Skip to main content

Cognitive and Cerebrovascular Effects Induced by Low Dose Ionising Radiation

Final Report Summary - CEREBRAD (Cognitive and Cerebrovascular Effects Induced by Low Dose Ionising Radiation)

Executive Summary:
1. CEREBRAD main findings:
I) Epidemiological evaluations of the risk of cerebrovascular disease following low dose exposures were based upon a cohort of 233 survivors of childhood cancer receiving radiation therapy before the age of 5 year, matched to an equal number of survivors not treated with radiation. The Excess of Odds Ratio (EOR) of stroke per Gy of average radiation dose to the cerebral arteries, was equal to EOR/Gy = 0.49 (95% CI: 0.22 to 1.17) in a linear model.
II) Cognitive impairments have been evaluated in a medical and in Chernobyl cohorts, in which exposure to radiation occurred either in utero, or during childhood below the age of one year, or at adult age in clean-up workers. Impairments appear to be age-dependent; in in utero exposed cohort, effects are observed below 0.1 Gy, in the medical cohort (exposure at childhood below the age of one year), impairments increased with increasing dose to the thyroid and cerebral hemispheres from thresholds of 0.12 Gy and 0.054 Gy, respectively. On the other hand, Cleanup workers demonstrated significant cognitive deficits when exposed to doses over 0, 25 Gy.
III) The shape of the dose-response curve for cognitive impairments in animal models shows a linear dose-response curve with age-dependent sensitivity. In in utero exposed mice, subtle changes in behavior can still be observed with the low dose 0.1 Gy, while early postnatal exposure showed impairments starting from 0.3 Gy on. More importantly, postnatal co-exposure with environmental toxicants (such as MeHg, nicotine and PBDE) showed defects at a dose below 0.1 Gy. In all, our data indicate there might be no threshold below which no effects are observed, warranting thus further investigations.
IV) Investigations in animal models on Blood-Brain Barrier (BBB) permeability and integrity after exposure to low and high doses of radiation indicate contribution of several components such as age at exposure, genetic background and basal inflammatory status in the reponse.
V) Innovative dosimetry calculations were developed on medical cohorts providing accurate retrospective estimates of doses to several brain structures and cerebral arteries. In parallel, animal dosimetry simulations allowed to calculate energy deposition in soft tissue compared to bone in mice exposed to either external radiation or internal contamination.
VI) The cellular and molecular investigations revealed obvious effects of low-dose ionizing radiation 'LD-IR' on the brain at multiple levels. In general, we could observe a clear dose-dependent effect and could unveil different anomalies induced by the lowest X-ray dose studied (0.1 Gy) in terms of cognition, cell death and neurogenesis. finally, mechanisms acting at low doses are different from those at high doses, while, processing of the late response could in part be mastered through epigenetic events, requiring thus additional future investigations.
VII) A dedicated computational bioinformatics platform was developed, called Brain Radiation Information Data Exchange (BRIDE). BRIDE fitted nicely the systems biology approach of CEREBRAD to explore pathways inference from omics data in the context of cognitive deficits.

Project Context and Objectives:
2. Overview of the CEREBRAD project and objectives
Epidemiological evidences about the occurrence of late cognitive and cerebrovascular diseases due to exposure to radiation early in life (in utero or during childhood) are scarce. However, A-bomb survivor data indicate a linear dose-response curve with a threshold around 200 mSv. This raises once more the concern regarding the uncertainty of low-dose radiation. This is in part due to the lack of sufficiently large cohorts to estimate the expected mild effects from low radiation doses, combined with a lack of understanding of the underlying mechanisms. Nevertheless, the increasing use of radiation in medical diagnostics urges the need for appropriate research to define precisely the effect of low dose radiation on the brain. The aim of the FP7 CEREBRAD project (GA n°295552), thus is to gather sufficient scientific evidence to increase the statistical power of epidemiological data. Moreover, the project aims to explain the related cellular and molecular events modulated after exposure and most probably responsible for possible late cognitive and cerebrovascular diseases.CEREBRAD was designed to answer key questions addressed by several reports of experts and recommendation bodies (HLEG, UNSCEAR, ICRP, and MELODI) in order to advance the know-how of radiation effects on the central nervous system:
• Key Question 1: Is the epidemiological data sufficiently robust in estimating the risk of cognitive and cerebrovascular effects below 500 mGy? -Solved in main finding I and II
• Key Question 2: Can the animal studies contribute in addressing the shape of the dose-response curve for cognitive and cerebrovascular effects at different stages of brain development as non-cancer detriment? -Solved in main finding III and IV
• Key Question 3: How to reduce the uncertainties of dose reconstruction in the brain? -Solved in main finding V
• Key Question 4: What are the underlying biological mechanisms for such non-cancer effects of radiation? -Solved in main finding VI and VII

3. CEREBRAD results: Increasing the statistical power of epidemiological data
3.1. Cerebrovascular diseases
We set up a case-control study, including 233 cases of strokes having occurred 5 years or more following a childhood cancer and 233 controls matched on cohort, gender, age and date of childhood cancer, and length of follow-up. We performed very detailed radiation dose estimation in all brain structures and in cerebral arteries. When all types of strokes were considered as “stroke” in general, the excess odds ratio (EOR) of stroke per Gy of average radiation dose to the cerebral arteries, was equal to EOR/Gy = 0.49 (95% CI: 0.22 to 1.17) in a linear model. Adding an exponential or a quadratic term to the model did not improve the fit of the data. The radiation dose received to brain structure other than brain arteries did not play any role.
This risk is in line with the ones observed in most of the other studies. In A-bomb survivors the EOR/Gy was equal to 0.09 (95% CI: 0.01 to 0.17) when considering stroke as underlying cause of death and EOR/Gy = 0.12 (95% CI: 0.05 to 0.19) when considering stroke as underlying or contributing cause of death (Shimizu, Kodama et al. 2010). However, when considering only survivors who were less than 10 years old at time of exposure, as in our cohort, the EOR/Gy was equal to 0.36 (Shimizu, Kodama et al. 2010). In a meta-analysis of several cohort studies, including the international nuclear workers study and the A-bomb survivor cohort, the EOR/Gy was estimated to be 0.27 (95% CI: 0.20 to 0.34) for stroke (Little, Tawn et al. 2010, Little, Azizova et al. 2012). Lastly, in Mayak workers, the EOR/Gy for external radiation was recently estimated as being 0.33 (95% CI: 0.19 to 0.50) (Simonetto, Schollnberger et al. 2015).
3.2. Cognitive effects of low doses
Several groups have been evaluated to estimate neurocognitive impairments at different stages of brain development.
-Exposure in utero
Subjects exposed in utero exhibited an excess of autonomic nervous system disorders (ICD-10: G90) at the age of 25-27. Neurological microsymptoms as well as neurotic, stress-related and somatoform disorders predominate. Nevertheless neither this excess, nor the value of any of the neurocognitive tests, could be related to the estimated radiation dose received in utero to brain, thyroid or whole body.
-Medical irradiation during childhood
The study was deliberately designed in a similar way to the study conducted by Hall et al. and Blomstrand et al. to enable comparisons (Hall et al. 2004; Haddy et al. 2011). A higher total radiation dose (brachytherapy and X-rays) to the cerebral hemispheres was significantly associated to a lower education (p = 0.035). Nevertheless the total radiation dose received in cerebral hemispheres, whatever the structure considered, was not significantly linked to any of the neurocognitive test used in our study, except for a near from significant result for the depression ‘HAD-D score’ when considering the left hemisphere. Higher average radiation dose to the cerebral hemispheres was also significantly or nearly significantly associated to a decline in the value of most of neurocognitive tests
Depression appeared to increase with increasing dose to the thyroid and hemispheres from thresholds of 0.12 Gy and 0.054 Gy, respectively, while cognitive impairments scores increased significantly with radiation dose to the left hemisphere (threshold 0.059 Gy). The maximum brachytherapy dose to the temporal lobes was also significantly associated to this score above a threshold of 0.054 Gy. Although we found an effect of low-dose radiation on verbal cognitive performance, it should be emphasized that the cohort was rather small (115 subjects), warranting caution before making statements about causality.
-Exposure at adulthood (Chernobyl cleanup cohort)
Subjects exposed to ionizing radiation at adulthood as cleanup workers exhibited symptoms of mild cognitive impairment according to the operational criteria of the mini–mental state examination (MMSE) (mean group scores range = 24-27). A dose-dependent increase in the incidence of mental disorders was evidenced according to the Brief Psychiatric Rating Scale (BPRS), Zung Index and Global Health Questionnaire (GHQ-28), but not for MMSE. This increase remained significant when restricting the analysis to the workers having received less than 250 mSv. In comparison with previous studies an excess of cognitive dysfunction was significant only for doses of 250 mSv and higher.
Cognitive function seems to be influenced by the radiation dose and age at exposure. The incidence seems higher in young adults exposed in utero than the exposed adults.
4. CEREBRAD results: animal studies
Cognitive and cerebrovascular studies were also conducted in animal models, we used either C57Bl/6 or NMRI mice that were exposed to prenatal irradiation at embryonic day 11 (E11) or to irradiation after birth at postnatal day 10 (PND10) or at adult age at week 10 (W10).
4.1. Cerebrovascular effects of low doses
Our data are of particular importance, since they are corroborated by previous research but also contradict other studies. A radiation-induced blood–brain barrier (BBB) breakdown has been supposed to explain the acute radiation syndrome and the delayed brain radiation injury, but it has been clearly demonstrated only at very high doses (Remler, Marcussen et al. 1986). In our study, local brain irradiation induced acute endothelial cell activation in the cortex, hippocampus and cerebellum in W10 irradiated mice, and in the cerebellum in PND10 irradiated mice, indicating a higher sensitivity of older mice to radiation induced acute inflammatory reactions compared to young mice. Next, a very important finding was the chronic radiation-induced BBB damage in the hippocampus and cerebellum of W10 irradiated animals and in the hippocampus of PND10 irradiated animals, which was induced both by low and high doses. Yet, it should be noted that only a small number of animals could be used for these experiments, leading to a relatively high standard deviation. As such, the trend towards a radiation-induced BBB damage, which is present even 1 month after irradiation, is very promising but will need further confirmation at later time points and with more biological replicates.
4.2. Cognitive defects
-In utero exposure to radiation
Impairments in behavior (reduced activity, increased sociability and a declined spatial learning) were observed in mice externally irradiated with high doses (0.5 and 1 Gy). Of interest, for internal exposure, the increased sociability/decreased anxiety could be recapitulated in the lower activities (500 and 1000 Bq/kg). Thus, these data clearly indicate persistent aberrations in cognition and learning as a result of prenatal exposure to irradiation. Although, no threshold dose could be adopted since depending on the investigated brain region, a trend of a linear dose-response curve could be observed for most of the behavioral tests performed, with in general significance for 0.5 and 1.0 Gy external doses. However, and even more importantly, some subtle changes in higher behavioral functions related to strategy could be observed at low dose of external IR (0.1 Gy). Thus, our results clearly indicate cognitive defects at low doses, thus warranting further investigations.
-Early postnatal exposure to radiation
Mice neonatally exposed to external radiation significantly displayed differences in behavior starting from 0.5 Gy of gamma-irradiation, while showing a clear dose-response at 1.0 Gy. In case of internal contamination, we again noted a difference in exploratory behavior (less anxiety) and, surprisingly, also an improved working memory in animals exposed at PND10 compared to other groups. Even though slight differences in dose-responses were observed, we still can conclude that behavior is affected in in utero and PND10 exposed animals. Therefore, we need to address this issue of low-dose ionizing radiation "LD-IR" induced persistent cognitive effects in our community, to improve health assessment.
4.3. Combined exposure to radiation and toxicants
The uncertainty of epidemiological studies at low doses is partly due to the interaction with confounding factors such as smoking, alcohol consumption and exposure to chemicals. Simultaneous exposures to IR and environmental and/or social toxicants are of concern but is commonly not directly addressed in epidemiological studies. A first attempt to gain momentum on this issue has been addressed in CEREBRAD.
Results obtained within CEREBRAD have confirmed that neonatal co-exposure of IR and the environmental agents MeHg, PBDE 99, paraquat and nicotine can cause a change/shift in the dose-response curve, thereby lowering the dose of IR to below 100 mGy for induction of functional disturbances. These functional defects can be seen at doses where neither IR nor the toxic agents alone caused any effects.
Based on this finding, and since we are living in a mixed environment combining different physical and chemical agents, a threshold theory cannot be adopted. Future research need to focus on combining more than two agents to be more in line with our real life. Additionally, extrapolation of this research to specific diets to investigate life style would emphasize other elements of our modern society that might contribute to radiation risk estimate.

5. Mechanistic studies
Next, based on our Omics investigations, we could differentiate effects at low and high doses as well early and late responses. Prenatal exposure to X-rays shows induction of apoptosis as early as 6 h after exposure with both low (0.1) and high (1.0 Gy) doses in a dose-dependent way, at the molecular level this response is preceded (at 2 h after exposure) by a transcriptomic induction of p53 and downstream responsive genes, from which we identify a set of unknown genes and are now under further investigations. The high apoptotic rate induced by the high doses (0.5 and 1.0 Gy) seems to induce morphological changes (reduced brain size, enlarged ventricles and reduced cortical thickness) persistent at adult age as well as defect in adult neurogenesis, both would most probably contribute to the consistent cognitive deviations observed in 0.5 and 1.0 Gy exposed animals. Although, the low dose (0.1 Gy) don't seem to induce any morphological changes, but appear to be more effective on the long term regarding redox state and mitochondrial functions, which are highly involved in ageing processes. Preliminary findings indicate the implication of epigenetic changes which might relate the early effects with late disease occurrence, although several validations are still ongoing and needs to be thoroughly investigated. Besides, further and dedicated studies eventually involving animal models with mutations in pathways of interest that could synergize with low doses of irradiation will be required to better understand the impact of our findings on human health and disease.
6. General conclusion
In all, the CEREBRAD consortium has moved forward the knowledge about cognitive and cerebrovascular disease outcomes, by increasing the statistical power of epidemiological data. With an appropriate retrospective dosimetry on medical cohorts, we provide better estimates of doses to several brain structures. Consequently we attempt to correlate the absorbed doses with either cerebrovascular or cognitive outcomes to define the most compromised brain regions. In parallel, cognitive and cerebrovascular effects have been evaluated in animal models. Thus, CEREBRAD provides extensive information unraveling molecular and functional details of LD-IR effects. Our results revealed obvious effects of LD-IR on the brain at multiple levels and opened prospective avenues for further exploration. In general, we could observe a clear dose-dependent effect and could unveil different anomalies induced by the lowest X-ray dose studied (0.1 Gy) in terms of cognition, cell death and neurogenesis. Of interest, next to the administered dose, also the age at which exposure occurs strongly influences radiation effects, which are mostly exacerbated when irradiation is induced at earlier stages. Yet, this is sometimes highly dependent on the brain region of interest. At the start of the project we did not believe to see effects at low doses up to six month after exposure, today CEREBRAD opened new windows and concepts to investigate late outcome and had established the methodologies for future research to push our curiosity to investigate the mechanisms in the real low dose region (below 0.1 Gy)
CEREBRAD results are emphasized in a number of published papers in the open scientific literature, and have been presented in large scientific events at major international conferences. Together with the Procardio project, our joint training activities have been successful for several graduate students (MSc & PhDs). Finally, the public CEREBRAD website was regularly updated with new data to inform the general public about our research.

Project Results:
SEE PDF ATTACHED "DELIVERABLE 1.4" FOR VERSION WITH FIGURES AND TABLES
1. Introduction
Up to now, the direct effects of ionizing radiation (IR) on the central nervous system remain elusive and are subject to many debates and uncertainties, especially concerning low doses of irradiation (LD-IR). In the context of the FP7 CEREBRAD (Cognitive and Cerebrovascular Effects Induced by Low Dose Ionizing Radiation, grant agreement n°295552) project, we set the stage to answer these questions by means of two approaches: (1) a direct health assessment through epidemiological studies on exposed individuals and (2) an investigation of dose-dependent and radiation-type dependent biological effects using a mouse model. Furthermore, to correctly inform on the risk estimates, we compared internal and external exposure paradigms and evaluated a possible synergistic effect of radiation with other environmental pollutants. This multidisciplinary approach was achieved by the joint effort of a European consortium including radiobiologists, epidemiologists, neurobiologists, bio-informaticians, paediatricians and dosimetrists. Finally, and importantly, our obtained results were shared with other researchers outside the radiobiology field, with stakeholders and with the general public via workshops, peer-reviewed publications, the creation of a website (http://www.cerebrad-fp7.eu/) and the organization of training courses and student's days.
2. Human data

2.1. Cerebrovascular effects of low doses

We set up a case-control study, including 233 cases of strokes having occurred 5 years or more following a childhood cancer and 233 controls matched on cohort, gender, age and date of childhood cancer, and length of follow-up. We performed very detailed radiation dose estimation in any structure of the brain and in cerebral arteries In a linear model, the EOR of stroke, all types together, per Gy of average radiation dose to the cerebral arteries, was equal to EOR/Gy=0.49 (95%CI: 0.22 to 1.17). To add an exponential or a quadratic term did not improve the fit of the data. The radiation dose received to brain structure other than brain arteries did not play any role.
Our findings strongly differed according to the type, ischemic or hemorrhagic of the cerebrovascular diseases. When considering hemorrhagic strokes, an exponential model fitted better the data. Therefore the risk due to low doses was low (EOR/1GY=0.13 (95%CI : 0.07 to 0.21). At the opposite, when considering ischemic strokes, a linear (negative) exponential model was the best model. In this model the linear for low dose was very high: 2.64 (95%CI: 0.39 ; 17.18) (for more detail see D.2.2).

The EOR/1Gy we evidenced for cerebrovascular diseases, considered together (EOR/Gy=0.49 (95%CI: 0.22 to 1.17)) is coherent with the ones observed in most of the other studies. In Hiroshima Nagasaki survivors, in whom the EOR/1Gy was equal to 0.09
(95%CI: 0.01 to 0.17) when considering stroke as underlying cause of death and EOR/1Gy=0.12 (95%CI: 0.05 to 0.19) when considering stroke as underlying or contributing cause of death (Shimizu, Kodama et al. 2010), but, this cohort the EOR/1GY was equal to 0.36 in survivors who were less than 10 years old at time of atomic bomb (Shimizu, Kodama et al. 2010), what was the age of most of the children at time of radiotherapy in our cohort. In a meta-analysis of several cohort studies, including the international nuclear workers study and the Hiroshima-Nagaski cohort, the EOR/1Gy has been estimated to 0.27 (95%CI:0.20 to 0.34) for stroke (Little, Tawn et al. 2010, Little, Azizova et al. 2012). Lastly, in Mayak workers, the EOR/1GY for external radiation has recently been estimated as being 0.33 (95C%I: 0.19 to 0.50) (Simonetto, Schollnberger et al. 2015).
Our results concerning ischemic strokes have to be considered carefully because of the relatively small number of cases (n=97). Nevertheless, the very high EOR/1Gy we evidenced for low doses, 2.64 (95%CI: 0.39 ; 17.18) could be of importance if verified. At our knowledge, up to now, no other study focused ischemic strokes.
In our study, we only evidence a risk of radiation dose to the cerebral arteries, and not to other brain structures or organs. In particular, we did not evidence a role of the radiation dose received to the kidneys, which is known to induce hypertension. This finding is coherent with the one of the cerebrovascular disease mortality study previously published by IGR/INSERM team (Haddy, Mousannif et al. 2011).
All these results are nevertheless based on average radiation dose to the cerebral arteries. It has to consider that the very strong gradients of dose near to borders of the radiation therapy fields, have as a consequence a very strong heterogeneity of dose within the cerebral arteries, whatever the average radiation dose. Therefore the results of any analysis based on only one parameter of the dose distribution within cerebral arteries (mean, median mode, minimal...) are strongly dependent of the dose distribution, and may not be extrapolated, at all, to uniform irradiation (for more detail see D.2.1). Further analyses will be performed using methods in course of development, able to take into account the distribution of the radiation dose within the cerebral
arteries, rather than only one parameter. These methods will include functional statistics and analyses of isovolumes.
2.2. Cognitive effects of low doses

2.2.1. Medical irradiation during childhood
The study was deliberately designed in a similar way to the study conducted by Hall et al. and Blomstrand et al. to enable comparisons (Hall, Adami et al. 2004, Haddy, Mousannif et al. 2011).
The study cohort consisted of the ANGIO cohort. These subjects were treated in the vast majority before the age of one year. This cohort was established between 1985 and 1995 by IGR/INSERM team to study radiation-induced pathologies (Dondon, de Vathaire et al. 2004, Haddy, Andriamboavonjy et al. 2009, Haddy, Dondon et al. 2010, Haddy, Mousannif et al. 2012). Doses of ionizing radiation received by all organs of the body were estimated for all these children, regardless of the site of the haemangioma.
Topics to be included have been selected to people living in the Ile-de-France region. One hundred sixty-seven individuals whose brain radiation received dose estimates are less than 1 Gy to the brain have been identified. A total to 115 subjects were interviewed, the average age at time of questionnaire tests being 50 (from 42 to 63). Neurocognitive assessments of participating subjects undergo and initial interview based on 7 standerdized questionnaires (for more detail see D.2.3).

The RAVLT test and particularly the “delay recall” task is specific to evaluate the episodic memory. Our finding concerning the role of the maximum brachytherapy dose to the temporal lobes in the RAVLT test scores seems relevant since the episodic memory uses neural networks in the hippocampus and more broadly in the inside of the temporal lobes. Indeed, The hippocampus appears to play a central role in the temporary and more durable storage explicit information related to different cortical structures (Hall, Adami et al. 2004).
The MoCA test involved many cognitive domains (executive function, language, memory) and most of patients lost points in memory task. It could explain the relationship between the temporal dose and the MoCA test score but this test is too general and uses several brain structures to conclude a causal relationship.
A higher total radiation dose (Brachytherapy and X-rays) to the cerebral hemispheres was significantly associated to a lower education (p=0.035). Nevertheless the total radiation dose received in cerebral hemispheres, whatever the structure considered was not significantly linked to any of the neurocognitive test used in our study, at the exception of a near from significant result when evaluating depression based on HAD-D score when considering left hemisphere (Table 2 6).
A higher average radiation dose to cerebral hemisphere was also significantly or nearly significantly associated to a degradation in the value of most of neurocognitive tests we used (Table 2 6, Table 2 7): The FactCog (Perceive cognitive impairments) and HAD-D scores were more degraded with higher average radiation dose to the brain hemispheres than with higher radiation dose to other structures, whereas RALVT Decay recall and MOCA scores were more impacted by average radiation doses to the temporal lobes. Among all dosimetric parameters, no one had a significant correlation with the HAD-A and FactCog (perceived cognitive ability).
For the HAD-D test there was a trend for increasing scores with increasing dose to the thyroid and with the maximum brachytherapy dose to the Hemispheres from thresholds equal to 0.12 Gy and 0.054 respectively. Approximately the same threshold (0.059 Gy) of the radiation dose to the left hemisphere lob is obtained to show a significant increase of the FactCog Perceive cognitive impairments scores. RALVT delay recall scores according the years schooling (threshold = 3 years). The maximum brachytherapy dose to the temporal lobes was also significantly associated to this test scores above 0.054 Gy.
Thyroid tissue is one of the most radiosensitive human tissues and hypothyroidism and hyperthyroidism were previously been associated with depression (Blomstrand, Holmberg et al. 2014) and our results concerning the correlation between the mean thyroid dose and the HAD_D scores could be explained by a possible impact of a low doses on the thyroid hormone regulation functions. Although we found an effect of low-dose radiation on verbal cognitive performance, it should be emphasized that we analysed few subjects (115), warranting caution before making statements about causality. Besides that, children with the largest hemangiomas on the face were likely treated with the highest doses, resulting in the most mutilating scars. This may have influenced their self-esteem or confidence and thereby also their test results and performance in the educational system (for more detail see D.2.3).
More detailed neuropsychological examinations on a larger patient dataset, e.g. analysis of executive functions, would likely have enabled us to draw further conclusions about the nature of the effects of low-dose radiation, if any, on cognitive performance.

2.2.2. Chernobyl studies
Cognitive function is influenced by the radiation dose and age at exposure. The level of subjective distress caused by traumatic event is higher in young adults exposed in utero/ There is some increase of somatoform symptoms and levels of anxiety, insomnia and social dysfunction (for more detail see D.2.4).
Subjects exposed in utero during the check at age of 25–27 years exhibit an excess of the disorders of autonomic nervous system (ICD-10: G90). Neurological microsymptoms as well as neurotic, stress-related and somatoform disorders (F40–F48) dominate.
Subjects exposed to ionizing radiation at adulthood as cleanup workers exhibit symptoms of mild cognitive impairment according to the operational criteria of the MMSE (mean group scores range =24–27). The cleanup workers have significantly higher level of mental disorders according to the BPRS in dose-related manner, than young adults. This could be the effect of the age and radiation dose. Cleanup workers exposed to doses over 250 mSv and, especially, 500 mSv demonstrate significant cognitive deficit in comparison with exposed below 250 mSv and non-exposed patients. In comparison with previous studies an excess of cognitive dysfunction was significant at doses of 250 mSv and higher (for more detail see D.2.4).
3. Animal studies
For all animal studies, we used either C57Bl/6 or NMRI mice that were exposed to prenatal irradiation at embryonic day 11 (E11) or to irradiation after birth at postnatal day 10 (PND10) or at postnatal week 10 (W10). Different modes of radiation were used, including whole-body irradiation (pre- and postnatal IR) or local cranial irradiation (postnatal IR). Afterwards, different post-irradiation time points were chosen, ranging from hours to days and months after radiation exposure. Mice were exposed to a range of low to moderate doses of X-rays or gamma-irradiation (Co-60/Cs-137) for external irradiation, and to Cs-137 for internal contamination. Dosimetry simulations using Monte-Carlo codes allowed to evaluate the distribution of the doses in soft tissues and bones to correctly estimate the doses absorbed by the embryonic brain (for more detail see D.3.2)
In the following paragraphs, the effects of LD-IR on the cerebrovascular system and cognition will be discussed, and possible underlying mechanisms are proposed.
3.1 Cerebrovascular diseases
To assess whether prenatal/neonatal radiation exposure exerts an effect on the brain vasculature, we studied the effect of local head irradiation on blood-brain barrier (BBB) damage and repair, known to contribute to a proper brain functioning and related to an increased cell ageing. To this end, whole-brain irradiation of animals and
humans has indeed been reported to lead to late delayed vascular damage (Yoshii and Phillips 1982).
Local brain irradiation induced acute endothelial cell activation in the cortex, hippocampus and cerebellum in W10 irradiated mice, and in the cerebellum in PN10 irradiated mice, indicating a higher sensitivity of older mice to radiation induced acute inflammatory reactions compared to young mice. Next, a very important finding was the chronic radiation-induced BBB damage in the hippocampus and cerebellum of W10 IR animals and in the hippocampus of PND10 IR animals, which was induced both by low and high doses. Yet, it should be noted that only a small number of animals could be used for these experiments, leading to a relatively high standard deviation. As such, the trend towards a radiation-induced BBB damage, which is present even 1 month after irradiation, is very promising but will need further confirmation at later time points and with more biological replicates (for more detail see D.3.4).
In any case, our data are of particular importance, since they are corroborated by previous research but also contradict other studies. A radiation-induced blood–brain barrier (BBB) breakdown has been supposed to explain the acute radiation syndrome and the delayed brain radiation injury, but it has been clearly demonstrated only at high doses (Remler, Marcussen et al. 1986). Furthermore, a previous study has shown that 20Gy and 40Gy brain irradiation produced an early permanent increase in BBB permeability in rats, while 10 Gy had no effect at all (Liu, Xiao et al. 2010). Finally, Mao and colleagues demonstrated a time- and dose-dependent loss of the vasculature following gamma and proton radiation exposure in rodents, and decrements in vessel growth were found and could be observed as long as 12 months after a single 8- or 28-Gy exposure (Mao, Archambeau et al. 2003).
3.2 Cognitive defects

3.2.1 Single radiaton exposure
To address persistent effects of external/internal irradiation at the embryonic or early postnatal stage, we subjected animals to a battery of behavioral tests (neuromotor, exploration and learning tests).
In case of in utero radiation at E11 (external IR: from 0 to 1.0 Gy, or internal IR: from 500 to 50000 Bq/kg), mice externally irradiated with 1 Gy were overall less active when compared to other groups. Further, this group showed an increased sociability and a declined spatial learning. Concerning internal IR, Cesium low/mid activity exposed animals show an improvement in the working memory, spontaneous behavior and less anxiety than control animals (for more detail see D.3.1).

We suggest that in future studies, the range of the activity of internal contamination should be expanded towards a range resulting in doses comparable to the external irradiation doses.
Thus, these data clearly indicate persistent aberrations in cognition and learning as a result of prenatal exposure to irradiation. Even more importantly, as was seen for swim strategies in the Morris water maze, a low dose of external IR (0.1 Gy) already showed difficulties in finding the hidden platform (Figure 1).
In contrast, mice neonatally exposed to external radiation only displayed differences in behavior starting from 0.5 Gy of gamma-irradiation, while showing a clear dose-response at 1.0 Gy. This discrepancy might be explained by the use of different behavioral paradigms (water maze strategies vs. spontaneous behavior), addressing different aspects of behavior. As such, additional behavioral tests for our groups would be of interest to soundly analyze radiation-induced cognitive aberrations. Importantly, by performing behavioral tests on both males and females, we could rule out a possible gender effect that could be attributable to the observed dose differences. Similarly, differences in behavior between mouse strains (C57Bl6 vs NMRI) could be ruled out.
In case of internal contamination, we again noted a difference in exploratory behavior (less anxiety).

Even though slight differences in dose-responses were observed, we still can conclude that behavior is similarly affected in in utero and PND10 exposed animals. Therefore, we need to address this issue of LD-IR induced persistent cognitive effects in our community, to improve health assessment (for more detail see D.3.1).
3.2.2 Combined exposure to radiation and toxicants
As a second main aim, based on the high risk for consequences of exposure to IR and toxic agents of the developing nervous system, we characterized the (synergistic) effect of radiation and toxicants such as PBDE, methylmercury, paraquat and nicotine on mouse behavior at the adult age of 2 and 4 months, preceded by irradiation at PND10.

External irradiation
The results obtained within CEREBRAD indicate a synergistically defective spontaneous behavior in IR+PBDE exposed mice, suggestive for an altered cognitive function in adult mice neonatally exposed to gamma irradiation at doses where the sole compounds did not cause any effect. The effects on single exposure are in agreement with earlier published reports on IR (Eriksson, Fischer et al. 2010, Buratovic, Stenerlow et al. 2014) and PBDE99 (Eriksson, Fischer et al. 2006).
In agreement with earlier published work (Fredriksson, Fredriksson et al. 1993), we additionally showed an interaction between IR and 0.4 mg/kg MeHg (Eriksson, Fischer et al. 2010) as well as with paraquat (dose-dependent) and nicotine (for more detail see D.3.5).

Dose-Response curve
In general, we propose a shift in the dose-response curve when such environmental toxicants are combined with IR exposure, resulting in a lowering of the threshold dose of about 300 mGy (Figure 2) (for more detail see D.3.3).
Based on this finding, and since we are living in a mixed environment combining different physical and chemical agents, a threshold theory cannot be adopted. Future research need to focus on combining multiple agents to be more in line with our real life. Additionally, extrapolation of this research to specific diets to investigate life style would emphasize other elements of our modern society that might contribute to radiation risk estimate.

Internal irradiation
A single subcutaneous dose of Cesium-137 was combined with paraquat, bisphenol A, nicotine or MeHg. While paraquat, nicotine or bisphenol A alone increased anxiety, co-exposure of either of those toxicants with lower doses of Cs caused a reversion of this effect. In addition, unexpectedly, Cs and co-exposed groups spent less time searching in
the water maze target quadrant, while paraquat alone showed an increased time spent in the target quadrant. Nonetheless, no clear additive or synergistic effect of IR and environmental toxicants could be observed using the current internal contamination paradigm. A more detailed analysis will thus be required to disentangle these remarkable findings, for example by using higher doses of internal contamination (for more detail see D.3.5). For an overview of all cognitive tests performed within CEREBRAD and the main conclusions drawn, we refer to table 1.
3.3 Underlying cellular and molecular mechanisms
To explain the observed cerebrovascular and cognitive defects that result from pre/neonatal radiation exposure, we investigated early and late cellular and molecular events that might be at the origin of these anomalies.

3.3.1 Early effects

Neurogenesis & corticogenesis
In the developing neocortex, we noted an impact of prenatal LD-IR on different aspects of brain development as well as on brain cytoarchitecture, as demonstrated by a defective hippocampal neurogenesis and differentiation. Our data also demonstrate that the developing neocortex is, next to the hippocampus, highly susceptible to LD-IR. Further studies will have to be designed to investigate permanent defects in this anatomical region following prenatal irradiation. In any way, these first indications could potentially explain the observed permanent behavioral changes that cannot solely be attributed to hippocampal aberrations.
Early genetic changes after pre- and postnatal radiation that are linked to a deviant neurogenesis and cortical development might be attributed to a p53-mediated DNA damage signaling and apoptosis, which is probably cell-type specific. Besides, we showed a dose-dependent alternative transcription of a shorter isoform for several genes in the irradiated embryonic brain 2 h after X-irradiation, for which p53 was shown to bind the promotor sequence of this short isoform (for more detail see D.4.6). On these premises, we believe that the exact mechanisms explaining LD-IR long-term effects at the organism level can be unraveled only by achieving a better understanding
of the early effects (hours to days) and at the level of the different neuronal populations, of which we provided first important insights.
In all, multiple developmental processes crucial for the correct wiring and functioning of the brain, such as apoptosis, cell division and cell differentiation, were altered following radiation with low and/or high doses of irradiation.
Yet, more focused research with emphasis on cell-type specific apoptosis and cell cycle arrest, or investigating a possible premature differentiation or a delayed migration will be key to fully comprehend the consequence of LD-IR on brain development.

3.3.2 Late effects

Neurogenesis
Similar as for early effects, we noted an impaired juvenile hippocampal neurogenesis with some defects persisting until 6 months after in utero and postnatal radiation exposure. Intriguingly, in animals irradiated at 10 weeks of age and analyzed 6 months after irradiation, the 0.1 Gy dose induced a significant increase in the number of neurons, pointing at potential hormetic mechanisms. Whether the changes elicited by such low doses are potentially detrimental, for instance by synergizing with other environmental or genetic factors, remains to be addressed (for more detail see D.4.2 D.4.3).
Chronic inflammation
Nowadays, there is a growing awareness that even low doses of irradiation can cause long-term defects and possibly also a chronic elevation of inflammation. We hypothesize that this neuroinflammatory milieu contributes to the observed neurocognitive defects, including inhibition of hippocampal neurogenesis and synaptic function (see further).
At 6 months after external irradiation, we found that apoptosis and microglia activation only occurred following 2 Gy IR and only in PND10 irradiated mice in comparison to W10 irradiated mice, suggesting a higher sensitivity of the young brain to radiation effects. In addition, we could suggest a radiation-induced neuroinflammation based on a persistent astrogliosis, again only after high dose irradiation at PND10 but not at W10. Finally, and in contradiction to the aforementioned, we could note a change in the microvessel length of the hippocampus only in W10 irradiated animals. Furthermore, even though radiation was shown to induce inflammation in irradiated animals, no synergistic effect was noted when animals were predisposed to inflammation (EAE mice). On the contrary, the effects were antagonistic, since irradiation decreased inflammation-mediated cytokine upregulation both after low (0.1 Gy) and high (2 Gy) radiation doses, at least for CD54 (for more detail see D.4.2 D.4.3).
At 2 and 4 months after internal Cs exposure at E11, only a few cytokines were shown to be differently expressed in the hippocampus, while BDNF expression showed the strongest increase at 20 weeks post-irradiation. Similarly, for PND10 irradiated groups, a few trends were apparent, especially for BDNF expression at 40 weeks post-irradiation, but a high standard deviation prevented us from making sound conclusions.

Neuronal plasticity and communication
Based on our proteomics studies, we discovered a persistent molecular fingerprint in the brain after neonatal irradiation. This fingerprint included both proteins and miRNAs that were always found to be deregulated in a similar manner in different mouse strains (NMRI and C57Bl/6), in both genders and in both the hippocampus and cortex. This finding was clearly observed after a dose of 0.5 Gy but not after a low dose of 0.1 Gy. The molecular components of this fingerprint included Rac1 (always downregulated) and cofilin (always upregulated), which are key players in the Rac1-Cofilin pathway. The latter is essential for neuronal spine morphology and maturation, hippocampal synaptic plasticity and spatial learning, and thus correlates well with the aberrant learning observed in adult mice irradiated at E11 or PND10 (for more detail see D.4.6).
Brain morphology
We investigated brain regional differences via a voxel-based MRI morphometric approach. From this, we revealed a clear decline in total brain volume, accompanied by enlarged ventricles and a relative decrease in volume of the prefrontal cortex in 1.0 Gy irradiated animals, which indeed indicates a correlation with the Morris water maze results. Yet, other factors might be in play, since behaviour was also affected at doses below 1.0 Gy. As such, additional analyses need to be performed to unveil all causes leading to an aberrant learning and cognition, e.g. by focusing on other brain regions or on more subtle effects as compared to a reduction in brain size (for more detail see D.4.2).
Mitochondrial function
Mitochondrial function and basal respiration in particular, was shown to be altered in whole body irradiated mice, 4 weeks post-IR. At 24 weeks post-IR, however, respiration at higher doses returned to normal levels and alterations were only significant in the 0.1 Gy exposed group. Remarkably, in this low dose irradiated group, respiration was found to be higher than in controls, which is suggestive for hormetic mechanisms activated by very low radiation doses.

Irradiation also altered the neuronal oxido-reductive state and its associated signaling, and effects were again detectable at the lowest 0.1 Gy dose. Interestingly, the dose-response relationship between mitochondrial respiration, oxido-reductive state, or signaling and LD-IR differed between animals analyzed 4 or 24 weeks after irradiation, and was only found to be linear in the former case. These data are consistent with a threshold model in which an adaptive response is activated only when factors signaling the perturbation surpass a certain edge (illustrated in Figure 3). Unambigious validation of this model, however, would require additional and extensive research, for instance monitoring the different parameters (e.g. mitochondrial function, oxido-reductive state, signaling) in real-time with advanced imaging methods (for more detail see D.4.3).
Experiments to investigate mitochondrial function and oxido-reductive changes in locally irradiated mice confirmed a different susceptibility to LD-IR at the anatomical level, being that the cerebellum is less sensitive than the cortex or the hippocampus. Alterations were often not linear in time and presented biphasic trends. Importantly, also in this case, the 0.1 Gy dose was sufficient to induce radiation-induced defects (for a general overview of changes in the cortex, cerebellum and hippocampus, see table below).
Of note, the mode of irradiation seems to influence the outcome, given that the results observed with total body irradiation do not fully overlap with those observed with local cranial irradiation. For instance, 24 weeks after 2 Gy irradiation, protein carbonylation was significantly decreased in animals subjected to local irradiation, while no significant differences could be observed at the same time point in animals that had been exposed to whole-body irradiation. The same applies to mitochondrial analyses, even though local and whole body irradiated samples have been generated and analyzed in different laboratories. Differences observed in total- versus local-irradiation might be attributable to systemic effects generated by LD-ID peripherally and related to processes such as inflammation and/or hormonal changes that might ultimately have consequences to the brain for more detail see D.4.3).
4. General conclusions
The consequences of LD-IR on the brain are dimly understood. CEREBRAD provided extensive information unraveling molecular and functional details of LD-IR effects.
Our results revealed obvious effects of LD-IR on the brain at multiple levels and opened prospective avenues for further exploration. In general, we could observe a clear dose-dependent effect and could unveil different anomalies induced by the lowest X-ray dose studied (0.1 Gy) in terms of cognition, cell death and neurogenesis. Of interest, next to the administered dose, also the age at which exposure occurs strongly influences radiation effects, which are mostly exacerbated when irradiation is induced at earlier stages. Yet, this is sometimes highly dependent on the brain region of interest, for example the cerebellum which is more vulnerable to neonatal radiation as compared to prenatal radiation.

Of interest for our studies, the transcriptional and protein landscape emerged from microarray and proteomics studies confirmed alterations in processes such as inflammation, mitochondrial electron transport, neuron development and axonogenesis. In addition, we could identify several new p53 target genes that are induced in a dose-dependent, transient manner at 2 h after prenatal radiation exposure. However, while the early responses are easy to decipher, the long-term effects are more difficult to interpret. This is most likely related to the nature of the insult (i.e. acute radiation), which results in a strong, yet transient biological effect (DNA damage), and in turn activates a multitude of downstream biological processes. Furthermore, the possibility of radiation-induced epigenetic changes is a challenging field of research, and needs to be thoroughly investigated. Besides, further and dedicated studies eventually involving animal models with mutations in pathways of interest that could synergize with low doses of irradiation will be required to better understand the impact of our findings on human health and disease.
The experimental data obtained within CEREBRAD were organized, integrated and elaborated with an extraordinary bioinformatics approach that, to our best knowledge, is unprecedented in the field of radiation biology (for more detail see D.4.1 D.4.4 D.4.5).
One aim was to classify and integrate functional, molecular and behavioral data in a custom made computational platform, which we called Brain Radiation Information Data Exchange (BRIDE). BRIDE is extremely rich as well as quite challenging to explore for pathways inference in the context of particular phenotypes under study, such as cognitive deficits in adult, prenatally exposed mice. Obviously, the BRIDE platform can be further improved to become an established data mining approach among radiation biologists. One potential future plan is to provide a text-mining suite for literature scans, so that the contents can be maintained current, with minimal effort by semi-automated monitoring of the literature. Furthermore, importantly, our systems biology approach confirmed that exposure with 0.1 Gy alters biological processes in the brain. Hereto, it is essential to reiterate that, at present, it is unclear whether these changes are detrimental and further focused studies will be critical to conclusively address this issue (for more detail see D.4.1 D.4.4 D.4.5)..
5. Future perspectives
At this point, numerous questions remain open and more research will be essential in several areas to complement the CEREBRAD findings and answer society’s question about cognitive and cerebrovascular risks of LD-IR.

In regard to the animal studies, the following issues need to be addressed for a future approach in LD-IR research
➢ It is imperative to determine whether the effects elicited by LD-IR are detrimental. This includes anomalies in processes that are fundamental for a proper brain functioning and that might lead to pathology when deranged (e.g. mitochondrial respiration).
➢ Future studies involving models for neurodegenerative diseases will be indispensable to fully estimate risks of LD-IR during infancy in adulthood and to provide accurate directions to the public.
➢ The threshold model requires further validation. Addressing the mechanisms activating an adaptive response and its kinetic is extremely important for neuroprotection. An adaptive response, in fact, could be preventively elicited to mitigate undesired off-target effects of medial approaches involving the use of radiation or to protect individuals in the case of a radiation environmental emergency. Studies to identify the specific signaling molecules translating radiation-induced damage into a concerted response are therefore essential.
➢ Is hormesis really happening? Our data on mitochondrial function and neurogenesis suggest that this might indeed be the case, and focused
experiments to test conclusively this possibility are inescapable. It will also be important to determine the specific doses and conditions eliciting putative hormesis.
➢ CEREBRAD analyzed molecular and cellular changes up to 24 weeks after irradiation. The presence of alterations at this time point strongly suggest that LD-IR might influence natural ageing and lays foundation for studies in aged mice. However, it is still unclear whether LD-IR could promote senescence and, eventually, in which neuronal cell type.
➢ The molecular and cellular findings are in high correlation with the observed cognitive deficits in pre- and neonatally irradiated mice. In particular, the defective cortical development that was observed together with a disturbed hippocampal neurogenesis nicely links to the decreased thickness of the prefrontal cortex at the long term. This thus urges for more experiments investigating higher cognitive functions related to the prefrontal cortex in irradiated animals.
➢ Are the observed early radiation-induced effects the only contributors to the persistent cognitive decline and inflammation? To address this matter, a wider investigation of different brain regions is desired, preferably accompanied by the use of transgenic animals deficient for certain brain developmental factors or cytokines.
➢ Our systems biology results require experimental validation. Is the identified signature specific and sufficient to reveal LD-IR exposure? The bioinformatics platform should be classified as an infrastructure to be expanded to include more experimental evidence. Additionally, further investment would be necessary to fully develop the usefulness of this tool for the community involved in radiation biology research, and to make it sustainable in the frame of CONCERT.
➢ Finally, future research need to focus on combining multiple agents to be more in line with our real life. Additionally, extrapolation of this research to specific diets to investigate life style would emphasize other elements of our modern society that might contribute to radiation risk estimate.
In general, our results provide critical information for the EU citizens and highlight the importance of further experiments in the same line of research. Identifying neuropathological signs after LD-IR is particularly relevant when framed in the context of medical imaging, which relies on LD-IR for diagnostic purposes. Ensuring that such diagnostic procedures do not pose substantial neurological risks for the patients at later life stages therefore constitutes a fundamental public health issue. This issue becomes even more compelling when approached from the perspective of ageing, which has become an emergency in the EU and in the advanced economies in general. Indeed, in a society where life expectancy is steadily increasing, addressing the potential consequences of LD-IR exposure in young life on elders has a significant societal and
economic relevance. Moreover, we are obliged to inform the community about the additional risks of IR exposure when predisposed to other environmental toxicants, for example nicotine, which are often underestimated and which need our focused attention.
In conclusion, CEREBRAD generated solid evidence and paved the road for future studies in the field.
6. References

Buratovic, S., B. Stenerlow, A. Fredriksson, S. Sundell-Bergman, H. Viberg and P. Eriksson (2014). "Neonatal exposure to a moderate dose of ionizing radiation causes behavioural defects and altered levels of tau protein in mice." Neurotoxicology 45: 48-55.
Eriksson, P., C. Fischer and A. Fredriksson (2006). "Polybrominated diphenyl ethers, a group of brominated flame retardants, can interact with polychlorinated biphenyls in enhancing developmental neurobehavioral defects." Toxicol Sci 94(2): 302-309.
Eriksson, P., C. Fischer, B. Stenerlow, A. Fredriksson and S. Sundell-Bergman (2010). "Interaction of gamma-radiation and methyl mercury during a critical phase of neonatal brain development in mice exacerbates developmental neurobehavioural effects." Neurotoxicology 31(2): 223-229.
Fredriksson, A., M. Fredriksson and P. Eriksson (1993). "Neonatal exposure to paraquat or MPTP induces permanent changes in striatum dopamine and behavior in adult mice." Toxicol Appl Pharmacol 122(2): 258-264.
Liu, Y., S. Xiao, J. Liu, H. Zhou, Z. Liu, Y. Xin and W. Z. Suo (2010). "An experimental study of acute radiation-induced cognitive dysfunction in a young rat model." AJNR Am J Neuroradiol 31(2): 383-387.
Mao, X. W., J. O. Archambeau, L. Kubinova, S. Boyle, G. Petersen and R. Grove (2003). "Quantification of rat retinal growth and vascular population changes after single and split doses of proton irradiation: translational study using stereology methods." Radiat Res 160(1): 5-13.
Remler, M. P., W. H. Marcussen and J. Tiller-Borsich (1986). "The late effects of radiation on the blood brain barrier." Int J Radiat Oncol Biol Phys 12(11): 1965-1969.
Yoshii, Y. and T. L. Phillips (1982). "Late vascular effects of whole brain X-irradiation in the mouse." Acta Neurochir (Wien) 64(1-2): 87-102.


Potential Impact:
CEREBRAD was designed in a way to increase the knowledge regarding the cognitive and cerebrovascular effects following exposure to low doses of ionising radiation. CEREBRAD has successfully reached the defined objectives. The epidemiological data provide additional information for cognitive and cerebrovascular diseases, increasing thus the statistical power. We increase thus the data for cerebrovascular diseases based on case-control study, including 233 cases of strokes having occurred 5 years or more following a childhood cancer and 233 controls matched on cohort, gender, age and date of childhood cancer, and length of follow-up. We performed very detailed radiation dose estimation in any structure of the brain and in cerebral arteries. In a linear model, the Excess of Odds Ratio EOR of stroke, all types together, per Gy of average radiation dose to the cerebral arteries, was equal to EOR/Gy=0.49 (95%CI: 0.22 to 1.17). In a meta-analysis of several cohort studies, including the international nuclear workers study and the Hiroshima-Nagasaki cohort, the EOR/1Gy has been estimated to 0.27 (95%CI: 0.20 to 0.34) for stroke (Little, Tawn et al. 2010, Little, Azizova et al. 2012). Lastly, in Mayak workers, the EOR/1Gy for external radiation has recently been estimated as being 0.33 (95C%I: 0.19 to 0.50) (Simonetto, Schollnberger et al. 2015). Regarding Cognitive outcome, correlation of the neuropsychological test with specific brain regions and the absorbed doses were evaluated. A threshold dose in the range of 50 to 120 mGy was evidenced for the hemangioma cohort consisting of 115 subjects treated before the age of one year and receiving less than 1 Gy to the brain. Subjects exposed to ionizing radiation at adulthood as cleanup workers exhibit symptoms of mild cognitive impairment according to the operational criteria of the neurocognitive tests clearly detectable over a dose of 250 mGy. Our epidemiological data are highly consistent with previous findings, providing additional new data following exposure to radiation at different stages of brain development from embryonic to childhood and adult age. Besides, our animal studies revealed a linear dose-response curve in which severe neurocognitive impairments are significant only for doses higher than 0.5 Gy for both prenatal as early postnatal exposures. Nevertheless, subtle changes in higher cognitive functions related to prefrontal cortex behavior seem to be disturbed at 100 mGy. In addition combined exposure of ionizing radiation with environmental toxicants (paraquat, nicotine, PBDE...) showed also a shift in the threshold dose below 100 mGy. This indicates that there might be no threshold below which no effects are observed, especially for the behavioural effects. These investigations raise once more the question regarding confounding factors and their role in the uncertainty in the low dose region. To our knowledge, CEREBRAD is the first project providing new insight on co-exposure of radiation and environmental toxicants. Additionally, this research should be extended for multiple agents and to cover the interaction with life style especially in terms of nutrition and diets and their combined contribution with radiation to risk estimate.
The cellular and molecular investigations revealed obvious effects of low-dose ionizing radiation 'LD-IR' on the brain at multiple levels and opened prospective avenues for further exploration. In general, we could observe a clear dose-dependent effect and could unveil different anomalies induced by the lowest X-ray dose studied (0.1 Gy) in terms of cognition, cell death and neurogenesis. Of interest, next to the administered dose, also the age at which exposure occurs strongly influences radiation effects, which are mostly exacerbated when irradiation is induced at earlier stages. Yet, this is sometimes highly dependent on the brain region of interest.

Impact on optimizing protection and improving regulation
The CEREBRAD consortium is committed to improve the information on cognitive and cerebrovascular effects at low doses. Based on suitable cohorts and accurate dosimetry calculations as well as the extensive animal studies, we correctly address the shape of dose-response curve. In one hand our data is in line with previous studies, on the other hand we have increased the statistical power of epidemiology as originally planned to allow advanced mathematical modeling of cognitive and cerebrovascular diseases. The considerable effort devoted to the mechanistic studies allowed improving our knowledge about the early biological mechanisms inducing damage to the brain tissue and leading to occurrence of late cognitive and cerebrovascular outcomes. Taken together, the data generated in CEREBRAD will certainly contribute to improve the radiation protection system.
Impact on key issues in research
CEREBRAD was designed to answer key questions addressed by several reports of experts and recommendation bodies (HLEG, UNSCEAR, ICRP, and MELODI) in order to advance the know-how of radiation effects on the central nervous system:
• Key Question 1: Is the epidemiological data sufficiently robust in estimating the risk of cognitive and cerebrovascular effects below 500 mGy?
CEREBRAD has increased the statistical power of epidemiology for estimating the risk of cognitive and cerebrovascular effects below 500 mGy, providing thus new data to the community required for future mathematical modeling investigations to improve risk estimate. Interestingly, we provide additional data regarding cognitive deviation following exposure to radiation at different stages of brain development, prenatally (in utero) from Chernobyl, during childhood from medically exposed Hemangioma cohort exposed below the age of one year and receiving less than 1 Gy to the brain, and at adult age from Chernobyl cleanup workers. Our cognitive assessments are of high importance especially regarding the implications for the use of radiation in diagnostic and therapy at different ages. On the other hand, the cerebrovascular investigations on childhood cancer survivors will contribute to increase the data about cerebrovascular risk, leading thus to future robust mathematical models on cerebrovascular risk estimate. All together, the CEREBRAD epidemiological investigations have setup the methodologies and paved the way for accurate assessment of cognitive and cerebrovascular diseases especially in utero and at childhood.
• Key Question 2: Can the animal studies contribute in addressing the shape of the dose-response curve for cognitive and cerebrovascular effects at different stages of brain development as non-cancer detriment?
The extensive animal investigations in CEREBRAD are highly consistent with the human data in terms of cognitive effects. Behavioural defects were mostly observed when exposure occurred with a high dose (over 0.5 Gy) for both prenatal as postnatal exposures. Nevertheless, two tests in particular, revealed alterations in 0.1-Gy prenatally exposed mice. Indeed, these low-dose irradiated animals showed a defective sociability, as well as changes in higher cognitive functioning. It is important to note that these defects could be unveiled because of the use of an extremely sensitive behavioural test battery. Such sensitive protocols are necessary to uncover possible low-dose consequences, as also pointed out in the MELODI SRA. These animals were additionally subjected to 3D T2-weighted Magnetic Resonance Imaging 'MRI' to assess overall brain volume and volumes of specific brain structures. This showed a microcephalic (reduced head size) phenotype from a dose of 0.33 Gy on. Interstingly, microcephaly after prenatal exposure with such a relatively low dose has never been seen before and could also be addressed using this sensitive MRI screening method. Furthermore, no difference in relative volumes of particular brain structures could be observed for doses below 0.5 Gy, additionally indicating that low and high doses might not evoke similar mechanisms. On the other hand, coexposure experiments with combined irradiation and environmental toxicants showed lowering of the threshold dose below 0.1 Gy. Thus, we have been able for the first time to evaluate the effect of combined radiation and environmental toxicants exposures, which might have serious implications on risk evaluation.
CEREBRAD thoroughly investigated low-dose effects in mouse behaviour both in utero and at early postnatal age. Importantly, this indicates that there might be no threshold below which no effects are observed, especially for the behavioural effects.
• Key Question 3: How to reduce the uncertainties of dose reconstruction in the brain?
Innovative methodologies of dose reconstructions on medical cohorts have been used in CEREBRAD and allowed precise retrospective reconstruction of radiation doses in several brain structures and cerebral arteries. In such a way we could accurately relate effects with absorbed doses in brain structures allowing to correctly addressing the shape of the dose-response curve for cognitive and cerebrovascular disease. In addition dose calculations in mice were assessed using Monte Carlo codes simulation and allowed to quantify the energy deposition in soft tissue and bone for both external irradiation and internal contamination. The small size of the embryonic brain did not allow measuring the doses within brain structures. To our knowledge, CEREBRAD is providing unique data regarding dose estimation in mice enabling accurate dose-response curve which has never been considered in the past, although this field requires additional investigations in future projects within Horizon 2020.
• Key Question 4: What are the underlying biological mechanisms for such non-cancer effects of radiation?
Increased apoptosis and inflammatory response early after exposure appear to be the most contributing events for reduced brain size and late cognitive impairments. It is known that irradiation can cause apoptosis and necrosis and impaired neuronal migration. However, it is not clear to which extent these processes contribute to the late effects observed and, very importantly, it seems that defects differ depending on the timing of development at which irradiation takes place. In addition, we have shown that apoptosis in the irradiated embryonic neocortex at 24 h after exposure is restricted to differentiated, not proliferating cells, which is in contradiction to the generally accepted assumption that dividing cells are most radiosensitive. Besides, defects in adult neurogenesis and axonogenesis seem to be compromised several months after exposure. In addition, transcriptomic and proteomic analyses indicate possible contribution of epigenetic events in the processing of the late effects, requiring thus additional future investigations. Our research has definitely contributed to a better understanding of the underlying mechanisms indicating that mechanisms acting at low doses are not identical to those at high doses. Nevertheless, further research is necessary to accurately define the mechanisms in the very low dose region (below 0.1 Gy).

Impact on clinical practices
CEREBRAD is providing clear evidences on potential cognitive and cerebrovascular late health effects. In one hand, children exposed below one year of age (Hemangioma cohort) and who received radiation doses to the brain less than 1 Gy, on the other hand, the childhood cancer survivor study. Both studies are highly relevant for diagnostic and therapeutic exposures to evaluate the potential risks and benefits of radiation therapy and diagnostic practices in utero and during childhood. Further analysis will help to clarify the related aging consequences. In Europe, the elderly population is rising steadily in comparison to other age groups. This raises the question whether the increased use of medical radiation might influence the ageing process. It has been previously shown that adult exposure to a therapeutic dose of radiation causes premature senescence in the brain (Suman et al., 2013). Additionally, mice age prematurely due to the accumulation of DNA damage during the embryonic period (Murga et al., 2009), suggesting that ageing is influenced to some extent by embryonic distress, like irradiation (Fernandez-Capitello, 2010). However, the possible link between prenatal radiation exposure and brain ageing was never studied in detail before. With this part of our research, we are addressing a societal relevant question which has never been studied previously. Ageing is the main risk factor for neurodegenerative diseases and dementia and requires more attention in low-dose radiation research, especially in regard to cognitive and cerebrovascular defects.
Impact on radiation science research
CEREBRAD addressed an area of research in radiation biology that has been neglected for a long period resulting in a considerable lack of information and knowledge. Statements in the HLEG report made recommendations for maintenance of competence in Europe as key issue. The CEREBRAD project substantially increased the know-how in the area of cognitive and cerebrovascular effects. Although, concluding on any significant effect that could be informative for reviewing radiation protection regulation is probably too early. But at present stage, and after taking into account results from other ongoing studies in these fields, CEREBRAD pave the way to define clearly how a future program should be built up based on the findings and recommendations of the present program in order to move forward our knowledge about low dose radiation and related diseases. During this project, new never or very little explored fields have been explored and should be more deepened in the future, as they are of big importance for public health, particularly combined exposures (unavoidable in the real life!) and long term CNS effects as well as ageing. Future research need to focus on combining multiple agents to be more in line with our real life. Additionally, extrapolation of this research to specific diets to investigate life style would emphasize other elements of our modern society that might contribute to radiation risk estimate. Overall CEREBRAD has made important contributions in exploring the low dose region and in understanding the risks in utero and for children due to radiation exposure. Longer follow-up times research to understand outcomes of cerebrovascular and age-related neurological diseases due to low dose radiation exposures are highly recommended.
Finally, CEREBRAD promoted scientific transfer to the community via international pee-reviewed articles, presentations at international conferences and training junior scientists in radiation sciences. The wide expertise gathered by CEREBRAD partners will help to reach new research expectations and will open new possibilities for long-term interdisciplinary collaboration.




List of Websites:
http://www.cerebrad-fp7.eu/en