1
Radiation and effective dose: Are cancer emergence rates consistent with
organ-specific factors?
Ivan Lin Yang
2
Statement of Originality
This document is written by Ivan Lin Yang who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract: The study attempts to investigate whether effective dose and its tissue-specific weights are relevant in describing how low levels of ionizing radiation relate to the increase of cancer risk in different human tissues. The findings do not support the ICRP recommendation of using effective dose as an indicator of risk in operational radiation situations because large inaccuracies were found when trying to measure the relationship of tissue-specific weight and cancer incidence in different organs. As tissue-specific weights were expected to represent the sensitivity of different organs to radiation damage.
The study applies two econometric models to analyze how well tissue-specific weights fit the data of cancer incidence in different organs. On the first model, it was analyzed the effect of these weights on the incidence of primary cancer tumors that were not explained by external factors. The coefficient for the tissue-specific weight were of statistical significance at 0.05 and had a linear correlation of 0.59 with the dependable variable (share of different types of cancer tumors unexplained by external factors). Implying that only 35.36% of the variation in
differences in the shares of cancer types are explained by the variation in the ICRP’s tissue-specific weights. Moreover, on the second model, where the external factors were used as control variables and the dependable variable was the share of different types of cancers. It was also measured the relationship between distribution of cancer types and the weights. The tissue-specific weights were found to be statistically insignificant, and this is a surprising result as a relationship between tissue-specific weights and the share of different types of primary cancer tumors was initially expected.
Hence, the findings were partially aligned with the initial expectations, as on the first model, the cancer emergence rates in different tissues are at least weakly correlated with the tissue-specific factors. But the second model yielded a surprising result, where the coefficient for tissue-specific weights were found to be statistically insignificant. Nevertheless, the implication of both findings is that the tissue-specific weights are not precise enough to justify the use of the effective dose concept in radiation protection protocols.
4
Introduction
With the development of radiation emitting technology and with the increase of scientific knowledge on radiation and its dangers. The risks brought by ionizing radiation exposure from man-made and natural sources urged government bodies to be concerned about the amount of radiation allowed in certain environments. The International Commission on Radiological Protection (ICRP) publishes recommendations that are adopted in academic studies and in radiation protection protocols to operationalize the risks of radiation exposure for the public. Therefore, the focus of this paper is at the effective dose, the ICRP’s concept that provide a summation of radiation doses to tissues and organs for radiological protection protocols (ICRP, 2008). It is used to assess risk of detriment to human health at exposure to low levels of ionizing radiation. Moreover, the effective dose comprises of the summing of organ-specific factors multiplied by organ equivalent doses, it represents the probability of radiation-induced cancer and genetic damage at low levels of ionizing radiation. The organ specific factors are expressed as the tissue-specific weighting factors on the following equation:
‘E’ stands for effective dose which is a value calculated for the whole body. ‘Wt’ is the tissue weighting factor defined by ICRP regulation and represents the differences in organ sensitivity to radiation damage. ‘Ht’ is the equivalent dose of radiation absorbed by tissue ‘t’, it is adjusted to account for the effectiveness of the type of radiation, as different sources of radiation have different effectiveness in damaging human tissue.
The aim of this study is to verify how well the tissue-specific factors or weights (Wt)
approximate the effects of radiation on cancer emergence rates in different tissues. It will be done by comparing the data on primary cancer tumors emergence rate of different organs (controlled for external factors) with the weight’s values assigned by the ICRP.
Low levels of ionizing radiation exposure and its risks
According to the United States’ Nuclear Regulatory Commission (2017), the average American is exposed to 3.1 mSv of background radiation every year. The Nuclear Regulatory Commission (NRC) sets a dose limit of 1 mSv for the general public and 50 mSv for workers that jobs require exposure to radiation (NRC, 2017). Hence, the assumption of late effects of radiation on normal tissue and the development of tumors is not unreasonable as the exposure of radiation for the average American is above the limit set by the regulation. And it is known how radiation has adverse effects on healthy tissue and its mechanism is well documented in scientific literature (Stone et al., 2003).
Moreover, at low doses of radiation, the risk of inducing solid cancer is small but as the overall lifetime exposure increases, this risk also increases due to overtime accumulative exposure. As some scientific evidence suggest there is no threshold of exposure that is harmless, which is
5
aligned with the “linear, no threshold risk model”. This model hypothesizes that even the smallest dose of ionizing radiation still has potential to increase health risks to humans (The National Academies, 2005). However, there are other studies that challenges the validity of a linear risk model in describing the relationship between harm caused by radiation and
magnitude of radiation exposure.
Remarks on the linear, no threshold risk (LNT) model
As mentioned, some authors disagree on the assumption that a “linear, no threshold risk model” (LNT) applies to explain the risk of low levels of ionizing radiation and question if it should be used in radiation protection protocols. For example, Pennington and Siegel (2019) argue that Analyses Errors were found in the previous Life Span Studies, and the data does not support the LNT model. They go further to claim that the use of the LNT model in risk
management in low dose radiation situations is wrong and potentially dangerous, as they argue that LNT model does not accurately describe the response of increase in health risks to low dose radiation exposure. Thus, concluding that it should not be used at all.
However, Shore et al. (2018) explain that The National Council on Radiation Protection and Measurements’ (NCRP) judges continue to use the LNT on the basis that the current evidence does not suggest that any alternative dose-response relationship model would be more pragmatic or prudent for radiation protection purposes than the LNT model. Therefore, the assumption of using the LNT model to estimate the risks of low levels of ionizing radiation is not based on undisputable evidence and should be approached with caution.
Effective dose purpose and tissue-specific weighting factors
The concept of effective dose was developed with the intend of being used to set limits for radiation exposure for workers exposed to occupational radiation and for the public. It is used as a risk-related quantity for the optimization of protection below constraints and reference levels. These reference levels are defined by the ICRP as the level of dose or risk above which exposure should not occur (ICRP, 2008, pp. 95). Which considers the necessity of exposure, as some type of jobs are essential and higher radiation exposure is justifiable for these jobs. However, the maximum value for the choice of constrain and reference level is at 100 mSv due to the significant risk of cancer at doses higher than 100 mSv.
Furthermore, the effective dose concept relies on the assumption of a low dose of ionizing radiation exposure and that a linear non-threshold relationship between dose and risk is applicable. Its advantage is that it enables the summation of all radiation exposures, from external exposure to internal inhalation and ingestion of radioactive particles. Nonetheless, the efficacy of quantifying risk caused by radiation at low doses is still disputable (Harrison, 2013). Additionally, on ICRP publication 103 (2008) is stated that the “main use of effective dose are the prospective dose assessment for planning and optimization in radiological protection […]”. It should not be used for epidemiologic evaluations or to detail specific investigation of
6
Regarding the weighting factors present in the effective dose equation, they are used to
express the sensitivity of the different tissues to radiation damage, they represent mean values for humans, averaged over both sexes and all ages. Twelve tissues and organs are specified by the ICRP with individual weights and an additional ‘remainder’ tissue is defined to encompass all unaddressed tissues, the values of the weights were later revised and proposed in ICRP publication 103 (2008). The reason behind the selection of these tissues is that ICRP deemed that there was sufficient epidemiologic information on the tumorigenic effects of radiation to estimate the risk of cancer emergence. But the remainder category also includes tissues that were not explicitly evaluated as individual cancer sites (ICRP, 2008, pp. 184).
The numerical value of the weights is defined by the ICRP as: “A dimensionless factor by which the organ or tissue absorbed dose is multiplied to reflect the higher biological effectiveness of high-LET radiations compared to low-LET radiations. […]” (ICRP, 2008, pp. 30). High-LET radiation basically consists of alpha particles, protons, and neutrons. These are the relevant particles when talking about risk of cancer at low levels of ionizing radiation because they are the particles that potentially cause the most biological damage at low levels of exposure. As opposed to the biological damage caused by low-LET radiation, which is largely restricted to exposure at high dose rates (“Health Effects of Exposure to Low Levels of Ionizing Radiation”, 1990, pp. 7).
Thus, it is expected that the rates of different primary cancer types controlled for external factors are somewhat proportional to these weights. For example, the weight attributed to the stomach area is of 0.12, then close to 12% of all primary cancer tumors not explained by external factors should be occurring in this organ. Surely, there are a few other factors that will distort the actual cancer rate for different organs, and it will not be present in the analysis. But if the weights are relevant in predicting the risk of cancer emergence in different tissues under the presence of low levels of radiation. It should have some association with the shares of each cancer types assuming a uniform exposure to radiation for the average person.
Data and study design
The data used was drawn from the US National Cancer Institute and their Surveillance, Epidemiology and End Results Program (SEER). It was organized by the share of each primary cancer tumor type and matched with the ICRP table for tissue-specific weights. The original data contains the number of primary cancer tumor cases from 2013 to 2017 in US territory but in the results section the table presented was already modified to display the share of each primary tumor type among the total primary tumor cases instead of the number of cases registered for the different cancer types.
Moreover, estimates of population attributable fraction (PAF) were added to the data (see appendix) and used as a control for modifiable external factors such as smoking and second-hand smoking, alcohol intake, obesity, insufficient physical activity, bacterial infection, and dietary factors. These estimates were found in Islami et al.’s (2017) study that reported PAF
7
values for 26 different types of cancer in the US context. The PAF’s values estimated the proportion of cancers cases that could be avoided if exposure to the causal factor in the entire population was reduced to the level of reference category. For example, if alcohol intake were to be reduced to zero (zero being the reference for this category), the potential reduction of cancer cases for each organ would be equivalent to the PAF’s values related to alcohol
consumption of each primary cancer type multiplied by the respective number of cases of each cancer type.
These values for PAF were estimated from large scale pooled analyses or meta-analysis from studies in the United States, or from worldwide studies when no studies from the US were available. To calculate the overall attributable proportion due to modifiable external factors of the cancer cases for a given type of cancer, it was assumed that the risk factors had no
interaction (Islami et al., 2017). The causal factors comprised in the PAF’s values are tobacco smoking (secondhand included), excess bodyweight, alcohol intake, dietary factors, insufficient physical activity, UV radiation and cancer-associated infections.
Theoretically, some cancers could potentially be eradicated if the external factors were to be controlled to the reference categories, therefore some PAFs’ values indicate that some cancers are entirely explained by modifiable external factors. For instance, the PAFs for Kaposi Sarcoma and Cervix cancer is 100%, thus controlling all theoretical modifiable external factors could have prevented all the cases of these cancers, suggesting that these two cancers are strongly
influenced by well documented external factors.
In addition, a linear regression model is applied to observe the statistical significance of the tissue-specific weights in explaining the differences in the share of cancers unexplained by external factors. The share of cancers unexplained was found by subtracting the explained portion of the original value for the share of cancers, the explained portion was found using the PAF’s values mentioned before. Afterwards, a second linear regression model is used where the PAF’s values for each cancer type is used as a control variable, the tissue-specific weights are the independent variable, and the dependable variable are the shares of cancer types (both explained and unexplained).
The first model is described as:
Where the independent variable is the ICRP’s tissue-specific weight and the dependent variable is the share of primary cancer types that was not explained by the mentioned external factors. The second one is described as:
8
Where the independent variable is the ICRP’s tissue-specific weight, the control variable are the PAF’s values for each cancer type and the dependable variable are the share of the different cancer types without any control. This mean that the dependable variable is the original shares of primary cancer tumors, both explained and unexplained. Whereas in the first model it was only the share that was left unexplained.
The basic assumptions are that the radiation exposure is uniform for the average person and the proportion of cancer emergence in different tissues would reflect the tissue-specific weights’ values assigned by the ICRP. Since it represents the sensitivity of different tissues to ionizing radiation damage, and the sensitivity is expressed by the value of the weight assigned. The analysis also relies on the assumption that effective dose has an important role in
predicting cancer, and the cancer cases after controlling for external factors are due to exposure to low levels of ionizing radiation (doses below 100 mSv).
The shortcomings of such assumptions are that the known and unknown factors that increase risk of cancers are many and not all of them are present in the analysis. Moreover, if these extra factors were to be added to the regressions as control variables, the amount of the dependable variable explained by the tissue specific weight would likely change and more accurately measure the statistical significance of tissue-specific weights in predicting the incidence of each type of cancer. Adding more control variables could change the statistical significance of the tissue-specific weights and this could be problematic if it increased the significance of the coefficient but it would not be an issue if it decreased the significance of the tissue-specific weight coefficient. Since decreasing the statistical significance of the coefficient corroborates with the results found in the regressions in the results section but increasing the significance would contradict the findings. Additionally, other issues are the assumption of uniform radiation and the lack of precision in measuring the cancer incidence caused by low levels of ionizing radiation. If the latter were possible, then perfectly correlated weights would have been drawn in the first place.
Also, the PAF approach faces the problem of possibly overlapping the risk factors among the general population. This could lead to overestimation of the value of PAF since the factors were assumed to be independent of each other, however more modifiable external factors exist but data on the impact of these factors is scarce (Islami et al., 2017, p. 50). Therefore, the PAF’s values could either be slightly overestimating or underestimating the proportion of cancer explained by external factors, depending on the extent of the overlap and on the impact of the other modifiable factors that were not included in Islami et al.(2017) estimates. Moreover, the PAFs reported in Islami et al. (2017) are representative for the year of 2014 only but are generalized to the years 2013-2017 in the data analysis for practical purposes.
Thus, more research is necessary to validate all the assumptions of this paper and the precision of the analysis could be improved if specific PAF values for each year and all the possible modifiable factors could be included. Nonetheless, the research is still relevant because improving the measurement of cancer cases due to external factors would likely reinforce the
9
findings of the results section. As the tissue-specific weights could become less significant with more controls available, and finding new solutions to explain the relationship of low levels of ionizing radiation and cancer emergence would potentially undermine the concept of effective dose which is also one of the outcomes that could be inferred from the results section.
Results Cancer Types Share of cancers Explained by external factors Share not
explained Weights Correlation Melanoma 0.0503404 0.0478497 0.0024906 0.01 0.594634 Brain 0.0136893 0 0.0136893 0.01 Esophagus 0.0098566 0.0072029 0.0026537 0.04 Liver 0.0213813 0.0126592 0.0087221 0.04 Bladder 0.0449824 0.0211006 0.0238818 0.04 Thyroid 0.0328863 0.0041013 0.0287849 0.04 Gonads 0.0192219 0.000576 0.0186459 0.08 Stomach 0.0164454 0.0092689 0.0071765 0.12 Colon & Rectum 0.0859312 0.0469667 0.0389645 0.12 Lung & Bronchus 0.1231445 0.105519 0.0176255 0.12 Breast 0.1536151 0.0437248 0.1098903 0.12 Bone marrow related 0.0958652 0.0080134 0.0878518 0.12 Remainder 0.3326405 0.0868122 0.2458282 0.14
Total 1 0.3937947 0.6062053 1
Table 1: Data used: the share of different types of primary cancer tumors (“Share of cancers”) is the number of cases of each primary type of cancer tumor divided by the sum of cases of all primary tumor cancers. The “Explained by external factors” represent the PAF value of each cancer type multiplied by the primary tumor cases found at “Share of cancers”. The share of cancer left unexplained are designated as “Share not explained”. The ICRP’s tissue-specific weights for each organ are represented by “Weights”. And the “Correlation” value displays the linear correlation between the “Share not explained” and the “Weights”.
10
Chart 1: comparison of tissue-specific weights and cancer incidence not explained by external factors (“share not explained”).
Initially, by looking at the graph (Chart 1), it is evident that some types of cancers are almost exactly proportional to the tissue-specific weight assigned but other types are remarkably different. One possible explanation for this inconsistency is that some cancers are strongly explained by external factors apart from radiation and this was not anticipated by the organ-specific weighting factors, which indicates that these weights need adjustments.
Another interesting observation is that organs closer to the middle of the body or covered by other tissues (bone marrow covered by bone tissue) have all being less frequent to cancer incidence than predicted by the weights. One hypothesis is that the position of these organs make them more resistant to damage from low levels of ionizing radiation because high-LET particles cannot penetrate more than one or two centimeters of tissue (“Adverse Reproductive Outcomes in Families of Atomic Veterans: The Feasibility of Epidemiologic Studies.”, 1995, pp. 22). Therefore, they are less exposed to low levels of ionizing radiation since these organs are distant from the extremities.
0 0.05 0.1 0.15 0.2 0.25 0.3
Weights vs Share of cancer not explained by
external factors
11
Table 2: regression model 1: the dependable variable is the share of cancers that was left unexplained. Which is the original share of cancers (“Share of cancers” from table 1) subtracted the share of cancers that could have been avoided by controlling for the external factors
(“Explained by external factors” from table 1).
Analyzing the first linear regression model (Table 2), the results indicates that the organ-specific weights are statistically significant at 0.05 and the linear correlation between the weights and the share of cancer types not explained by external factors is only of 0.59. Revealing that only 35.36% of the variation in share of cancers unexplained by external factors can be explained by the variation in specific weights in this regression model. Which suggests that the tissue-specific weights are only loosely associated to the distribution of cancer types and as a
stochastic risk assessment they have a very weak predictive ability. Hence, from this model can be inferred that the weights are insufficient in representing the differences of tissue sensitivity to radiation damage.
Table 3: regression model 2: original distribution of cancer types (“Share of cancers” from table 1) with external factors as control (“Explained by external factors” from table 1).
On table 3, a different regression is used, in this regression the dependable variable is the total share of primary cancer types (both explained and unexplained). The tissue-specific weights are the independent variable, and a representative of the PAF values for the different cancers are
_cons -.0183521 .0308921 -0.59 0.564 -.0863452 .0496409 weights .8447829 .3443921 2.45 0.032 .0867809 1.602785 Sharenotex~d Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .055860984 12 .004655082 Root MSE = .05729 Adj R-squared = 0.2948 Residual .036109122 11 .003282647 R-squared = 0.3536 Model .019751862 1 .019751862 Prob > F = 0.0321 F(1, 11) = 6.02 Source SS df MS Number of obs = 13 . reg Sharenotexplained weights
_cons -.0197497 .0314306 -0.63 0.544 -.0897814 .050282 Explainedbyexternalfactors 1.460925 .5680801 2.57 0.028 .1951634 2.726686 weights .6814416 .403653 1.69 0.122 -.2179533 1.580836 Shareofcancers Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .096083818 12 .008006985 Root MSE = .05821
Adj R-squared = 0.5769 Residual .033878793 10 .003387879 R-squared = 0.6474 Model .062205025 2 .031102512 Prob > F = 0.0054 F(2, 10) = 9.18 Source SS df MS Number of obs = 13
12
used in the control variable. The PAF value of each cancer was multiplied by the respective cancer share of each cancer to represent its size in explaining each cancer type. To illustrate, if the PAF value of cancer ‘X’ was 90% and cancer ‘X’ share was 20% of the total cancers, then the respective data value used in the regression for the control variable was 90% times 20% (0.18). On the table 1, the values used in the control variable correspond to the values of the
“Explained by external factors” column.
The differences in the dependable variable in the first model and the second model is that the dependable variable in the first model was the shares of cancer types after subtracting the explained portion from it. Where in the second model the dependable variable is the original share of cancer types without any reduction. Since the second model had no modification to the share of cancer types, the “Explainedbyexternalfactors” variable is added to contain the control values for each cancer type. While in the first model there was no control variables because the dependable variable was already adjusted to represent only the portion unexplained of each share of cancer.
The result of the second regression shows that the tissue-specific weights have become
statistically insignificant while the variation in the independent variables explains 57.69% of the variation in the dependent variable. The result suggests that the tissue-specific weights are of no relevance in predicting the distribution of cancer types in this model. This is an unexpected outcome that further weakens the credibility of effective dose as risk assessment and predictive tool since no association between tissue-specific weights and shares of cancer types can be inferred from the second regression model, whereas only weak association can be inferred from the first model. However, it is important to notice that on both models, the amount of variation explained by the independent variables are far from 100%, which indicates that other factors are missing in the analysis. Thus, further research is necessary to strengthen the
findings.
Discussion and final remarks
The concept of effective dose was introduced with the aim to define a quantity that can be related to the probability of a detriment to the human health caused by exposure to ionizing radiation where only stochastic effects occur (Jacobi, 1975). It focuses on late effects caused by low levels of ionizing radiation, as it is known that high levels of ionizing radiation would have acute effects on human tissue (Stone et al., 2003). Additionally, it was also created to provide a value representative of the combination of organ doses exposed to radiation. Considering that before the introduction of the effective dose concept, the system of basic radiation protection standards only had maximum dose allowed for single organs and for uniform irradiation of the whole body.
However, the concept was not designed to be applied for risk estimates to individual cases but only for risk management and limitation in operational situations (Dietze, 2004). And for these reasons, Brenner (2008) argues that the concept should be replaced for a newer concept, one
13
that is able to provide estimation of risk to individuals where it considers gender and age specific factors. But others like Harrison (2015) and Dietze (2004) argue that it is still useful as an indicator of risk, and for example, adequate for controlling exposure received by people in medical diagnostic and interventional procedures. Hence, they argue in defense of the use of effective dose in operational radiation protection and in situations of practical relevance. As well as reinforcing that the effective dose concept was never meant to be used to provide estimates of risk to individual patients.
Yet, as seen in the results section, it was evident that the tissue-specific weights only had a weak association with the differences in cancer types. Its variation could only explain about 35.36% of the variation in differences in cancer emergence in different organs on the first model. While it was found to be statistically insignificant on the second regression model, demonstrating to be completely inadequate to be used as a predictive and risk assessment tool. Moreover, Pennington and Siegel (2019) had identified issues in the Data analysis of previous Life Span Studies of atomic bomb survivors. The ICRP relied on these same studies of atomic bomb survivors to draw the values for most of their tissue-specific weights. This also
compromises the accuracy of their modelling on the differences in radiation damage to different tissues since it was based on unreliable studies. Furthermore, the assumption of a linear relationship between dose and risk was already mentioned to be contentious and could undermine the entire validity of the model. As Jacobi (1975) stated this assumption as one of his main assumption when designing the effective dose concept.
Despite the ICRP recommending using the concept only to assess the risk of biological damage at low levels of radiation exposure. The tissue-specific weights failed in explaining the cancer distribution among the different tissues, suggesting that it is not suited to explain the
differences of tissue’s sensitivity to ionizing radiation damage. As the results section demonstrated, the values of the tissue-specific weights are insufficient to predict the association between the weights and the distribution of primary cancer tumor types in low radiation environments. Therefore, If the weights do not capture the difference in tissue sensitivity to radiation damage, it is safe to assume that the effective dose will not have correct measurements of risk of biological damage caused by low levels of ionizing radiation.
Thus, even putting aside all the other potential issues with the effective dose concept, with the current values of the tissue-specific weights. The concept of effective dose cannot be applied to give an estimate on the risk of cancer induced by exposure to low levels of ionizing radiation because the weights are wrong, and this would produce wrong estimates. As the results suggest, the weight’s values need to be adjusted and these adjustments are paramount if the ICRP desires to continue using the effective dose concept. Only after the adjustments of the tissue-specific weights the use of effective dose as a tool for stochastic risk assessment can become defensible.
14
Reference list
1. Brenner, D. J. (2008). Effective dose: a flawed concept that could and should be replaced. The British Journal of Radiology, 81(967), 521–523.
https://doi.org/10.1259/bjr/22942198
2. Dietze, G. (2004). Dose quantities in radiation protection and their limitations. Radiation Protection Dosimetry, 112(4), 457–463. https://doi.org/10.1093/rpd/nch097
3. Harrison, J., & Lopez, P. O. (2015). Use of effective dose in medicine. Annals of the ICRP, 44(1_suppl), 221–228. https://doi.org/10.1177/0146645315576096
4.
Howlader N, Noone AM, Krapcho M, Miller D, Brest A, Yu M, Ruhl J, Tatalovich Z, Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA (eds). SEER Cancer Statistics Review, 1975-2017, National Cancer Institute. Bethesda, MD, https://seer.cancer.gov/csr/1975_2017/, based on November 2019 SEER data submission, posted to the SEER web site, April 2020. 5. ICRP. Publication 103: Recommendations of the ICRP, Ann ICRP. 37(2–4), (2008). 6. Institute of Medicine (US) Committee to Study the Feasibility of, and Need for,Epidemiologic Studies of Adverse Reproductive Outcomes in the Families of Atomic Veterans. Adverse Reproductive Outcomes in Families of Atomic Veterans: The Feasibility of Epidemiologic Studies. Washington (DC): National Academies Press (US); 1995. 4, Basic Principles of Radiation Biology. Available from:
https://www.ncbi.nlm.nih.gov/books/NBK232435/
7. Islami, F., Goding Sauer, A., Miller, K. D., Siegel, R. L., Fedewa, S. A., Jacobs, E. J., … Jemal, A. (2017). Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA: A Cancer Journal for Clinicians, 68(1), 31–54. https://doi.org/10.3322/caac.21440
8. Jacobi, W. (1975). The concept of the effective dose a proposal for the combination of organ doses. Radiation and Environmental Biophysics, 12(2), 101–109.
https://doi.org/10.1007/bf01328971
9. NRC: Backgrounder on Biological Effects of Radiation. (2017, May 22). Retrieved from https://www.nrc.gov/reading-rm/doc-collections/fact-sheets/bio-effects-radiation.html 10. Parkin, D. M., Mesher, D., & Sasieni, P. (2011). 13. Cancers attributable to solar
(ultraviolet) radiation exposure in the UK in 2010. British Journal of Cancer, 105(S2), S66–S69. https://doi.org/10.1038/bjc.2011.486
11. Pennington, C. W., & Siegel, J. A. (2019). The Linear No-Threshold Model of Low-Dose Radiogenic Cancer: A Failed Fiction. Dose-Response, 17(1), 155932581882420.
https://doi.org/10.1177/1559325818824200s
12. National Research Council (US) Committee on the Biological Effects of Ionizing Radiation (BEIR V). (1990). Health Effects of Exposure to Low Levels of Ionizing Radiation: Beir V. National Academies Press (US).Shore, R. E., Beck, H. L., Boice, J. D., Caffrey, E. A., Davis, S., Grogan, H. A., … Dauer, L. T. (2018). Implications of recent epidemiologic studies for the linear nonthreshold model and radiation protection. Journal of Radiological
15
13. The National Academies. (2005, July 26). Low Levels of Ionizing Radiation May Cause Harm. ScienceDaily. Retrieved May 20, 2020 from
www.sciencedaily.com/releases/2005/07/050725185400.htm
14.
Stone, Helen B., et al. "Effects of radiation on normal tissue: consequences and mechanisms."The lancet oncology 4.9 (2003): 529-536.16
Appendix
The data used in the study was extracted from Howlader et al. (2019) and was combined with medians of PAF taken from Islami et al. (2017). Please note that all PAF values are in
percentages.
Female Male Total PAF F PAF M
Other cancers 194659 0 0
Oral
Cavity 22934 54410 77344 65.7 82.3
Esophagus 6512 22381 28893 67.5 74.7
Stomach 19018 29189 48207 60.6 53.6
Colon & Rectum 120631 131263 251894 50.8 58.2 Liver &
IBD 17882 44794 62676 21.9 74.1
Pancreas 43015 44,489 87504 25 26
Larynx 3,971 15,764 19735 79 84.4
Lung & Bronchus 178,136 182,843 360979 82.8 88.5
Melanoma 60,798 86,767 147565 93.7 96
Breast 446,594 3,705 450299 28.7 0
Cervix
uteri 22,834 22834 100 0
Corpus & Uterus 101630 101630 71 0
Ovary 39263 39263 4.3 0 Prostate 354648 354648 0 0 Testis 17083 17083 0 0 Bladder 31,895 99,964 131859 39.1 49.4 Kidney 39,087 68,993 108080 56.4 52.4 Brain 17,706 22,422 40,128 0 0 Thyroid 72,023 24,378 96401 12.8 11.5 Hodgkin Lymphoma 7,121 8,631 15752 1.5 8 Non-Hodgkin lymphoma 57,483 70,287 127770 2.4 14.1 Myeloma 26,222 20,612 46834 11.8 10.9 Leukemia 38,035 52,623 90658 5.71 7.392 Kaposi Sarcoma 319 2,594 2913 100 100 Mesothelioma 1,469 4,268 5737 0 0 All sites 1,471,341 1,460,004 2931345
