically exposed groups

In Canada (Zablotska et al. 2014) and in Massachusetts (Little et al. 2016), mortality due to circulatory disease has been examined in two groups that received repeated fluoroscopic doses in connection with the lung collapse treatment for tuberculosis (TB). A pooled analysis of the two cohorts has also appeared (Tran et al. 2017) and we shall use only this henceforth. Lung dose was used as a surrogate for heart dose in both cohorts, and in the Massachusetts cohort additional analyses used thyroid dose (a surrogate for dose to the carotid artery) and red bone marrow dose. In the Canadian data there was a significant increase in risk per unit dose with increasing dose fractionation for IHD (Zablotska et al. 2014), but only when using a 10-year lag. There was nothing similarly observed in the Massachusetts data (Little et al. 2016) or overall (Tran et al. 2017). There is no overall dose-response in the two datasets, but analysis of those with low-moderate doses (<0.5 Gy) suggested much steeper (and statistically significant) dose-response trends for all circulatory disease and IHD (Table 1). Lifestyle and medical information is limited in both cohorts, but more complete in the Massachusetts data, where there is data for smoking, alcohol consumption, thoracoplasty, and pneumolobectomy (Little et al. 2016).

2.3. High/moderate/low-moderate/low-dose exposed groups

2.3.1. Atomic Bomb Survivors

Stroke and heart disease mortality are significantly associated with radiation exposure in the LSS (Table 1) (Shimizu et al. 2010; Takahashi et al. 2017). Takahashi et al (Takahashi et al. 2017) document increased risk for valvular heart disease (VHD), and in particular rheumatic VHD, and hypertensive organ damage (Table 1), with no significant change in excess relative risk (ERR) over time for all three endpoints, but with the excess deaths concentrated mostly before 1968. Indeed, the ERR/Gy for rheumatic heart disease is the highest of those examined, 0.96 Gy⁻¹ (95% CI 0.28, 1.92), while there is only a slight increase for IHD, 0.03 Gy⁻¹ (95% CI −0.08, 0.15) (Takahashi et al. 2017) (Table 1). In contrast the ERR for non-rheumatic VHD significantly increases (p=0.03) over time, although excess deaths are still largely concentrated before 1968 (Takahashi et al. 2017). There are distinct temporal trends for rheumatic heart disease in the LSS compared with the Japanese population, with a particularly high peak at age 80+ in the period 1950-1965 in the LSS, which is not seen nearly so strongly in the national population (Ozasa et al. 2017). The discrepant trends with the Japanese population suggests problems of diagnosis, but also perhaps illustrate the difficulties of drilling down too far in mortality rubrics.

In the higher exposed parts of the LSS cohort survivors suffered from a number of acute effects, in particular epilation, burns and other acute injuries induced by blast, heat and radiation associated with the bombs, and it is possible that these injuries may contribute to risk of late-occurring non-cancer diseases over and above the direct effects of radiation, and may also introduce selection. A degree of selection bias is a priori quite likely, although there is little evidence for this within the LSS (Little 2002); the general consistency of risks in the Japanese and other groups (Table 1) suggest that it is unlikely to have a major impact. (see (Little, Azizova, et al. 2012) for a more formal analysis) There have been pronounced changes in morbidity and mortality from circulatory disease in the Japanese population, and in the LSS cohort, in the period since the atomic bombings, possibly due to partial westernization of diet and substantial increases in cigarette smoking prevalence (Ozasa et al. 2016). Hypertension remains the major risk factor for circulatory disease in the Japanese population, and in the LSS (Ozasa et al. 2016); although hypercholesterolemia is an important risk factor in many western populations (see Appendix A Table A1), it is relatively unimportant in the older Japanese population (Ueshima 2007). The introduction of International Classification of Disease 10th revision (ICD10) coding led to other changes in disease coding in Japan, with the result that heart failure became much less commonly diagnosed after 1995 (Ozasa et al. 2016).

2.3.2. Occupationally Exposed Groups

2.3.2.1. INWORKS study and component subcohorts

The International Nuclear Worker study (INWORKS), comprises substantially extended follow-up of three large national groups of workers (for France, UK, US) (Gillies et al. 2017); it is a successor to the IARC 15-country study (Vrijheid et al. 2007). This study has demonstrated significant risks of circulatory disease, IHD, acute myocardial infarction (AMI) and CeVD (Gillies et al. 2017). In the dose response for CeVD there is some evidence of downward curvature, but not for any other endpoint (Gillies et al. 2017). As in the LSS there are indications (albeit non-significant) of increasing trends over time for non-IHD heart diseases in the UK radiation workers, particularly 40 or more years after first exposure (Zhang et al. 2019a). Gillies et al observed significant heterogeneity in all circulatory disease ERR by facility (p=0.006) although not by country (p=0.089), the heterogeneity being due to four workplaces (Sellafield/Chapelcross, UK Atomic Energy Authority (UKAEA), Portsmouth Naval Shipyard, Idaho National Laboratory), removal of which left the aggregate risk estimate unchanged, but without such heterogeneity (p>0.50) (Gillies et al. 2017). However, there was no such heterogeneity for IHD by country (p=0.183) or facility (p=0.087), nor by country (p>0.50) or facility (p=0.116) for CeVD, which suggests that the circulatory disease heterogeneity may be driven by differences in ERR between these disease endpoints and by proportion of numbers of deaths between cohorts/facilities (Gillies et al. 2017). The British heart disease study (Zhang et al. 2019b) also found significant (p=0.012) inter-facility heterogeneity of ERR, with particularly marked effects due to Sellafield/Chapelcross and UKAEA; however the effect of leaving out any one cohort did not result in large changes in ERR/Gy, which remained in the range 0.18-0.50. Zhang et al (Zhang et al. 2019b) suggested that lifestyle variations between facility, which were only partially taken into account by the crude industrial/non-industrial classification, may be the cause of this. Unfortunately this publication (Zhang et al. 2019b) (a letter to the editor) is short of certain vital details, which limit its interpretability. [Note: there were a number of occupational studies that did not appear until after the literature search had been conducted, and therefore are technically ineligible for inclusion. These include a case-control study of ischemic heart disease mortality within the British Nuclear Fuels (BNF) fuel cycle workers (working at the Sellafield and Springfields plants), which had an excess odds ratio (EOR) of 0.50 Gy⁻¹ (95% CI −0.2, 1.1) (de Vocht et al. 2020), similar to that observed 0.70 Gy⁻¹ (90% CI 0.33, 1.11) in the older BNF worker cohort analysis of McGeoghegan et al (McGeoghegan et al. 2008), which was based on a larger group of workers (including also the BNF Capenhurst and Chapelcross workers), with slightly longer follow-up (1950-1998 vs 1946-2005). The central estimate of risk was very much higher than the overall mortality risk in the LSS (Takahashi et al. 2017) (Table 1), albeit with substantial uncertainties, and was also unusual in using a very long lag, of 15 years for the main analyses, which was also the main lag period employed in the older BNF cohort analysis of McGeoghegan et al (McGeoghegan et al. 2008). The study cohort from which the case-control study was drawn was entirely subsumed within that of the INWORKS study (Gillies et al. 2017). A study of US uranium enrichment workers yielded very uncertain estimates of mortality risk ERR in relation to external dose for IHD of 19 Gy⁻¹ (95% CI −77, 260) and for CeVD of −130 (−420, 440) (Anderson et al. 2020). Again, the study cohort was largely subsumed within that of the INWORKS study (Gillies et al. 2017). A cohort study of uranium processing workers at the Mallinckrodt plant yielded an IHD mortality ERR Gy⁻¹ of 1.3 (95% CI −0.1, 2.8) (Golden et al. 2019). A study of US military personnel exposed to eight atmospheric nuclear tests yielded an IHD mortality ERR Gy⁻¹ of −0.01 (95% CI −1.2, 1.1) (Boice JD et al. 2020). A cohort of diagnostic medical radiation workers in South Korea yielded an all circulatory morbidity ERR Gy⁻¹ of 1.4 (95% CI −5.7, 9.9), an IHD morbidity ERR Gy⁻¹ of 12.2 (95% CI −7.1, 47.3) and a CeVD morbidity ERR Gy⁻¹ of 31.0 (95% CI −7.5, 115.9) (Cha et al. 2020). A problem with this cohort in comparison with some others studied in this section is the modest size (n=11,500) and relatively short follow-up (mean 8.1 years), both of which limit its statistical power.]

2.3.2.2. Russian Mayak nuclear worker cohort

The Mayak worker cohort has been subject to various recent analyses (Azizova et al. 2014; Moseeva et al. 2014; Simonetto et al. 2014; Azizova, Grigorieva, et al. 2015; Azizova, Grigoryeva, et al. 2015; Simonetto et al. 2015; Azizova et al. 2016; Azizova, Batistatou, et al. 2018), of which the most recent studies of IHD and CeVD are used here (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015), also a study of lower extremity arterial disease (Azizova et al. 2016), and the Mayak part of a pooled analysis (with the Sellafield workers) (Azizova, Batistatou, et al. 2018) which are cited in Table 1. There are significant trends with dose for IHD and CeVD morbidity, although the mortality trends for these endpoints are much lower, and generally non-significant (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Azizova, Batistatou, et al. 2018) (Table 1), the reasons for which are discussed below. Unusually, doses to certain organs, specifically the liver and lung, are predominantly from internally deposited radionuclides, in particular, the α-particle-emitting radioisotopes of plutonium. Doses in this study are moderate (for external γ rays on average 0.5-0.6 Gy (Table 1)), in contrast to the generally low doses in other occupationally-exposed groups. However, the external whole-body dose rates are low (<5 mGy/hour) (Wakeford and Tawn 2010; United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2018), as in other occupational cohorts. [Note: although this study did not appear until after the literature search had been conducted, and therefore technically ineligible for inclusion, hypertension in the Mayak worker cohort is associated with external γ dose to the liver, with an ERR of 0.14 Gy⁻¹ (95% CI 0.09, 0.20), although there is no trend with internal α-particle dose (Azizova T et al. 2019). This risk is somewhat lower than that seen in the LSS (Takahashi et al. 2017) and in a group of Chernobyl liquidators (Ivanov et al. 2006) (Table 1).]

The heterogeneous and large internal α-particle dose from plutonium complicates interpretation of the results. For IHD and CeVD morbidity trends with external γ dose and internal (α-particle) dose to the liver are generally both significant (Azizova et al. 2014; Moseeva et al. 2014), although such trends are mostly not significant for mortality; however mortality trends in relation to internal dose for IHD are statistically significant, while incidence trends are not (Moseeva et al. 2014). Few other cohorts have individual α-particle internal doses which are large enough to merit analysis. Another complication in interpreting the results of the Mayak worker data is the contrast of the nearly null findings for IHD and CeVD mortality there with the very strong significant radiation-associated excess risk for IHD (although not for CeVD, which is non-significant) in the Sellafield workers (Azizova, Batistatou, et al. 2018) (Table 1).

Russian national mortality data is likely to be particularly unsound, with marked differences in disease coding practices over the country (Wasserman and Varnik 1998; Danilova et al. 2016), and should therefore probably not be used for epidemiological analysis, in particular for any of the Russian worker studies considered here (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Azizova et al. 2016; Kashcheev et al. 2016; Kashcheev et al. 2017). Mortality data on Mayak workers who had never left Ozyorsk city for another place of residence were derived only in Ozyorsk. However, 41% of the cohort had migrated out of the city by the end of 2005 (Azizova, Grigoryeva, et al. 2015) and information on fact of and cause of death for them were derived not in Ozyorsk, but from all regions of the former Soviet Union. The analysis of Deltour et al (Deltour et al. 2015) makes clear that mortality rates for Ozyorsk residents and Mayak workers in the years 1998-2010 were much lower than those elsewhere in Russia, suggesting the continued effects of selection in the Ozyorsk groups, and access to higher quality healthcare, although a role for misclassification of cause of death outside Ozyorsk cannot be excluded. We judge this to be a potential source of bias in the analyses of the Mayak data, in particular because of the contrast between the routine and thorough medical surveillance of workers who carried on living in Ozyorsk city and had access to a high-quality health care system compared with the migrants living in other Russian regions with varying (and generally much lower) quality health care systems. The magnitude of the bias is revealed by more detailed investigation of mortality trends by residence status. Indeed, for IHD Azizova et al (Azizova, Grigoryeva, et al. 2015) demonstrated a mortality ERR among Ozyorsk residents of 0.09 Gy⁻¹ (95% CI <0, 0.21) (or adjusted for internal dose to the liver 0.19 Gy⁻¹ (95% CI 0.07, 0.36)), which is comparable to the incidence ERR of 0.14 Gy⁻¹ (95% CI 0.09, 0.21) (or adjusted for internal liver dose 0.10 Gy⁻¹ (95% CI 0.04, 0.17)), both being somewhat higher than the same thing for migrant mortality of >0 Gy⁻¹ (95% CI −0.08, 0.10) (or adjusted for internal liver dose −0.03 Gy⁻¹ (95% CI −0.10, 0.07)). Less complete information is available for CeVD, but nevertheless Moseeva et al (Moseeva et al. 2014) showed that the CeVD mortality ERR for Ozyorsk residents was 0.061 Gy⁻¹ (95% CI −0.040, 0.162) (or adjusted for internal liver dose 0.168 Gy⁻¹ (0.025, 0.311)), while the incidence ERR was 0.511 Gy⁻¹ (95% CI 0.408, 0.614) (or adjusted for internal liver dose 0.119 Gy⁻¹ (0.051, 0.187)), so that at least when adjusted for internal liver dose the ERR were comparable.

2.3.2.3. Chernobyl worker cohorts

Chernobyl recovery workers exhibit radiation-associated excess CeVD and IHD morbidity (Kashcheev et al. 2016; Kashcheev et al. 2017)(Table 1). The very high rates of circulatory disease in this cohort is an unusual feature, including as it does 22,220 cases of IHD and 23,264 cases of CeVD in a cohort of 53,772 people (Kashcheev et al. 2016; Kashcheev et al. 2017), which may reflect the markedly elevated mortality and morbidity rates of circulatory disease in the Russian population compared with other developed countries (World Health Organization (WHO) 2015). An early study of standardized mortality ratio has been published for this cohort (Ivanov et al. 2011), but no radiation risks can be derived from this. There are substantial concerns about many design aspects of this study, including cohort selection, diagnosis confirmation and source of dose information.

2.3.2.4. Other radiation worker studies

There are a number of groups of uranium miners and processing workers (Lane et al. 2010; Kreuzer et al. 2013; Drubay et al. 2015; Zhivin et al. 2018; Anderson et al. 2020) also workers at a US plant in which workers were exposed to ²¹⁰Po, ³H and various plutonium isotopes (Boice JD, Jr. et al. 2014), in none of which were there significant correlations between dose and circulatory disease (Table 1). However, a strong and statistically significant radiation-associated excess risk of AMI morbidity was observed in a group of Russian male nuclear workers in Seversk (Karpov et al. 2012).

2.3.3. Environmentally Exposed Groups

A study of essential hypertension morbidity in a Kazakhstan group exposed to fallout from nuclear weapons tests at the Semipalatinsk site, a small subset of a previously studied population (Grosche et al. 2011), found weak indications (with large uncertainties) of excess risk (Markabayeva et al. 2018) (Table 1). The dosimetry in this cohort was, however, based on assessments of residence, diet and estimates of time spent outdoors, all via interview more than three decades after the bomb tests. This study may be therefore less informative than others considered here. The Chinese high natural background radiation study (Tao et al. 2012) yields positive but non-significant trends of mortality from the major circulatory disease subtypes (Table 1).

2.4. Risk modifying factors

As noted previously, radiation-associated ERR for circulatory disease decreases with increasing age at exposure with borderline significant decreasing trends also with attained age in the LSS; however, sex or time since exposure does not substantially modify radiation risk (Little, Azizova, et al. 2012). As above, there are increasing time trends for certain non-IHD endpoints in the LSS (Takahashi et al. 2017) and in the UK nuclear workers (Zhang et al. 2019a).

3. Epidemiological data for ocular disease

The total economic burden of vision loss and blindness was estimated at $139 billion in the US in 2013, and treatment of eye-related disorders incurred annual direct medical costs of more than $68.8 billion (Wittenborn and Rein 2013). Most of the costs are associated with cataract, macular degeneration and glaucoma (Wittenborn and Rein 2013). Macular degeneration is the primary cause of vision loss in developed countries, including the US (Eye Diseases Prevalence Research Group 2004; Gunnlaugsdottir et al. 2008; Klein et al. 2013), but worldwide glaucoma, the four main types of which comprise (a) primary open-angle glaucoma (the most common type in the US), (b) primary angle-closure glaucoma, (c) secondary glaucoma, and (d) congenital glaucoma (National Eye Institute (NEI) 2017), has more substantial impact on vision loss, second only to cataract (Kingman 2004).

Opacification is the principal pathology of the lens, and this is labeled as ‘cataract’ when it is most pronounced (van Heyningen 1975). Three main types of cataract, depending on their location in the lens, are PSC, CC and NC. NC develops first in the inner embryological and fetal lens fiber cells. CC involves the outer, more recently formed lens fiber cells. PSC develops from the dysplasia of transitional zone epithelial cells, and results in an opacity at the posterior pole (Kuszak and Brown 1994). The most significant risk factors for cataracts comprise solar ultraviolet radiation (UVR) (Söderberg et al. 2016), high body mass index (BMI), diabetes, ocular trauma, cigarette smoking, and use of corticosteroid medicine (Christen et al. 1992; Cruickshanks et al. 1992; Hodge et al. 1995; Floud et al. 2016).

Metabolic disease such as diabetes are correlated with some types of glaucoma (Zhao et al. 2015; Lee et al. 2017), whereas cigarette smoking and age are the major lifestyle and environmental risk factors linked to macular degeneration (Lim et al. 2012); there is weaker evidence that macular degeneration is also associated with obesity, certain nutritional components and other variables associated with circulatory disease (Snow and Seddon 1999; Seddon, Cote, Davis, et al. 2003; Seddon, Cote, Rosner 2003; Reynolds et al. 2010). There is weak data suggesting correlations of macular degeneration with solar exposure (with uncertainty as to the direction of the association) (Nano et al. 2013; Millen et al. 2015), but even less data suggesting a link of solar exposure with glaucoma. Glaucoma (in particular neovascular glaucoma) is associated with very high dose (>5 Gy) radiotherapeutic procedures (Krema et al. 2013; Popovic et al. 2017). The National Council on Radiation Protection and Measurements (NCRP) recommended that parts of the eye other than the eye lens need to be taken into account in the system of radiation protection (National Council on Radiation Protection and Measurements (NCRP) 2016).

3.1. Cataract

Table 2 summarizes the risks for cataract in various groups and demonstrates that risks of PSC and CC are reasonably consistent. Similar risks for PSC and CC were observed in the Mayak nuclear workers (Azizova, Hamada, et al. 2018), similar also to those in other groups (Table 2), but slightly higher risks (albeit with considerable uncertainties) were seen in the Chinese high natural background area study of Su et al (Su et al. 2021), and slightly lower risks for both endpoints in a group of Chinese industrial radiographers (Lian et al. 2015). Although there was no breakdown by cataract type in the US radiologic technologists (USRT) cohort, the risk is very similar to those in other cohorts, with an ERR Gy⁻¹ of 0.69 (95% CI 0.27, 1.16) (Little Mark P. et al. 2018) (Table 2). A notable feature of the USRT cohort is that there is a significant excess cataract risk under 100 mGy, in part as a consequence of the huge size of this dataset, comprising over 12,000 cases (Little Mark P. et al. 2018; Little, Cahoon, et al. 2020; Little, Patel, et al. 2020). Similar cataract risks were also seen after high to very high dose RT, with mean doses of 2.8 Gy (maximum 66 Gy), yielding an EOR Gy⁻¹ of 0.92 (95% CI 0.65, 1.20) (Chodick et al. 2016).

Table 2.

Risks for cataract in radiation-exposed cohorts

Cohort	Dose (Gy), mean (range)	Ascertainment	Endpoint	Cases	Excess hazard ratio Gy−1 or excess odds ratio Gy−1 (95% CI)
Swedish skin hemangioma(Hall et al. 1999)	0.400 (0 - 8.4)	LOCS I	Cortical	111a	0.50 (0.15, 0.95)
			Posterior subcapsular	17a	0.49 (0.07, 1.08)
Icelandic airline pilots (Rafnsson et al. 2005)	NA (0 - 0.048)	WHO	Nuclear	71	20 (0, 30)
			Cortical	102	<0 (<0, >0)
			Posterior subcapsular	32	<0 (<0, >0)
Japanese atomic bomb survivor AHS (Nakashima et al. 2006)	0.522 (0 - 4.94)	LOCS II	Cortical	618b	0.30 (0.10, 0.53)
			Posterior subcapsular	214b	0.44 (0.19, 0.73)
			Nuclear opacity	415b	0.07 (−0.11, 0.30)
			Nuclear colour	358b	0.01 (−0.17, 0.24)
Chernobyl recovery worker (Worgul et al. 2007)	NA (0 - >1.095)	Merriam-Focht	Non-nuclear stage 1-5	3369a,c / 274d	0.65 (0.18, 1.30)
			Posterior subcapsular stage 1	2781a,c / 252d	0.42 (0.01, 1.00)
			Nuclear	382a,c / 113d	0.07 (−0.44, 1.04)
			All cataract stage 1-5	3751a,c / 384d	0.70 (0.22, 1.38)
Finnish interventional radiologists (Mrena et al. 2011)	0.060 (0.010 - 0.304)	LOCS II	Cortical or posterior excluding nuclear	7b	4 (−20, 28)
			All opacity (excluding nuclear color)	15b	13 (−2, 28)
Japanese atomic bomb survivor AHS cataract surgery (Neriishi et al. 2012)	0.500 (0 - 5.14)	Surgical removal	All cataract removal	1028	0.32 (0.17, 0.52)d, e
Chinese industrial radiographers (Lian et al. 2015)	0.07002 (>0 - 0.23611)	LOCS III	Posterior subcapsular	42	0.14 (−0.90, 0.76)
			Cortical	54	0.16 (−0.04, 0.36)
Childhood cancer survivors (Chodick et al. 2016)	2.8 (0 - 66)	Questionnaire	Cataract history	483	0.92 (0.65, 1.20)
US Radiologic Technologists (Little Mark P. et al. 2018)	0.056 (0 - 1.514)	Questionnaire	Cataract history	12,366	0.69 (0.27, 1.16)
			Cataract surgery	5509	0.34 (−0.19, 0.97)
Mayak nuclear workers (Azizova, Hamada, et al. 2018)	0.526 (0 - >2.0)	Slit lamp exam	Cortical	3132	0.62 (0.50, 0.75)f
			Posterior subcapsular	1239	0.90 (0.67, 1.19)f
			Nuclear	2033	0.47 (0.35, 0.60)f
Mayak nuclear workers (Azizova TV et al. 2019)	0.515 (0 - >2.0)	Slit lamp exam	Cataract surgery	701	0.09 (−0.02, 0.22)f
Chinese high natural background area (Su et al. 2021)	NA (0.0221 - 0.3104)	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)

Acronyms: AHS = Adult Health Study; LOCS = Lens Opacities Classification System; WHO=World Health Organization

^asummed over cataracts in left and right eyes.

^ball cases with LOCS II grade I and above.

^cprevalent cataract.

^dincident cataract.

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

^flag period of 5 years.

Estimates for NC are generally not significantly elevated – for instance the EOR in the Chinese high natural background area study the EOR was −1.9 Gy⁻¹ (95% CI −3.6, 0.1) (Su et al. 2021). Although no dose response information is given for NC in the Chinese industrial radiographers, the hazard ratio for NC in relation to fact of exposure was 0.93 (95% CI 0.78, 1.11) (Lian et al. 2015). However, larger and more significant risks were reported in the Mayak nuclear workers, with EOR Gy⁻¹ of 0.47 (95% CI 0.35, 0.60) (Azizova, Hamada, et al. 2018) (Table 2). For surgically removed cataracts the trends are non-significant in the US radiologic technologists and in the Mayak workers, with an EOR of 0.34 Gy⁻¹ (95% CI −0.19, 0.97) (Little Mark P. et al. 2018) and 0.09 Gy⁻¹ (95% CI −0.02, 0.22) (Azizova TV et al. 2019) respectively (Table 2).

Although there were no individualized radiation doses available, two large population-based cohort studies of cataract in relation to exposure to computed tomography (CT) both suggested that increased risk of cataract was associated with increasing number of CT procedures (Yuan et al. 2013; Gaudreau et al. 2020).

3.2. Glaucoma and macular degeneration

Table 3 summarizes the evidence for glaucoma and macular degeneration following moderate and low dose exposure. As can be seen, there is very little evidence that other types of ocular disease, in particular glaucoma and macular degeneration are radiation inducible, whether in the USRT (Little M. P. et al. 2018), the Mayak nuclear workers (Bragin et al. 2019) or the Japanese atomic bomb survivors (Yamada et al. 2004; Kiuchi et al. 2013). Only in the Japanese atomic bomb survivors are there indications of excess risk, for normal tension primary open-angle glaucoma, with an EOR Gy⁻¹ of 0.31 (95% CI 0.11, 0.53) (p=0.001) although not for high pressure primary open-angle glaucoma, with an EOR Gy⁻¹ of −0.11 (95% CI −0.48, 0.21) (p=0.28), or primary angle-closure glaucoma, with an EOR Gy⁻¹ of −0.46 (95% CI −0.71, 0.02) (p=0.06) (Kiuchi et al. 2013) (Table 3).

4. Epidemiological data of other non-malignant endpoints

This has been reviewed elsewhere (United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008b). The LSS data implies that there may be excess risk for non-malignant respiratory and digestive diseases (Ozasa et al. 2012). However, similar excess risks have not been seen in other radiation-exposed groups (United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008b), suggesting that the LSS findings may be due to chance, or possibly due to confounding by some correlate of radiation dose in the two cities such as heat, blast or trauma effects.

As discussed previously (Little 2013), there is radiation-associated reduction in head size and intelligence quotient (IQ) among those exposed in utero to the atomic bombs in Hiroshima and Nagasaki (Otake and Schull 1993, 1998), and reductions in cognitive test performance were also seen in another group exposed in early infancy (Hall et al. 2004). Azizova et al (Azizova et al. 2020) observed a significant radiation-associated excess risk of Parkinson’s disease incidence in a cohort of Mayak workers, with ERR / Gy = 1.03 (95% CI 0.59, 1.67) based on n=300 cases. Wong et al (Wong et al. 1993) observed a somewhat lower and marginally significant excess risk of Parkinson’s disease incidence in AHS with an ERR at 1 Gy = 0.44 (95% CI −0.06, 1.57) based on n=50 cases. It is clear that these risks are statistically compatible. Gillies et al (Gillies et al. 2017) reported a significant radiation-associated excess mortality risk of mental disorders in INWORKS, with an ERR/Gy = 1.30 (90% CI 0.23, 2.72) based on n=705 deaths, 53% of which was accounted for by dementia. Wong et al (Wong et al. 1993) reported a lower and statistically non-significant excess incidence risk of dementia in AHS, with an ERR at 1 Gy = 0.11 (95% CI −0.18, 0.64) based on n=84 cases. Loganovsky et al (Loganovsky et al. 2020) reported a highly significant increase in neuropsychiatric disorders in a group of male Ukrainian liquidators, with ERR/Gy = 2.5 (95% CI 0.86, 7.27, p<0.001). However, this very small study (n=137) was problematic in a number of ways, including the fact that the cohort was self selected, and participants will have had some idea of their exposure status, so that selection bias is possible.

Chronic bronchitis has been assessed in the Mayak nuclear workers (Azizova et al. 2017) and largely null results obtained, with an overall ERR/Gy of 0.03 (95% CI −0.05, 0.14) although some groups of workers, for example in a plutonium plant, or among ex-smokers, exhibited larger and borderline significant excess risks. There are few other radiation-exposed cohorts in which this endpoint was studied. Chronic obstructive pulmonary disease (COPD) morbidity has been assessed in the Japanese atomic bomb survivors (Pham et al. 2013), yielding an ERR / Gy of 0.08 (95% CI −0.14, 0.37), and mortality form bronchitis, emphysema and COPD has been examined in the UK nuclear workers (Muirhead et al. 2009), yielding an ERR/ Gy of −1.041 (95% CI −1.40, −0.48), implying a significant negative trend with dose. Overall, these three studies do not suggest any strong relation of chronic bronchitis with radiation exposure.

5. Discussion and conclusions

Circulatory disease

We have reviewed moderate and low dose studies, comprising a single medical diagnostic study (Table 1), the pooled TB fluoroscopy cohorts (Tran et al. 2017), with mean dose of about 1.2 Gy, included here partly because of the emphasis on trends for doses <0.5 Gy. Most of the other studies considered here involved low-to-moderate mean cumulative radiation doses (≤0.2 Gy), with participants in the occupational studies exposed at low dose rates (Table 1). Nonetheless, in most cohorts the observed trends are driven by the small numbers of participants exposed at high cumulative doses (0.5 Gy or above) (Table 1).

The ERRs that we observe (Table 1) are generally consistent with those of a previous systematic review and meta-analysis of moderate/low dose studies (Little, Azizova, et al. 2012), also with those of a successive non-systematic review (Little 2016). Risks in studies of RT-treated groups, with relevant organ doses usually well above 0.5 Gy, which we have not considered, are also generally consistent with those in populations exposed at the much lower doses/dose rates discussed above (Mulrooney et al. 2009; Little, Azizova, et al. 2012; Little, Kleinerman, et al. 2012; Darby et al. 2013; Little et al. 2013). This suggests that the mechanisms at high doses and high dose rates and at low doses and dose rates may not be dissimilar. Mean dose to the heart may be the most relevant metric for predicting radiation-associated IHD, as suggested by the fact that IHD risks in relation to mean heart dose in high-dose/partial-body exposed groups are similar to risks in the generally uniformly whole-body-exposed groups in relation to whole-body dose discussed above (Table 1) (Little et al. 2013). Specific risk factors for circulatory disease, including diabetes, cigarette smoking, obesity, high blood pressure, increased low-density lipoprotein cholesterol, and decreased high-density lipoprotein cholesterol plasma levels, also male sex, and family history of heart disease have been identified by epidemiological research (Wilson et al. 1998; Burns 2003). A limitation of many of the studies of circulatory disease is the lack of information on these lifestyle/medical risk factors. Only the Japanese atomic bomb survivors (Shimizu et al. 2010) and Mayak workers (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Azizova et al. 2016) had information on lifestyle factors, particularly alcohol consumption, cigarette smoking and obesity. In the LSS a few other variables associated with circulatory disease (education, diabetes and household occupation) were available (Shimizu et al. 2010), and in the Mayak cohort data was also available on blood pressure (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Azizova et al. 2016). A case-control study of UK nuclear fuel cycle workers (de Vocht et al. 2020) had information on smoking, obesity, blood pressure, occupational noise and shift work. In most groups, however, there is little evidence that these lifestyle risk factors, when available, appreciably modify radiation-associated circulatory disease risk, and in particular do not confound (Yamada et al. 2004; Shimizu et al. 2010; Little, Kleinerman, et al. 2012; Darby et al. 2013; Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Cutter et al. 2015; Maraldo et al. 2015; Azizova et al. 2016; van Nimwegen F.A. et al. 2016; Tran et al. 2017; van Nimwegen F. A. et al. 2017; Jacobse et al. 2019; de Vocht et al. 2020). However, the fact that these lifestyle factors do not confound in these datasets does not rule out the possibility of them confounding the radiation dose response more generally; it is possible that in some of the occupational studies which do not have such measures they may partially confound the dose response, as suggested by the inter-facility heterogeneity in risk of heart disease in the UK study (Zhang et al. 2019b).

A previous meta-analysis, which included some but not all these studies, but also some studies given radiotherapeutic exposure, found little evidence of heterogeneity for IHD and non-ischemic heart disease (p>0.2), but highly significant heterogeneity for CeVD (p<0.0001), which was eliminated if adjustment was made for any of age at exposure, mean dose, or dose fractionation (Little 2016). The evidence for heterogeneity within INWORKS is, as we judge above (Gillies et al. 2017), not compelling. There is scanty evidence of inter-facility heterogeneity within the UK heart disease mortality data, unfortunately only presented in a letter to the editor (Zhang et al. 2019b). As is clear from Table 1 overall there is fairly clear evidence of excess risk of both CeVD and IHD, although their relative magnitudes differ, so that for example in the LSS the ERR for IHD is higher than for CeVD (Shimizu et al. 2010; Takahashi et al. 2017), whereas in INWORKS it is the other way round (Gillies et al. 2017). The low ERR for CeVD in the Sellafield workers is noteworthy (Azizova, Batistatou, et al. 2018) (Table 1), but the confidence intervals are wide, so that risks are statistically consistent with the much larger ERR in the INWORKS study (Gillies et al. 2017) (Table 1). Similar variations in magnitude of site-specific ERR exist for cancer between cohorts (United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008a), the reasons for which are likely similar, relating to the differing baseline levels of the various types of disease in these groups. However, as above the role of confounding as an explanation of part of these differences cannot be discounted.

Although morbidity rather than mortality data were preferentially used here, due to the likely greater diagnostic accuracy of the former, it is nevertheless possible that disease ascertainment might vary with dose within a cohort, as for instance might be the case if the investigating medical professional recognized the radiation history of the subject resulting in the higher dose subjects getting special attention, so that a case might be made for preferring mortality data. In the Mayak worker data (Moseeva et al. 2014; Azizova, Grigoryeva, et al. 2015; Azizova et al. 2016) this is unlikely. A significant strength of this cohort is the fact “that all workers, regardless of occupation, site, and radiation dose, were subjected to mandatory annual medical health examinations which included checks performed by medical specialists, imaging, and laboratory examinations” (Azizova T et al. 2019). The health examinations also included blood pressure, height, and weight, as well as ascertainment of smoking and alcohol consumption, and these data are available for the majority of the study cohort workers. Azizova et al. state that “neither doctors who performed medical health examinations nor workers knew anything about radiation doses accumulated by each … worker”, rendering unlikely dose-dependent selection of individuals for health examinations (Azizova T et al. 2019). However, this is more likely to be an issue with the Chernobyl recovery workers (Ivanov et al. 2006; Kashcheev et al. 2016; Kashcheev et al. 2017) analyzed here. The unreliability of Russian national mortality registration noted above suggests that less weight should be attached to mortality analysis in any of the Russian cohorts considered here.

There have been various recent reviews of candidate biological mechanisms (Schultz-Hector and Trott 2007; Little et al. 2008; McMillan et al. 2010; Bernstein et al. 2020; Tapio et al. 2021). After very high (>5 Gy) dose acute exposures a number of tissue reaction (formerly deterministic) effects are observed. The blood circulatory system is plausibly affected by a number of inflammatory mechanisms at high doses, but these are still not completely understood, (Schultz-Hector and Trott 2007). Among such effects are damage to parts of the heart – including pronounced fibrotic damage, especially of the pericardium and myocardium, microvascular damage, pericardial adhesions, and stenosis of the valves – and to the coronary and carotid arteries; these occur both in patients receiving RT and in experimental animals receiving comparable partial-body doses (Adams et al. 2003; United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008b). Except for pericarditis that occurs on timescales of months, most of these endpoints occur ≥10 years after irradiation (Adams et al. 2003).

Many inflammatory markers are up-regulated long after exposure to moderate to high doses of radiation (0.5–5 Gy), in humans and in experimental studies. However, after low-moderate exposures of below ~0.5 Gy, the balance shifts toward anti-inflammatory effects (Little et al. 2008; Mitchel et al. 2011; Baselet et al. 2019). Intriguingly, there is a steeper dose-response slope for various types of circulatory disease under 0.5 Gy in two cohorts that received fractionated fluoroscopic X-ray exposures (Tran et al. 2017). This may reflect particular medical issues associated with the high dose group, who had longer duration of treatment, and whose underlying TB is likely to have been more serious. Recent analysis of the LSS mortality data taking account of dose measurement errors found that radiation-exposure induced death risks per unit dose at high doses (with a test dose of 1 Gy) were 1.15-1.28 x 10⁻² Gy⁻¹, about half those at low doses (with a test dose of 0.01 Gy), 1.76-2.84 x 10⁻² Gy⁻¹ (Little, Pawel, et al. 2020).

Correlations of T-cell and B-cell population numbers with dose in the Japanese atomic bomb survivors suggest that the immune system may be affected by acutely delivered radiation (Kusunoki et al. 1998). However, there is only weak evidence for involvement of the immune system in cardiovascular disease (Ridker 1998; Whincup et al. 2000; Danesh et al. 2002; Cannon et al. 2005; Grayston et al. 2005), and so it is unlikely that whole-body dose or bone-marrow dose could be the most relevant to radiation effects. Another proposed mechanism is monocyte cell killing in the arterial intima, based on predictions of a quasi-biological mathematical model (Little et al. 2009); this mechanism remains speculative, but there is suggestive evidence for radiation-induced endothelial cell senescence and associated monocyte adhesion (Lowe and Raj 2014; Rombouts et al. 2014), and senescence is known to be associated with atherosclerotic lesions (Vasile et al. 2001; Minamino et al. 2002). Because vascular endothelial cells (VECs) regulate vascular function and structure, dysfunctions in VECs are thought to be a critical initiating stage in many types of circulatory disease (Creager et al. 2003), and will likely play a role in both macrovascular (e.g., aorta, carotid) and microvascular disease (Bairey Merz et al. 2006; Taqueti and Di Carli 2018). It has been hypothesized that radiation may lead to changes in oxygen metabolism, resulting in long-lasting elevations in reactive oxygen species (ROS) and other parts of an inflammatory cascade (Spitz et al. 2004); there is evidence for long-lasting elevations in vascular ROS in the LSS (Nakachi et al. 2008). Mitochondrial dysfunction is associated with circulatory disease (Ballinger 2005). Mitochondria play a major role in oxygen metabolism, and murine data has shown that localized heart doses of 2 Gy lead (over a period of 4-8 weeks) to proteomic changes in mitochondria, although 0.2 Gy has no such effect (Barjaktarovic et al. 2011; Barjaktarovic et al. 2013).

There is weak evidence in the LSS that the kidney may be a target tissue for hypertension (Adams et al. 2012; Sera et al. 2013), supported by experimental animal data (Lenarczyk et al. 2013; Lenarczyk et al. 2020). The consistency of IHD risk in the peptic ulcer cohort in relation to kidney dose, 0.033 Gy⁻¹ (95% CI 0.012, 0.056) (Little, Kleinerman, et al. 2012) (Table 1), with that in the LSS, 0.02 Gy⁻¹ (95% CI −0.10, 0.15) (Shimizu et al. 2010) (Table 1) also suggests that the kidney may be a target tissue.

Diabetes and obesity are major risk factors for circulatory disease (Wilson et al. 1998; Hajifathalian et al. 2015), the former implying that the pancreas may be a relevant target tissue. Many of the metabolic alterations that occur in diabetes, including hyperglycemia, excess free fatty acid liberation, and insulin resistance, mediate abnormalities in endothelial cell function (Creager et al. 2003). There is other evidence suggesting a role for RT, and specifically dose to the pancreas, in causing diabetes, in a number of RT-exposed groups (Kleinerman et al. 2010; de Vathaire et al. 2012; van Nimwegen F. A. et al. 2014), also at lower doses in the AHS (Hayashi et al. 2003). There are trends of parathyroid hormone (PTH) with dose in the Japanese atomic bomb survivors, suggesting that there may be radiation-associated hyperparathyroidism (Fujiwara et al. 1994). Primary hyperparathyroidism results in elevations in PTH, and increasing levels of calcium in the bloodstream (Andersson et al. 2004). This elevation results in hypertension (via disturbances in the renin-angiotensin-aldosterone system), cardiac hypertrophy, and myocardial dysfunction (Andersson et al. 2004). PTH receptors are present in the myocardium and exert hypertrophic effects on cardiomyocytes (Andersson et al. 2004). These associations suggest plausible mechanisms whereby the elevated PTH concentrations associated with hyperparathyroidism may lead to certain types of circulatory disease. However, the low prevalence of hyperparathyroidism (7/1459) in the AHS (Fujiwara et al. 1994) suggests it cannot account for more than a small fraction of the LSS circulatory disease cases.

ICRP has classified circulatory disease as a tissue reaction effect (International Commission on Radiological Protection 2012), with a threshold level of 0.5 Gy below which minimal excess risk would be expected. Schöllnberger et al (Schöllnberger et al. 2012) reanalyzed the older LSS mortality data (Preston et al. 2003) which implied a threshold dose of 0.62 Gy for CeVD and 2.19 Gy for cardiovascular disease. Schöllnberger et al. (Schöllnberger et al. 2012) used multi-model inference (MMI) (Burnham and Anderson 2002; Claeskens and Hjort 2008), which is somewhat related to Bayesian model-averaging and similar Bayesian techniques (Wang et al. 2012),; arguably the method of Schöllnberger et al does not adequately assess the model parameter uncertainties, which are better addressed by fully Bayesian techniques (Wang et al. 2012).

Cataract

Our review suggests that there is radiation-associated excess risk of both PSC and CC (Table 2). NC is possibly not so strongly radiogenic, although at least two (out of five) studies that we review exhibit significant radiation-related risk (Rafnsson et al. 2005; Azizova, Hamada, et al. 2018). The ICRP has classified cataract disease as a tissue reaction effect (International Commission on Radiological Protection 2012), with a threshold dose of 0.5 Gy below which there would be minimal excess risk. This is bordering on inconsistency with the findings in a number of datasets, in particular the USRT, in which there is significant excess risk under 100 mGy, although no association was seen for cataract extractions (Little Mark P. et al. 2018; Little, Cahoon, et al. 2020).

Solar UVR exposure (Söderberg et al. 2016), diabetes, high BMI, cigarette smoking, corticosteroid medicines and ocular trauma are well-established risk factors for cataracts (Christen et al. 1992; Cruickshanks et al. 1992; Hodge et al. 1995; Floud et al. 2016). These factors have not been suggested as likely confounders of the radiation dose response in the studies considered here that collected relevant information (Worgul et al. 2007; Neriishi et al. 2012; Little Mark P. et al. 2018; Little, Cahoon, et al. 2020; Su et al. 2021). There is epidemiological (Delcourt et al. 2005; Zoric et al. 2008) and experimental (Devi et al. 1965) evidence that cataract is correlated with malnutrition. However, while not impossible we judge it unlikely that dose and malnutrition would be correlated in the LSS, implying likely minimal possibilities for confounding within the Japanese cohort. Increasing exposure age results in reduction of radiation-induced relative risk of cataract in many exposed groups (Hall et al. 1999; Neriishi et al. 2012; Little Mark P. et al. 2018).

Other ocular endpoints

As the analysis of Table 3 makes clear there is little evidence of excess risk of glaucoma and macular degeneration in any moderate or low dose study. There is some evidence of excess glaucoma risk in groups treated with radiotherapy, in particular among those being treated for uveal melanoma (Krema et al. 2013; Popovic et al. 2017). Hamada et al (Hamada et al. 2019) argue from this and other data that there may be grounds for considering glaucoma to be a non-stochastic (deterministic) effect, with a dose threshold of at least 5 Gy, but possibly much higher.

Other non-malignant endpoints

A significant excess risk for mortality from non-malignant respiratory and digestive diseases has been seen in the LSS (Ozasa et al. 2012), but not in other exposed cohorts (United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) 2008b); as above, it is possible the LSS findings are due to chance or confounding. Studies of childhood cancer survivors (principally of leukemia) have consistently demonstrated cognitive impairment associated with high dose irradiation of the head (Syndikus et al. 1994). More surprisingly, Hall et al (Hall et al. 2004) documented cognitive impairment in a Swedish cohort treated for hemangioma in infancy, with >100 mGy posterior dose associated with a ~50% reduction in high school attendance, and reductions in cognitive test performance associated with > 250 mGy exposure. In utero exposed Japanese atomic bomb survivor data also suggest cognitive impairment at high dose, but there is no cognitive impairment (e.g., reduction in IQ) below 100 mGy (Otake and Schull 1998; Schull and Otake 1999). The extent of statistical compatibility of these two datasets is uncertain given the different metrics used; the differences between the exposure-age ranges in the two studies also make useful comparisons difficult.

Biographies

Mark P. Little, D.Phil, joined the National Cancer Institute, Radiation Epidemiology Branch (REB) in 2010, and was promoted to a Senior Investigator in 2012. Previously (2000-2010), he worked in Imperial College London, and before that (1992-2000) at UK National Radiological Protection Board (now part of Public Health England). He is a member of Council of NCRP, and has served as consultant to UNSCEAR, to IAEA, to ICRP, to the UK COMARE, and to NCRP committees SC 1-21 and 1-26. He has particular statistical interests in machine learning algorithms and dose measurement error models. He has over 280 publications in the peer-reviewed literature.

Tamara V. Azizova, MD, Ph.D, is the head and the leading researcher of the clinical department, science deputy director at the Southern Urals Biophysics Institute of the Federal Medical Biological Agency, and the head of the Occupational Radiation Pathology Medical Centre of Ozyorsk, Chelyabinsk region, Russia. She serves on ICRP Committee 1, Task Groups 91 and 102, and is an Alternative Representative of a Russian delegation to UNSCEAR. She has published >200 papers in peer reviewed international journals.

Nobuyuki Hamada, RT, Ph.D, is Senior Research Scientist at CRIEPI Radiation Safety Research Center and Visiting Professor at Hiroshima University Research Institute for Radiation Biology and Medicine. He serves on ICRP Task Groups 102 and 111, NCRP PAC 1, IRPA Phase 3 Task Group on the implementation of the eye lens dose limits, and Consultation Committee on AOP development for space flight health outcomes (Canadian project). He has published >125 papers in peer reviewed international journals.