Table S-1. Quotations about dose response excerpted from m

1

Supplemental Digital Content for,

TABLE OF CONTENTS

Table titles:

Table S‐1. Quotations about dose response excerpted from modern studies of lung cancer in Mayak and

European workers.

Table S‐2. Percentage pooling by IREP smoking categories to obtain IREP‐based predictions of lung

cancer ERR/Gy that could be compared to epidemiologic results.

Table S‐3. Percentage change in imputed 99th‐percentile ERR/Gy study values for lung cancer due to

transfer to US populations, assuming country‐by‐country lung cancer mortality rates for 1960,

1995, and 2005.

Table S‐4. Relative risks (RR) or approximate relative risks (~RR) for workers receiving internal

plutonium doses less than 0.2 Gy.

Table S‐5. Gillies et al. cohort study, Sellafield plutonium workers only: lung cancer excess relative risks

(ERRs) at low plutonium doses relevant to Palomares veterans using fast absorption rate for

plutonium nitrate based on data from volunteers at the UK’s Public Health England.

Table S‐6. Gillies et al. cohort study, Mayak plutonium workers only: lung cancer ERRs at low plutonium

doses relevant to Palomares veterans using slow absorption rate for plutonium nitrate based on

data for Mayak workers.

Table S‐7. Relative risk data for lung cancer incidence from the 2013 Mayak worker study by Labutina et

al.

IMPLICATIONS OF RECENT EPIDEMIOLOGICAL STUDIES FOR COMPENSATION OF VETERANS EXPOSED

TO PLUTONIUM

2

Table S‐8. Lung cancer mortality rates in 1960, 1995 and 2005 per 100,000 population.

Text titles:

Text S‐1. Point values of worker RBEs published or calculated without confidence limits.

Text S‐2. Similarities between ERR/Gy for mortality and incidence found in studies of lung cancer in

plutonium workers

Text S‐3. Other risk data for plutonium doses below 0.05 Gy.

Text S‐4. Quotations from the Air Force report.

Text S‐5. Smoking complexities

Text S‐6. Uncertainties in dose.

References

Figure titles

Fig. S‐1. Excess relative risks of lung cancer incidence in Mayak and Sellafield male workers for

plutonium doses below 0.05 unweighted Gy, assuming slow, as opposed to fast, plutonium

nitrate absorption.

3

Table S‐1. Quotations about dose response excerpted from modern studies of lung cancer in Mayak and

European workers.

Study Quotation

(Gilbert et al. 2013) “Highly statistically significant dose‐response relationships were found even

when analyses were restricted to plutonium doses less than 0.2 Gy.

Estimates of the ERR per Gy were similar regardless of the dose range

restriction.”

(Labutina et al. 2013) “There was a statistically significant increase in the relative risk of lung cancer

with increasing cumulative dose to the lung from alpha radiation as low as 0.1‐

0.2 Gy. The lung cancer risks were best described by a linear dose response

relationship for internal dose to the lung. Adding a quadratic term to the model

did not significantly improve the fit (p = 0.25).”

(Gillies et al. 2017a) “While there does appear to be some nonlinearity at low dose (< 10 mGy),

the uncertainties on these estimates are large. The likely main reason for this

apparent nonlinearity is that the SWC [Sellafield worker cohort] dominates

the findings in this dose range.”

“….This pattern [for Mayak workers] could perhaps be interpreted as evidence

of a threshold effect, but it is more likely related to the higher limit of detection

in the early MWC, which results in an inability to detect variation in risk below a

certain dose level”

(Grellier et al. 2017a) “We detected no evidence of departures from linearity in the associations for

total alpha, plutonium, or uranium.”

4

Table S‐2. Percentage pooling by IREP smoking categories to obtain IREP‐based predictions of lung

cancer ERR/Gy that could be compared to epidemiologic resultsa)

IREP smoking categories

Worker cohort Never Former Current b)

Sellafield malesc) 4 96

Mayak malesd) 24 76

Sellafield males +

Mayak males Results obtained by pooling the above cohorts

Mayak femalese) 96 4

Mayak smokersf) 100

Mayak

nonsmokers

Results obtained by pooling IREP female never smokers with IREP male

former smokers

a) All age 60 exposed at age 35. To generate the results presented in Tables 5 and 6 in the main

text, the listed percentages were used to scale the number of replications for each lognormal

distribution that were added together (pooled). When data was not available for precise

assignment of smoking category percentages, a choice was made that would increase pooled

ERR/Gy results, with alternate results presented in Table footnotes.

b) 20‐39 cigarettes per day, which is an illustrative choice.

c) 4% value stated for the European cohort as never smokers (Grellier et al. 2017a) . The

European cohort consisted primarily of Sellafield workers. Assigning the residual 96% to

former smokers increases the IREP results, albeit modestly, since the slopes for former and

current smokers in IREP are not that different (Table 4 of the main text).

d) Mayak workers started smoking later than Sellafield workers (Gillies et al. 2017a) , so these

assignments assume there would be more never smokers in the Mayak workers than in

Sellafield workers. The assignment of all male Mayak nonsmokers to never smokers increases

the resulting risk values, because never smokers have the highest IREP‐derived lung cancer

ERR/Gy values.

e) 4% female smokers (Gilbert et al. 2013). The residual 96% were assumed to be never

smokers.

5

Table S‐3. Percentage change in imputed 99th‐percentile ERR/Gy study values for lung cancer due to

transfer to US populations, assuming country‐by‐country lung cancer mortality rates for 1960, 1995,

and 2005.a)

Study cohort 1960 1995 2005

Males, Gillies et al., 2017, fast Pu nitrate

solubilityb)

SWC < 0.05 Gy + 89 0 ‐ 8

SWC all doses + 85 0 ‐ 9

MWC < 0.05 Gy ‐ 20 + 22 + 18

MWC all doses ‐ 12 + 27 + 22

Females, Gillies et al., 2017

MWC all doses 0 ‐ 19 ‐ 19

Gilbert et al., 2013

MWC males ‐ 12 + 27 + 21

MWC females 0 ‐ 19 ‐19

a) Unadjusted 99th‐percentile values are listed in Tables 5 and 6 in the main text. In the simplest

transfer model used in IREP, the scaling factor to use with an ERR determined by a

multiplicative regression model would be, 0.5 x (1+B_US/B_Russia), where B_US and B_Russia

= cancer mortality rates, as listed in Table S‐8. In the second transfer model, the two terms

are allocated unevenly using a simple allocation distribution (Kocher et al. 2008). The model

result most favorable to a specific claimant is chosen to determine a compensation cutoff

dose.

b) Gillies et al. made calculations with both a fast and slow solubility rate for plutonium nitrate,

with slow based on Mayak worker numbers and fast based on measurements in volunteers

from the UK’s Public Health England (Gillies et al. 2017a). Subsequent to the publication of

the study by Gillies et al., the UK fast absorption parameters were also adopted for Mayak

workers (Vostrotin et al. 2018).

6

Table S‐4. Relative risks (RR) or approximate relative risks (~RR) for workers receiving internal

plutonium doses less than 0.2 Gy.a) 90% CI for Gillies et al. and Grellier et al.; 95% CI for the rest

(italicized).

Dose

Study Incidence RR Mortality RR

Account

for

smoking?

0

1

>0 ‐0.001 MWC Gilllies

0.88 (90% CI: 0.31, 2.15)

SWC Gillies

1.14 (90% CI: 0.90, 1.46)

>0 ‐ 0.002 MWC Gilllies 0.69 (90% CI: 0.46, 0.99)

SWC Gillies 1.27 (90% CI: 1.02, 1.56)

0.001‐0.002 MWC Gilllies 1.14 (90% CI: 0.39, 2.83)

SWC Gillies 1.15 (90% CI: 1.14, 2.08)

0.002 ‐ 0.005 MWC Gilllies

0.73 (90% CI: 0.26, 1.7)

SWC Gillies 1.25 (90% CI: 0.92, 1.66) 1.2 (90% CI: 0.98, 1.59)


0.69 (90% CI: 0.30,1.37)

SWC Gillies 1.15 (90% CI: 0.72, 1.22) 0.88 (90% CI: 0.52, 1.34)

0.01 ‐ 0.02 MWC Gilllies 0.84 (90% CI: 0.53, 1.26) 0.69 (90% CI: 0.39, 1.14)

SWC Gillies 1.23 (90% CI: 1.02, 2.73) 2.15 (90% CI: 1.39, 3.25)

0.01 ‐ 0.025

EU Grellier

(~ RR)

1.07 (90% CI: 0.67, 1.73) Y

>0 ‐ 0.049

MWC

Labutina

(female) 0.5 (95% CI: 0.1, 2.0)

Y

MWC

Labutina

(male) 0.98 (95% CI: 0.7, 1.3)

Y

0.02 ‐ 0.05 MWC Gilllies 1.03 (90% CI: 0.72, 1.42)

7

0.025‐50

EU Grellier

( ~ RR)

4.2 (90% CI: 1.56, 12) Y

>0 ‐ 0.1 MWC Gilbert

0.99 (95% CI: <1, 1.2) Y

MWC Stram

0.97 (95% CI: 0.8, 1.19) Y


1.03 (90% CI: 0.75, 1.39)

SWC Gillies

1.58 (90% CI: 0.77, 3.24)

0.05‐0.09

MWC

Labutina

(female) 1.6 (95% CI: <0, 7.9)

Y

MWC

Labutina

(male) 1.1 (95% CI: 0.7,1.7)

Y

0.1‐0.19

MWC

Labutina

(female) 6.6 (95% CI: 1.6, 26.2)

Y

MWC

Labutina

(male) 2.0 (95% CI: 1.3,3.1)

Y

0.1‐0.2 MWC Gilbert

2.4 (95% CI: 1.6,3.4) Y

MWC Stram

1.64 (95% CI: 1.16, 2.3) Y

MWC Gilllies

1.13 (95% CI: 0.71, 1.72)

SWC Gillies

1.55 (90% CI: < 1, 4.34)

a) “Approximate relative risks” = odds ratios, which in cases like lung cancer are approximately

the same as relative risks (Text S‐3). SWC = Sellafield worker cohort. SWC results are

presented for the fast Pu nitrate absorption coefficient found in volunteers at the UK’s Public

Health England. MWC = Mayak worker cohort. MWC results from Gillies et al. are presented

for the slow Pu nitrate absorption coefficient, as found in Mayak workers. Some entries

were adjusted to age 60, so as to match entries in Tables 5 and 6 in the main text.

8

Table S‐5. Gillies et al. cohort study, Sellafield plutonium workers only: lung cancer excess relative risks

(ERRs) at low plutonium doses relevant to Palomares veterans using fast rather than slow absorption

rate for plutonium nitrate based on data from volunteers from the UK’s Public Health England.a, b)

Mortality Incidence

Pu dose (Gy) ERR (90% CI) n ERR (90% CI) n

0 0.0 234 0.0 204

0 – 0.001 0.14 (‐0.10, 0.46) 61

0 – 0.002 0.27 (0.02, 0.56) 99

0.001 – 0.002 0.55 (0.14, 1.08) 35

0.002 – 0.005 0.20 (‐0.02, 0.59) 41 0.25 (‐0.08, 0.66) 43

0.005 – 0.01 ‐0.12 (‐0.48, 0.34) 13 0.15 (‐0.28, 0.22) 17

0.01 – 0.02 1.15 (0.39,2.25) 16 0.23 (0.02, 1.728) 13

a) (Gillies et al. 2017b). Data taken from their Tables S‐7 & S‐8. Data rounded to two significant

figures.

b) From (Gillies et al. 2017a): “Briefly, to model activity in organs and tissues, a modified version of

the International Commission on Radiological Protection (ICRP) Publication 66 deposition model

(15) and ICRP Publication 30 gastrointestinal tract models were used (16). The dosimetric

models used in the calculation of organ doses were also those currently recommended by the

ICRP. The estimation of the level of slow absorption ratio for plutonium nitrate (ss) was the only

parameter for which no consistent value was found between the cohorts. For this reason, the

dosimetrists provided two sets of doses based on two different values of this ratio: one obtained

on the basis of measurements of Mayak PA workers ….(hereafter referred to as ss_slow doses)

and the other using data from a study of human volunteers at PHE …(hereafter referred to as

ss_fast doses)”. PHE stands for Public Health England.

9

Table S‐6. Gillies et al. cohort study, Mayak plutonium workers only: lung cancer ERRs at low

plutonium doses relevant to Palomares veterans using slow as opposed to fast absorption rate for

plutonium nitrate based on data for Mayak workersa, b)

Mortality Incidence

Pu dose (Gy) ERR ( 90% CI) n ERR (90% CI) n

0 0 258 0 96

0 – 0.001 ‐0.12 (‐0.69, 1.15) 3

0 – 0.002 ‐0.31 (‐0.54, ‐ 0.01 33

0.001 – 0.002 0.14 (‐0.61, 1.83) 3

0.002 – 0.005 ‐0.27 (‐0.74, 0.70) 4

0.005 – 0.01 ‐0.31 (‐0.70, 0.37) 8

0.01 – 0.02 ‐0.31 (‐0.61, 0.14) 16 ‐0.16 (‐0.47, 0.26) 30

a) (Gillies et al. 2017b). Data taken from their Tables S‐7 & S‐8. Data rounded to two significant figures.

b) From (Gillies et al. 2017a): “Briefly, to model activity in organs and tissues, a modified version of the

International Commission on Radiological Protection (ICRP) Publication 66 deposition model (15) and

ICRP Publication 30 gastrointestinal tract models were used (16). The dosimetric models used in the

calculation of organ doses were also those currently recommended by the ICRP. The estimation of the

level of slow absorption ratio for plutonium nitrate (ss) was the only parameter for which no consistent

value was found between the cohorts. For this reason, the dosimetrists provided two sets of doses

based on two different values of this ratio: one obtained on the basis of measurements of Mayak PA

workers ….(hereafter referred to as ss_slow doses) and the other using data from a study of human

volunteers at PHE …(hereafter referred to as ss_fast doses)”. PHE stands for Public Health England.

10

Table S‐7. Relative risk data for lung cancer incidence from the 2013 Mayak worker study by Labutina

et al.a)

Dose to lung from Pu (Gy) Relative risk (male) Relative risk (female)

0 1 1

>0 ‐ 0.049 0.98 (95% CI: 0.7, 1.3) 0.5 (95% CI: 0.1, 2.0)

0.05‐0.09 1.1 (95% CI: 0.7,1.7) 1.6 (95% CI: <0, 7.9)

0.1‐0.19 2.0 (95% CI: 1.3,3.1) 6.6 (95% CI: 1.6, 26.2)

0.2‐0.49 2.3 (1.4; 3.7) 29.5 (9.2; 104.5)

0.55‐0.99 5.2 (3.0; 8.5) 23.2 (4.5; 102.9)

1.0‐1.99 13.3 (6.1; 22.0) 52.0 (10.2; 220.7)

2.0‐2.99 21.6 (7.5; 49.5) 136.8 (6.3; 219.0)

3.0‐3.99 ‐‐ ‐‐

4.0+ ‐‐ 108.4 (22.4; 511.9)

a) Values from Table 2 of (Labutina et al. 2013).

11

Table S‐8. Lung cancer mortality rates in 1960, 1995 and 2005 per 100,000 populationa,b)

Country

Cohort Japan US UK Russian Federation

Males 1960 10 29 65 15

Females 1960 4 4 7 4

Males 1995 32 50 50 70

Females 1995 8 26 20 6

Males 2005 30 40 32 53

Females 2005 9 26 20 6

a) Extracted from Figs. 5 & 6 in (Islami et al. 2015).

b) The results of using these rates to transfer risk from study values to a US population are shown

in Table S‐3.

12

TEXTS.

Text S‐1. Point values of worker RBEs published or calculated without confidence limits.

Although RBE values without confidence intervals cannot be compared with 99th‐percentile

predictions from IREP, they are still of interest to help delineate the range in which central values fall.

Point values for Mayak RBEs, if not given directly in a plutonium epidemiologic study, can be extracted

by dividing central values for plutonium risks by central values for low‐LET external‐radiation risks; the

ratios for lung cancer fell in the range 16 to 35 (Labutina et al. 2013, Stram et al. 2021), but there was no

obvious way to obtain confidence intervals when not available in the published papers, other than to try

simulation.

An attempt was made for this paper using simulation to estimate RBE confidence limits for

worker studies that did not report them. However, lognormal simulations assuming no correlation

between external photon and internal plutonium risks produced confidence intervals much tighter than

reported in 2 studies that did report confidence limits (results not shown), suggesting that simple

simulations were of no value. This is not surprising, because the internal plutonium dose and the

external dose are sharing the same set of cancers, with the cancer data being fit to a sum of internal and

external terms. If the plutonium share increases, the external radiation share is likely to decrease. Such

a negative correlation should increase confidence limits for the ratio, and hence for the worker RBEs.

In a 2017 study of pooled Mayak and Sellafield workers (Gillies et al. 2017a), point values for

lung cancer RBEs ranging from 10 to 25 were given for 4 modeling cases, including mortality, incidence,

and two plutonium nitrate solubility values, slow or fast. Only one had confidence limits, as discussed in

the main text. IRCP report No. 150 has made best point estimates of RBEs based on the plutonium

Mayak results, using the A‐bomb risks as the photon reference (high energy photon mix). These RBEs

13

fell in the range 14‐16, assuming a DDREF of 2 for the reference prompt gamma exposures, but

confidence limits were not given (Tirmarche et al. 2021).

How do these point values compare to the corresponding IREP prediction for the relative

effectiveness factor (REFH) that is used when A‐bomb risks are the reference source? REFH is calculated

here to be 11 (1.5, 75) by dividing REFL by the DDREF distribution and 8.2 (1.0, 62) after further adjusting

for uncertainty and bias in dosimetry in the A‐bomb epidemiologic risks. Thus, the IREP central values

are lower, but the confidence limits easily include the Report‐150 central values of 14‐16.

Text S‐2. Similarities between lung cancer ERR/Gy for mortality and incidence in plutonium workers

Published results show similarities between lung cancer ERR/Gy for mortality and incidence in

plutonium workers. For instance, Gillies et al. presented similar results for both cancer incidence and

cancer mortality in the Mayak or full cohort. The central values for incidence ERR/Gy were only 20%

higher than mortality as reported in their Tables S7 & S8 for either of the Pu Nitrate absorption rates,

slow or fast, assumed in their study (Gillies et al. 2017b). The upper confidence limits differed by 25%.

Labutina et al. found a lung cancer incidence ERR/Gy and 95% CI for males of 7.1 (4.5, 10.9) (Labutina et

al. 2013), quite similar to a finding of 7.4 (5.0‐11) in Gilbert et al. for mortality, when autopsy data were

included to generate doses for 406 workers, and a finding of 6.5 (4.4 to 9.4), when the autopsy data

were not used (Gilbert et al. 2013). And, the magnitude of the mortality findings of Gilbert et al. were

numerically similar to the incidence findings of Gillies et. al. Therefore, it makes little difference to

include or exclude the Gilbert results for mortality in Table 6 of the main text.

14

Text S‐3. Other lung cancer risk data for plutonium doses below 0.05 Gy.

There is additional epidemiologic information on lung cancer ERR/Gy values at low doses, which,

due to the absence of results for a single worker age, could not be compared quantitatively to the IREP‐

based predictions. Nevertheless, the results are informative. For instance, as stated in the main text,

the European lung cancer mortality case/control study has approximate relative‐risk values that are

consistent with the Sellafield results given in Gillies et al., which is not surprising because the Sellafield

data dominates the European plutonium cohort. Note that the European cohort study did control for

smoking, while recognizing that the smoking data was limited (Grellier et al. 2017a).

The European cohort results are labeled as approximate relative‐risk values, because they were

reported in the paper as excess odds ratios (EOR), the standard metric for case/control studies.

However, the two statistics are related. Odds are calculated by dividing number of cases by number of

controls, whereas risks are calculated by dividing by numbers of cases plus controls, which means that

the odds ratio and relative risk are approximately the same, when the number of cases are small

compared to the number of controls. Thus, odds ratios can be labeled as approximate RR, as in Table S‐

3, because lung cancer mortality is not a large fraction of total mortality.

An earlier case/control study of lung cancers following plutonium exposures at Rocky Flats

(Brown et al. 2004) presented many different types of analyses, raising a potential multiple‐comparisons

problem. Such explorations are necessary for understanding the underlying science, and preferable to

publishing only a preferred model, but the possibility of some of the results occurring by chance makes

simple comparisons with other studies difficult. One reason for the multiple analysis and tests was the

finding of inconsistently low risks at the highest exposures. The dosimetry analysis was based on an

older software program, CINDY, which used publication 30 of the International Commission for

Radiological Protection, so is now outdated. The authors suggested that, “The inverse relation between

this interval and risk suggests that the CINDY code has overestimated lung doses disproportionately for

15

subjects with high cumulative lung doses.” They went on to say, “Since the CINDY code is based on a

dosimetry model (21) that has been improved recently (31), dose‐response relations may be improved

in analyses with dose estimates made with the newer model.” Whether or not use of the latest lung

models would have changed the results is unknown. The Brown et al. study is mentioned here, because,

even though it predates the final development of IREP, it may not have been considered in IREP’s

development, because of its newness and because of somewhat unclear results.

In any case, for the regression dose range in the study by Brown et al. of most interest to groups

like the Palomares veterans, 0.1 – 0. 4 Sv, as expressed in weighted Gy, the central risk estimates

obtained in the study of Brown et al. were always elevated in multiple regression, but not always

elevated at the lower confidence limit. Controlling for smoking did not affect radiation odds ratios by

much, in agreement with the plutonium results for the Mayak cohort studies and the European

case/control study.

A 2021 study of workers at Los Alamos National Laboratory (LANL) receiving low plutonium

doses had only 19 lung cancers among the 447 workers with >5 weighted mGy lung doses from

plutonium intake, so had limited statistical ability to assess risk with precision (Boice et al. 2021). The

hazard ratio of 1.15 (95%CI 0.66. 1.99) obtained for lung cancer per (weighted) 100‐mGy of internal

plutonium exposure was compatible with plutonium risks obtained by regression to Mayak data,

according to Boice et al. Note that weighted‐Gy units rather than Sv’s were used and were obtained by

multiplying unweighted Gy by an effectiveness factor of 20. Thus, 100 mGy of weighted dose

corresponds to 5 mGy of unweighted plutonium dose.

To obtain a relative risk at a dose, d, from a hazard ratio, it needs to be multiplied by, d, and

exponentiated. The hazard ratio corresponding to 1 unweighted Gy and its confidence limits would be

enormously high, due to exponentiation, making the confidence interval so large as to be uninformative.

Thus, the value of this data will likely come from future contribution to meta‐analysis or future pooling.

16

Moreover, the LANL study did not provide risk data by plutonium dose categories after

controlling for external radiation, so study results presented by the authors in their Table 8 cannot easily

be compared with the analysis presented here in Fig. S‐1, and in the main text in Fig. 1, Table S‐4 and S‐

5.

Text S‐4. Quotations from the Air Force report.

The following two quotes are taken from Air Force report, the first from the executive summary

and the second from the report’s main text (Rademacher 2020).

“The study of former Soviet Union plutonium workers, which received some of the highest

internal exposures of internally‐deposited plutonium, found a threshold‐like dose‐response

characteristic for risk of lung cancers when corrected for smoking, a liver cancer dose‐response

curve with better fits to quadratic and linear‐quadratic models, and for bone cancer a threshold‐

like dose‐response curve. The results from these studies provide further conservative influence

on compensation decisions.”

“a) Lung cancer. Lung cancer was found to have a statistically significant increase in the relative

risk of lung cancer with increasing cumulative dose to the lung from ‐radiation for doses in the

range of 0.1 – 0.2 Gy for MPA workers when corrected for smoking (Labutina et al. 2013). At this

range and above, the relationship was linear. In terms of effective dose, as used by the ICRP for ‐

particle emitters, this range would be 200 – 400 rem. Similar finding were previously

documented by Tokarskaya et al. (1997), where the authors noted, “the dose‐response relations

for plutonium incorporation is more intricate: it has a non‐linear threshold character.”

Additional work was conducted by Tokarskaya et al. (2002) on the MPA cohort, where the

authors concluded that “lowdose risk estimates for radiation‐induced lung cancer derived

without adjusting for the influence of cigarette smoking could be greatly overestimated.” A

17

more recent pooled analysis of MPA and British plutonium workers found a linear correlation

between lung cancer and plutonium burdens (Gillies et al. 2017). The study, however, did not

correct for smoking habit.” (Rademacher 2020)

Text S‐5. Smoking complexities

The most recent Mayak study went further in its smoking analysis than earlier studies to include

estimates of pack‐years in its regressions (Stram et al. 2021). No result showed any major change in

plutonium‐associated ERR/Gy, for lung cancer with or without controlling for smoking, although the

study by Stram et al. did find a change for the ERR/Gy for external gamma rays. These results suggest

that the absence of control for smoking in the pooled study by Gillies et al. may not have affected their

ERR results for plutonium by much, but conceivably might have affected their value for RBEs.

Gillies et al. did not control for smoking because of a conclusion that the smoking data for

Sellafield was inadequate. The authors of the European case/control study, however, decided to make

use of the limited Sellafield smoking data, finding large excess odds ratios per Gy (EOR/Gy) that varied

little with smoking adjustment as shown in eTable 4 of Grellier et al. (Grellier et al. 2017b). The authors

conceded, however, that the degree to which confounding by smoking was controlled “… was

potentially limited by a low proportion of lifetime nonsmokers and relatively high proportion of subjects

for which definitive smoking data were not available.” Based on the lack of smoking sensitivity in the

European cohort study, which was primarily Sellafield workers, the same insensitivity would presumably

have been found, had the limited Sellafield smoking data been used in the pooled Mayak and Sellafield

study by Gillies et al.

Why might lung cancer ERR/Gy for plutonium be insensitive to controlling for smoking when

considering the whole population? At Mayak, the overwhelming number smoked, in fact, 88%

according to one study (Labutina et al. 2013). At Sellafield, although percentage of current smokers was

18

lower, former smoker were very common. Thus, it is possible that the insensitivity of plutonium risk

estimates to controlling for smoking was due to the scarcity of never smokers. Alternatively, the dose

response may be highly multiplicative, with response proportional to baseline cancer rates, although

there were indications of behavior intermediate between additive and multiplicative (Gilbert et al.

2013).

Working against the idea of multiplicative model dominance is the finding of Stram et al., with

their more sophisticated smoking analysis, that although the lung cancer ERR/Gy results for plutonium

risks did not depend on smoking intensity, the ERR/Gy results for external radiation were smoking

dependent (Stram et al. 2021). The results for external radiation were also complex in the study of

Rocky Flats plutonium workers (Brown et al. 2004). Separating out external radiation effects from

external plus plutonium effects, which depends on the analytical approach (Gilbert et al. 2013), appears

to be more sensitive to study limitations than extracting plutonium effects alone.

Text S‐6. Uncertainties in dose.

IREP takes into account uncertainties in the Japan A‐bomb dosimetry, but the risk transfers

made here from the UK and the Russian Federation do not account for possibly larger uncertainties in

Mayak and Sellafield dosimetry. Stram et al. have considered the impact of dose uncertainty for a

multiplicative model applied to the full Mayak sample without restriction to age 60, finding a reduction

in central value for plutonium‐associated ERR/Gy for lung cancer of about a factor of two, but an

unchanged upper 95%‐confidence limit (Stram et al. 2021). Although the upper 95% confidence limit

did not change in the analysis by Stram et al., the 99th percentile would be expected to change

somewhat, because the standard deviation has increased. To get an idea of how big this effect might

be, the Stram et al. finding was assumed to apply to Mayak workers at age 60. Specifically, the central

value in the normal simulations of published age‐60 lung cancer results of Gilbert et al., 2013, was

19

reduced by half and the 95th percentile kept the same. Based on a normal simulation, the 99th

percentile increased by about 8%, which would not be sufficient to explain any difference in Tables 5

and 6 in the main text between epidemiologic findings and IREP‐based predictions.

REFERENCES

Boice JD, Cohen SS, Mumma MT, Golden AP, Howard SC, Girardi DJ, Ellis ED, Bellamy MB, Dauer LT,

Samuels C, Eckerman KF, Leggett RW. Mortality among workers at the Los Alamos National

Laboratory, 1943–2017. International Journal of Radiation Biology: 1‐28; 2021.

Brown SC, Schonbeck MF, McClure D, Baron AE, Navidi WC, Byers T, Ruttenber AJ. Lung cancer and

internal lung doses among plutonium workers at the Rocky Flats Plant: a case‐control study. Am

J Epidemiol 160: 163‐72; 2004.

Gilbert ES, Sokolnikov ME, Preston DL, Schonfeld SJ, Schadilov AE, Vasilenko EK, Koshurnikova NA. Lung

cancer risks from plutonium: an updated analysis of data from the Mayak worker cohort.

Radiation Research 179: 332‐342; 2013.

Gillies M, Kuznetsova I, Sokolnikov M, Haylock R, O'Hagan J, Tsareva Y, Labutina E. Lung cancer risk from

plutonium: a pooled analysis of the Mayak and Sellafield worker cohorts. Radiation Research

188: 645‐660; 2017a.

Gillies M, Kuznetsova I, Sokolnikov M, Haylock R, O'Hagan J, Tsareva Y, Labutina E. Supplement: Lung

cancer risk from plutonium: a pooled analysis of the Mayak and Sellafield worker cohorts.

Radiation Research 188: 645‐660; 2017b.

Grellier J, Atkinson W, Bérard P, Bingham D, Birchall A, Blanchardon E, Bull R, Guseva Canu I, Challeton‐

de Vathaire C, Cockerill R, Do MT, Engels H, Figuerola J, Foster A, Holmstock L, Hurtgen C, Laurier

D, Puncher M, Riddell AE, Samson E, Thierry‐Chef I, Tirmarche M, Vrijheid M, Cardis E. Risk of

lung cancer mortality in nuclear workers from internal exposure to alpha particle‐emitting

radionuclides. Epidemiology (Cambridge, Mass) 28: 675‐684; 2017a.

Grellier J, Atkinson W, Bérard P, Bingham D, Birchall A, Blanchardon E, Bull R, Guseva Canu I, Challeton‐

de Vathaire C, Cockerill R, Do MT, Engels H, Figuerola J, Foster A, Holmstock L, Hurtgen C, Laurier

D, Puncher M, Riddell AE, Samson E, Thierry‐Chef I, Tirmarche M, Vrijheid M, Cardis E.

20

Supplement: Risk of lung cancer mortality in nuclear workers from internal exposure to alpha

particle‐emitting radionuclides. Epidemiology (Cambridge, Mass) 28: 675‐684; 2017b.

Islami F, Torre LA, Jemal A. Global trends of lung cancer mortality and smoking prevalence. Transl Lung

Cancer Res 4: 327‐338; 2015.

Kocher DC, Apostoaei AI, Henshaw RW, Hoffman FO, Schubauer‐Berigan MK, Stancescu DO, Thomas BA,

Trabalka JR, Gilbert ES, Land CE. Interactive Radioepidemiological Program (IREP): A web‐based

tool for estimating probability of causation/assigned share of radiogenic cancers. Health Physics

95: 119‐147; 2008.

Labutina EV, Kuznetsova IS, Hunter N, Harrison J, Koshurnikova NA. Radiation risk of malignant

neoplasms in organs of main deposition for plutonium in the cohort of Mayak workers with

regard to histological types. Health Physics 105: 165‐176; 2013.

Rademacher SE. Topical issues for assessment of dose to Palomares accident recovery workers (available

at:

https://www.safety.af.mil/Portals/71/documents/Weapons/Palomares%20Dose%20Assessment

%20Rpt%207%20Jul%202020.pdf?ver=2020‐08‐05‐113936‐200 Accessed 19 August, 2021).

United States Air Force Safety Center; 2020.

Stram DO, Sokolnikov M, Napier BA, Vostrotin VV, Efimov A, Preston DL. Lung cancer in the Mayak

workers cohort: risk estimation and uncertainty analysis. Radiation Research 195: 334‐346;

2021.

Tirmarche M, Apostoaei I, Blanchardon E, Ellis ED, Gilbert E, Harrison JD, Laurier D, Marsh JW,

Sokolnikov M, Wakeford R, Zhivin S. ICRP Publication 150: Cancer risks from plutonium and

uranium exposure. Ann ICRP 50: 1‐143; 2021.

Vostrotin VV, Napier BA, Zhdanov AV, Miller SC, Sokolova AB, Bull RK, Suslova KG, Efimov AV, Smith MA,

Vvedensky VE. The Mayak worker dosimetry system (MWDS‐2016): internal dosimetry results

and comparison with MWDS‐2013. Radiation Protection Dosimetry 184: 201‐210; 2018.

FIGURE CAPTION FOR SDC

Fig. S‐1. Excess relative risks of lung cancer incidence in Mayak and Sellafield male workers for

plutonium doses below 0.05 unweighted Gy as reported in Gillies et al., 2017. 90% confidence limits on

data points as reported in Table S10 of Gillies et al. Fitted linear lines have upper and lower 90%

confidence limits around them. The linear fits were made to combined data for Mayak and Sellafield

21

workers. Risk values shown were calculated in the Gilles et al. study assuming a slow absorption

solubility ratio for plutonium nitrate. Circles = Sellafield data; Triangles = Mayak data. Solid lines = fit to

all dose data. Dashed lines = fit to dose data below 0.05 Gy. Dotted lines are illustrative IREP‐based

predictions for male Mayak and Sellafield workers combined at age 60 exposed at age 35 (Table 5 in

main text).

Slow Pu Nitrate

−0.5

0.0

0.5

1.0

1.5

0.00 0.01 0.02 0.03 0.04 0.05Dose (Gy)

ER

R

Table S-1. Quotations about dose response excerpted from m

Documents