NOTE: End date changed to 2/25/2021 per NSSC information (Ed., 9/3/20)

This task consists of four major components: The first component involves developing mechanistically-motivated mathematical models for heavy ion-induced carcinogenesis. It includes both targeted effects (TE), caused by DNA damage resulting from traversal of cells by ionizing tracks, and non-targeted effects (NTE), caused by radiation-induced perturbation of molecular signaling pathways between traversed and non-traversed cells. The second component involves estimating site-specific and consensus dose response functions for heavy ions produced by model-based analysis of NSCOR experimental data. The third component involves generating realistic uncertainty estimates for the functions from component two. Finally, in the fourth component, we will compare our results and uncertainties with current risk estimates and uncertainties from NASA.

To estimate heavy ion-induced cancer risks in astronauts engaged in long-distance space exploration such as a flight to Mars, we developed and are refining a mechanistically-motivated mathematical model of space radiation induced carcinogenesis. Our model (Shuryak et al., 2017) combines TE and NTE components. The TE component over the dose range of interest for space missions is reasonably described by a linear dependence. In contrast, the NTE component for heavy ions tends to be non-linear with a concave shape.

The recently updated mouse tumorigenesis data from our collaborators at Georgetown University show that not only overdispersion relative to the Poisson distribution (where variance/mean > 1), but also underdispersion (variance/mean < 1) are encountered, depending on radiation type and dose. Consequently, we generated a new detailed error distribution approach for the variability of tumor count data based on the weighted negative binomial (WNB) distribution. The motivation for using this more complex model is to reduce the errors on model-based radiation quality assessments and risk estimates by improved handling of the data variances.

Reference: Shuryak, I., Fornace, A.J., Datta, K., Suman, S., Kumar, S., Sachs, R.K., Brenner, D.J., 2017. Scaling Human Cancer Risks from Low LET to High LET when Dose-Effect Relationships are Complex. Radiat. Res. 187, 476–482.

As an example of potential utility of this approach, we applied the best-fit functions for TE and NTE to calculated microdosimetric lineal energy deposition distributions for the space environment and for the surface of Mars. The results predicted TE and the NTE low-dose metrics for the space environment and for the Mars surface. The metrics were somewhat higher for the space environment than for the Mars surface because the space environment spectrum contained a larger contribution of high lineal energies (Northum et al., 2015). Of interest is that the low-dose metrics for NTE were 2-3 fold higher than those for TE.

[Ed. Note. Reference: Northum, J.D., Guetersloh, S.B., Braby, L.A., Ford, J.R., 2015. Simulated response of a tissue-equivalent proportional counter on the surface of Mars. Health Phys. 109, 284–295. doi:10.1097/HP.0000000000000335 .]

NOTE: End date changed to 2/25/2021 per NSSC information (Ed., 9/3/20)

This task consists of four major components: The first component involves developing mechanistically-motivated mathematical models for heavy ion-induced carcinogenesis. It includes both targeted effects (TE), caused by DNA damage resulting from traversal of cells by ionizing tracks, and non-targeted effects (NTE), caused by radiation-induced perturbation of molecular signaling pathways between traversed and non-traversed cells. The second component involves estimating site-specific and consensus dose response functions for heavy ions produced by model-based analysis of NSCOR experimental data. The third component involves generating realistic uncertainty estimates for the functions from component two. Finally, in the fourth component, we will compare our results and uncertainties with current risk estimates and uncertainties from NASA.

To estimate heavy ion-induced cancer risks in astronauts engaged in long-distance space exploration such as a flight to Mars, we developed and are refining a mechanistically-motivated mathematical model of space radiation induced carcinogenesis. Our model (Shuryak et al., 2017) combines TE and NTE components. The TE component over the dose range of interest for space missions is reasonably described by a linear dependence. In contrast, the NTE component for heavy ions tends to be non-linear with a concave shape.

The recently updated mouse tumorigenesis data from our collaborators at Georgetown University show that not only overdispersion relative to the Poisson distribution (where variance/mean > 1), but also underdispersion (variance/mean < 1) are encountered, depending on radiation type and dose. Consequently, we generated a new detailed error distribution approach for the variability of tumor count data based on the weighted negative binomial (WNB) distribution. The motivation for using this more complex model is to reduce the errors on model-based radiation quality assessments and risk estimates by improved handling of the data variances.

Reference: Shuryak, I., Fornace, A.J., Datta, K., Suman, S., Kumar, S., Sachs, R.K., Brenner, D.J., 2017. Scaling Human Cancer Risks from Low LET to High LET when Dose-Effect Relationships are Complex. Radiat. Res. 187, 476–482.

Here we apply our modeling approach to recently updated and expanded data obtained by our collaborators from Georgetown University at the NASA Space Radiation Laboratory (NSRL). This is a detailed data set on APC(1638N/+) mouse tumorigenesis induced by space-relevant doses of protons, 4He, 12C, 16O, 28Si, or 56Fe ions, or gamma rays. A customized WNB distribution was used to model the data variability, which exhibited either under- or over-dispersion relative to the Poisson distribution, depending on radiation dose and type. This data set and modeling approach allowed detailed quantification of dose-response shapes, NTE and TE model parameters, and radiation quality metrics (relative biological effectiveness, RBE, and radiation effects ratio, RER, relative to gamma rays) for each radiation type.

The best-fit dose response for each radiation type was a smooth function, asymptotically linear at very low doses where NTE dominate and also asymptotically linear at high doses where TE dominate, but concave at intermediate doses. Both NTE and TE parameters increased with radiation linear energy transfer (LET) from gamma rays to Si ions, with evidence for saturation/decrease at very high LET >70 keV/µm. RBE and RER were asymptotically the same at very low and very high doses, but RBE exceeded RER by up to several-fold in the intermediate space-relevant dose range.

The proposed modeling approach can enhance current knowledge about quantification of health risks from space radiation. RBE and RER can be used to scale gamma ray-induced human colon cancer risks to space radiations, generating dose-dependent risk estimates for protons and each type of heavy ions for astronauts.

Bibliography (publications reported in previous years):

Shuryak, I., Brenner, D.J., 2019. Mechanistic modeling predicts no significant dose rate effect on heavy-ion carcinogenesis at dose rates relevant for space exploration. Radiat. Prot. Dosimetry 183, 203–212.

Shuryak, I., Fornace, A.J., Datta, K., Suman, S., Kumar, S., Sachs, R.K., Brenner, D.J., 2017. Scaling Human Cancer Risks from Low LET to High LET when Dose-Effect Relationships are Complex. Radiat. Res. 187, 476–482.

In order to answer the critical question of how to reliably estimate heavy ion-induced cancer risks in astronauts embarking on long-distance space exploration missions such as a flight to Mars, we are developing a mechanistically-motivated mathematical model that can predict radiogenic carcinogenesis as function of dose and dose rate using both targeted effect (TE) and non-targeted effect (NTE) contributions. Importantly, such models are needed to predict low dose rate risks based on data at higher dose rates because the very low heavy ion dose rates relevant for space missions are difficult to achieve in terrestrial experiments. Our goal is to calibrate the carcinogenesis model using available human and animal data and to generate scaling factors such as the recently proposed radiation effects ratio (RER), which compares carcinogenic effectiveness of heavy ions and gamma rays at the dose of interest. The scaling factors would then be used to estimate human heavy ion-induced cancer risks, based on human gamma-ray-induced risks. An important focus of our work is generation of realistic uncertainties for model parameters and predictions, which ultimately translate into realistic uncertainties on astronaut risk estimates.

Because the complex space radiation mixtures (consisting of protons and various types of heavy ions, as well as photons and neutrons) are difficult to recreate experimentally on Earth, mechanistically-motivated mathematical models represent valuable tools that help to enhance the interpretation of terrestrial experiments, generate quantitative predictions of risks from space exposures, and scale risk estimates from experimental animals to humans. We developed and tested such models on several data sets [7-10]. Here we apply our modeling approach to new data obtained by our collaborators from Georgetown University at the NASA Space Radiation Laboratory (NSRL). Model-based analysis of these expanded data sets on animal carcinogenesis, which cover a broad space-relevant LET range of 2-148 keV/µm, allows a more thorough investigation of how TE and NTE mechanisms depend on LET, and thereby enhances our understanding of space radiation-induced cancer risks.

The results suggest that LET values around 100 keV/µm correspond to maximal values for both TE and NTE processes. However, the shapes of the LET dependences for TE and NTE were different, with NTE increasing more rapidly with LET over 2-22 keV/µm, compared with TE. These findings provide a rationale for further research into how NTE phenomena depend on microscopic energy deposition patterns, and why these dependences may differ from those for TE.

Our model also provides information on dose response shapes, which can prove useful for assessing additivity vs non-additivity of effects from space-relevant radiation mixtures. Additivity is expected at low doses, below the dose range where NTE saturation occurs. Doses encountered in space during a prolonged interplanetary mission are likely to overlap the range of where additivity may change to sub-additivity.

References

7. Shuryak I, Fornace AJ, Datta K et al. Scaling Human Cancer Risks from Low LET to High LET when Dose-Effect Relationships are Complex. Radiat Res 2017;187:476–82.

8. Shuryak I, Sachs RK, Brenner DJ. Biophysical Models of Radiation Bystander Effects: 1. Spatial Effects in Three-Dimensional Tissues. Radiat Res 2007;168:741–9.

9. Shuryak I. Quantitative modeling of responses to chronic ionizing radiation exposure using targeted and non-targeted effects. PLoS One 2017;12:e0176476.

10. Shuryak I, Brenner DJ. Mechanistic modeling predicts no significant dose rate effect on heavy-ion carcinogenesis at dose rates relevant for space exploration. Radiat Prot Dosimetry 2019;183:203–12.

In order to answer the critical question of how to reliably estimate heavy ion-induced cancer risks in astronauts embarking on long-distance space exploration missions such as a flight to Mars, we are developing a mechanistically-motivated mathematical model that can predict radiogenic carcinogenesis as function of dose and dose rate using both targeted effect (TE) and non-targeted effect (NTE) contributions. Importantly, such models are needed to predict low dose rate risks based on data at higher dose rates because the very low heavy ion dose rates relevant for space missions are difficult to achieve in terrestrial experiments. Our goal is to calibrate the carcinogenesis model using available human and animal data and to generate scaling factors such as the recently proposed radiation effects ratio (RER), which compares carcinogenic effectiveness of heavy ions and gamma rays at the dose of interest. The scaling factors would then be used to estimate human heavy ion-induced cancer risks, based on human gamma-ray-induced risks. An important focus of our work is generation of realistic uncertainties for model parameters and predictions, which ultimately translate into realistic uncertainties on astronaut risk estimates.

For tractable model development, we assumed that traversal of a cell by densely ionizing radiation such as the core of a heavy ion track can cause the release of NTE signals. Under continuous irradiation, such as during a space mission, the signal concentration is expected to quickly (in much less time than the duration of exposure) reach a steady-state equilibrium value in the target organ(s), which is proportional to the radiation dose rate. This concentration determines the equilibrium probability for cells susceptible to NTE signals to enter into and remain in an “activated” state, e.g., a state of perturbed signaling, altered gene expression, and/or oxidative stress. We assume, for simplicity, that two components contribute to the excess cancer yield due to NTE: (1) the effects of cell activation during irradiation, which occur over the irradiation time, and (2) the cumulative (integrated) effects of cell activation after irradiation is finished and NTE signals are decaying. We also assume that the TE contribution to the heavy ion-induced cancer risk is proportional to dose and independent of dose rate, and that cell killing effects (which are unlikely to be dramatic at doses relevant for space missions) of heavy ions are also dose rate independent.

We fitted the model to lung carcinogenesis data in radon-exposed miners and rats. These data sets are valuable because they provide information on human and animal lung carcinogenesis induced by protracted exposure to densely ionizing radiation at doses and exposure durations that overlap the range expected during space exploration missions. In addition, radiation energy deposition patterns in cell nuclei in the bronchial epithelium are similar for radon and several types of space-relevant heavy ions such as iron.

The model was able to describe the shapes of dose and dose rate dependencies observed in data sets of human and animal cancer risks after protracted densely ionizing radiation exposure. We generated model-based DRE estimates, relative to acute exposures, on heavy ion-induced carcinogenesis at doses/dose rates expected during a Mars mission. A small and not statistically-significant DRE was predicted for human data and for combined human and rat data.

Our results suggest that the carcinogenic effectiveness of heavy ions at space-relevant dose rates and at high dose rates used in terrestrial experiments may be comparable. Consequently, scaling factors for heavy ion carcinogenesis estimated from moderate/high dose rate experimental data may be applicable for scaling human gamma-ray-induced cancer risks to heavy ions at situations relevant for space exploration. However, animal experiments using multiple small dose fractions and/or very low dose rates of densely ionizing radiation are needed to reduce prediction uncertainties by better quantifying NTE responses at space-relevant dose rates.

In order to answer the critical question of how to reliably estimate heavy ion-induced cancer risks in astronauts embarking on long-distance space exploration missions such as a flight to Mars, we are developing a mechanistically-motivated mathematical model that can predict radiogenic carcinogenesis as function of dose and dose rate using both targeted effect (TE) and non-targeted effect (NTE) contributions. Importantly, such models are needed to predict low dose rate risks based on data at higher dose rates because the very low heavy ion dose rates relevant for space missions are difficult to achieve in terrestrial experiments. Our goal is to calibrate the carcinogenesis model using available human and animal data and to generate scaling factors such as the recently proposed radiation effects ratio (RER), which compares carcinogenic effectiveness of heavy ions and gamma rays at the dose of interest. The scaling factors would then be used to estimate human heavy ion-induced cancer risks, based on human gamma-ray-induced risks. An important focus of our work is generation of realistic uncertainties for model parameters and predictions, which ultimately translate into realistic uncertainties on astronaut risk estimates.

Here we present such a model, which quantifies both TE and NTE contributions. NTE can dominate at low doses/dose rates encountered in space. To estimate an important model parameter – the dose rate at which NTE are 50% activated – we fitted the model to lung carcinogenesis data in radon-exposed miners, which are very useful for heavy ion risk estimation because they contain information on human carcinogenesis induced by densely-ionizing radiation at doses and exposure durations that overlap the range expected during space exploration missions. To calibrate the model for highly-energetic space-relevant heavy ions (C, Si, Fe) we used data from mouse experiments. We then generated model-based radiation effects ratios (RER) values, which compare carcinogenic effectiveness of heavy ions and gamma rays, at doses/dose rates expected during space exploration. The RER is a new metric, conceptually similar to a relative risk, which is useful for risk scaling when dose-effect relationships are complex. We used RER values to scale human gamma ray-induced cancer risks to estimate heavy ion cancer risks in astronauts.

We sought to address the following question: is the carcinogenic effectiveness of heavy ions at the very low dose rates encountered in space substantially different from that at the high dose rates used in terrestrial experiments? For example, is there a direct dose rate effect (i.e., reduction in carcinogenic effectiveness at low dose rates), or an inverse one (i.e., a possible increase in carcinogenic effectiveness)? Providing quantitative answers to such questions by generating RER and risk estimates and uncertainties is important for the planning and design of new terrestrial experiments and – ultimately – manned missions in space.

Our mechanistically-motivated model of densely-ionizing radiation carcinogenesis was able to describe the shapes of the dependences of lung cancer risk in radon-exposed miners on dose rate (exposure duration) and dose. Consistently with the data, the model predicted an inverse dose rate affect – an increase in cancer risk from a given densely ionizing radiation dose when this dose was delivered at a lower dose rate, i.e., over a longer duration.

The model-based interpretation of this phenomenon is as follows. When the exposure is protracted over a longer period, the duration of NTE signal activation is prolonged and NTE-driven carcinogenesis therefore increases. In other words, over a range of doses/dose rates that are sufficient to cause nearly maximal susceptible cell activation, there is an inverse dose rate effect for cancer risk from NTE. At very low doses, however, ionizing particle tracks are too sparse and occur too rarely to maintain high NTE signal levels because NTE signals induced by one track have time to decay away before the next track traverses the target area. Under such conditions reducing the dose rate causes a direct dose rate effect – NTE-driven carcinogenesis at a given dose decreases with decreasing dose rate.

The model parameter responsible for the dose rate dependence of the NTE contribution to cancer risks is the dose rate at which NTE signal levels reach 50% of maximal. The fit to the radon-exposed miner data suggested that this parameter is small. Consequently, substantial NTE activation was predicted at dose rates relevant for a mission to Mars. The combination of dose and dose rate expected for a Mars mission lies close to the transition from direct to inverse dose rate effects, according to model predictions. For such exposure scenarios, the model predicted a small and not statistically-significant inverse dose rate effect, relative to an acute exposure with the same total dose.

Compared with gamma-rays, the estimated carcinogenic effectiveness (RER) of heavy ions at space-relevant doses and dose rates was ~33, with considerable uncertainties. This large RER value and uncertainties, and their dependence on dose rate, were caused by NTE. If NTE terms were absent from the model, RER would be equal to the ratio of TE dose response slopes for ions vs gamma-rays, which would be approximately 1.7 in this case, and there would be no dependence of RER on dose rate. When the RER was used to scale human gamma-ray risks to heavy ions, the estimated heavy ion-induced sex-averaged colon cancer risk for a 40-year-old astronaut was ~2.8%, with considerable uncertainties. The uncertainties are fundamentally caused by paucity of information about NTE response at very low dose rates.

Important components of the current modeling efforts include the following: (1) The radiation carcinogenesis model included both TE and NTE. The latter can be particularly important at the low doses and dose rates of densely-ionizing radiation encountered during space exploration. (2) The model was able to fit the shapes of dose/dose rate dependencies observed in a valuable data set of human cancer risks after protracted densely-ionizing radiation exposure – lung cancers in radon-exposed miners. (3) Carcinogenesis data from animal experiments using space-relevant heavy ions (C, Si, Fe) at appropriate energies were also used to calibrate the model. (4) Monte Carlo simulations and error propagation were used to realistically approximate the errors associated with model parameters and other output. (5) Model predictions were used to scale human gamma-ray-induced cancer risks to heavy ions at doses/dose rates expected during space missions using a recently-developed metric: radiation effects ratio (RER).

Although several simplifying assumptions were used in the modeling, the resulting predictions were consistent with other data sets that were not used in the current analysis: (1) No strong dose rate effects were predicted. This is consistent with animal experimental data using heavy ions at tested dose rates. (2) NTE signals were predicted to be long-lasting, with decay rates on the order of months-years. This is consistent with data that show long-term consequences of irradiation such as chronic inflammation, oxidative stress, and genomic instability.

Our modeling approach and its application to human and animal radiation carcinogenesis data quantify and help to explain the complex effects of dose rate on the dose responses for densely-ionizing radiation such as galactic cosmic rays. Whereas most cellular damage induced by sparsely-ionizing radiation such as gamma-rays is rapidly repaired, leading to direct dose rate effects, damage induced by densely-ionizing radiation is more difficult to repair and often induces NTE. Persistent activation of NTE signaling during/after irradiation can result in inverse dose rate effects.

Our results suggest that the carcinogenic effectiveness of heavy ions at space-relevant dose rates and at high dose rates used in terrestrial experiments may be comparable. The model predicts a small inverse dose rate effect, but the uncertainties overlap a scenario with no dose rate effect.