Quantification of DNA Double Strand Breaks and Oxidation Response in Children and Adults Undergoing Dental CBCT Scan

Assessing the possible biological effects of exposure to low doses of ionizing radiation (IR) is one of the prime challenges in radiation protection, especially in medical imaging. Today, radiobiological data on cone beam CT (CBCT) related biological effects are scarce. In children and adults, the induction of DNA double strand breaks (DSBs) in buccal mucosa cells and 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dG) and antioxidant capacity in saliva samples after CBCT examination were examined. No DNA DSBs induction was observed in children nor adults. In children only, an increase in 8-oxo-dG levels was observed 30 minutes after CBCT. At the same time an increase in antioxidant capacity was observed in children, whereas a decrease was observed in adults. Our data indicate that children and adults react differently to IR doses associated with CBCT. Fully understanding these differences could lead to an optimal use of CBCT in different age categories as well as improved radiation protection guidelines.

Power analysis. Power analysis was based on validations experiments that were performed prior to this study 55 . The number of participants (N) are the numbers described in this manuscript. The results of the power analysis indicate that the sample size was sufficient for all analyses, i.e. equal to or greater than 0.9 (Supplementary Table 1).

DNA double strand break detection in exfoliated buccal mucosal cells before and after CBCT examination.
The results from co-localized γH2AX and 53BP1 foci, which are a measure for DNA DSBs, show no changes in the amount of DSBs after CBCT examination, neither in children nor adults (Fig. 1).
Contrary to the children, the number of foci per cell remained increased 24 hours after CBCT when compared to before CBCT (0.0061 ± 0.0051 foci/cell; p > 0.9999). Between 30 minutes after CBCT and 24 hours after CBCT no significant difference was observed (p > 0.9999).
Interestingly, the amount of foci per cell was significantly higher in children than in adults at every time point. Before CBCT 0.25 ± 0.054 foci/cell were observed in children and 0.0014 ± 0.0014 foci/cell were observed in adults (Before CBCT: Mann-Whitney U value = 121, p = 0.0020; 30 minutes after CBCT: Mann-Whitney U value = 145, p = 0.0146; and 24 hours after CBCT: Mann-Whitney U value = 170, p = 0.0487).
8-oxo-dG levels in saliva samples. 8-oxo-dG levels were measured in saliva samples collected before and after CBCT examination. They were increased in children but not in adults 30 minutes after CBCT (Fig. 2).
In the group of children, data were split based on gender (Table 1). Both in boys and girls the amount of 8-oxo-dG increased significantly after CBCT examination (N = 35, p = 0.024 and N = 33, t-value = 2.91, DF = 32, p = 0.0065, respectively). Furthermore, no differences between boys and girls was observed (Table 1). This was  Excretion of 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dG) into saliva is increased after cone beam computed tomography (CBCT) examination in children but not in adults. Only data from patients of which results were obtained for both time points were included. In children there is a significant average increase of 121% in 8-oxo-dG excretion 30 minutes after CBCT examination (N = 68, DF = 67, t value = 4, p < 0.0001). In adults there is an average increase in 8-oxo-dG excretion of 59% (N = 19, DF = 18, t value = 1.58, p = 0.1317). Green dotted line = average; ***p < 0.0001. confirmed when the proportional change between values before and after CBCT were compared between boys and girls (p = 0.6907) (see Supplementary Data 2).
Plotting the proportional change in 8-oxo-dG levels of children against the absorbed dose received by the patients showed no visible trend or dose response (Fig. 3).
Total antioxidant capacity in saliva samples. Ferric Reducing Antioxidant Power (FRAP) values were measured in saliva samples before and 30 minutes after CBCT examination. They were significantly increased in children and decreased significantly in adults 30 minutes after CBCT examination (Fig. 4).
Children showed a slight, but significant increase in FRAP value after CBCT examination. Thirty minutes after CBCT examination, FRAP values increased from 260. 80 Table 1. Comparison between boys and girls for 8-oxo-dG excretion before and after cone beam computed tomography (CBCT) examination. *For inter-group testing a paired student T-test was performed; for intragroup testing: an unpaired student T-test was performed.   Results were also analysed based on gender ( Table 2). In children, both boys and girls showed an increase in FRAP values, but the increase was only significant in girls (N = 62, t-value = 0.81, DF = 61, p = 0.4194 and N = 55, t-value = 2.28, DF = 54, p = 0.0268, respectively). Additionally, in both adult men and women a decrease was observed, but this was also only significant for women (N = 4, Wilcoxon test, p > 0.9999 and N = 13, t-value = 2.27, DF = 12, p = 0.0428, respectively). Furthermore, in children it was observed that the baseline levels were lower in the morning (225.10 ± 12.48) than baseline levels in the afternoon (282.30 ± 21.04) (Welch-corrected t-value = 2.34, DF = 82.42, p = 0.0217). The same was observed in adults (baseline morning: 174 ± 21; baseline afternoon: 269 ± 42), although this difference was not statistically significant (Mann-Whitney U value = 12, p = 0.0897). Therefore, the data from children were split into a morning and afternoon group. The salivary FRAP values did not significantly differ after CBCT examination if data were corrected for time of sample collection. In the morning groups, there was no significant change in both boys and girls (N = 24, Wilcoxon test, p = 0.97 and N = 10, t-value = 0.81, DF = 9, p = 0.7394, respectively). In the afternoon group, FRAP levels in boys did not change (N = 17, Wilcoxon test, p = 0.89). However, in girls from the afternoon group FRAP levels increased significantly (N = 24, t-value = 2.14, DF = 23, p = 0.0431).

Discussion
Determining the biological effects of exposure to low doses of IR, such as those used in medical imaging, is of paramount concern in radiation protection today. This study aimed to characterize the short-term radiation-induced effects associated with CBCT examinations, specifically in children. To this end, the number of DNA DSBs was monitored in BMCs and 8-oxo-dG levels as well as total antioxidant capacity were monitored in saliva samples using previously optimized protocols 55 .
Exposure to IR can result in DSBs, which are considered very harmful, since inaccurate repair could result in mutations, chromosome rearrangements, chromosome aberrations and loss of genetic information 31,32,35,36 . Our results indicate that exposure to radiation doses used in CBCT examinations (0.184 mGy -9.008 mGy in this study) does not induce DNA DSBs in BMCs from children and adults, as observed using a microscopic γH2AX/53BP1 co-localization assay. Previously, both the γH2AX assay and the γH2AX/53BP1 assay were used to detect DNA DSBs after exposure to radiation doses used in diagnostic and interventional radiology, such as CT scans [58][59][60] . These studies report a significant increase in γH2AX foci in lymphocytes 1 hour after CT examination, which uses higher radiation doses than CBCT. Furthermore, our group recently showed that low doses associated with CBCT examinations are capable of inducing DNA DSBs in vitro in dental stem cells 61 . BMCs have also been used successfully as a biomarker for genotoxic effects, including using the γH2AX assay to detect radiation-induced DNA DSBs 48,62,63 . These studies report increase of genotoxic effects in BMCs after low dose IR exposure. Gonzalez et al. (2010) showed that in vitro exposure of BMCs to IR induces γH2AX foci 48 . Our findings are in line with previous publications focusing on genotoxicity induced by radiological examinations. In these studies, no genotoxic effects, i.e. micronucleated cells, were observed after low doses of IR, such as panoramic dental radiology and CBCT. These studies, however, all reported increases in other nuclear alterations (e.g. pyknosis, karyorrhexis and karyolysis) that are associated with increased cytotoxicity [63][64][65][66]  Our data show 0.0014 ± 0.0014 co-localized γH2AX/53BP1 foci per cell in BMCs from adults at baseline. This number is remarkably lower than the 0.08 ± 0.02 γH2AX foci per cell in non-irradiated BMCs reported previously by Gonzalez et al. (2010) 48 . These different observations can be explained by the higher sensitivity of the γH2AX/53BP1 co-staining, which eliminates the detection of γH2AX foci observed during S-phase replication fork stalling 68 . In addition, Gonzalez et al. (2010) treated the BMCs differently, e.g. after collection they incubated the BMCs in cell growth medium at 37° Celsius, which can also affect the number of foci counted 48 .
Interestingly, we found that before CBCT examination, but also 30 minutes and 24 hours after CBCT examination, the average number of γH2AX/53BP1 foci per cell was higher in children than in adults. This observation  www.nature.com/scientificreports www.nature.com/scientificreports/ contradicts what has been published before, namely that aging is associated with accumulation of DNA damage 69,70 . One would expect the level of DNA damage, at least before CBCT examination, to be higher in adults than in children. However, BMCs are the first barrier in the inhalation and ingestion routes. Therefore, they are exposed to several genotoxins. These can be found in environmental and lifestyle factors such as diet, mouthwash, smoke, air pollution, etc. [71][72][73] . These factors can, at least partially, explain our observation, since children are more sensitive to these type of genotoxins compared to adults due to age-related differences in absorption, metabolism, development and body functions 72 .
Finally, we observed that the response after CBCT examination in children did not differ significantly from that of adults. This indicates that BMCs from children after CBCT examination do not show an increased radiosensitivity compared to BMCs from adults [25][26][27] . These findings are in line with results from Ribeiro et al. (2008). They compared the genotoxic and cytotoxic effects of dental radiography between children and adults and found no significant differences in micronucleus frequency or cytotoxicity 74 . However, the radiation doses used in radiography are lower than those used in CBCT, thus this should be interpreted with caution.
This study shows that 8-oxo-dG levels excreted in saliva increased in children but not in adults 30 minutes after CBCT. Because of its mutagenic potential, excretion of 8-oxo-dG depends on cellular DNA repair mechanisms, such as nucleotide excision repair, nucleotide incision repair and Nudix hydrolase activity 75 . Therefore, a reduced DNA repair capacity may result in accumulation of 8-oxo-dG in the cells, thus resulting in a decrease in 8-oxo-dG excretion. Since DNA repair capacity was shown to decrease with age, this could explain why the concentration of 8-oxo-dG in saliva samples of adults was not increased significantly after CBCT examination, as it was in children 76,77 . Despite the significant increase in children and the limited increase in adults, no statistical differences were observed between both groups. This is most likely due to the limited group size of the adult group.
Previously, an association between the excretion of 8-oxo-dG and high radiation doses was described 78 . This association was not linear and showed saturation between 0.5 and 1 Gy. However, such dependency was not observed in this study, for example children that were exposed to 0.8 mGy showed a similar increase in 8-oxo-dG excretion as children exposed to 0.2 mGy. These data indicate that there is a high variability in individual radiosensitivity in our study population. Alternatively, it could be that the very low IR doses associated with CBCT elicit a small biological response which is unrelated to the IR dose, like an all-or-nothing mechanism. This is similar to the use of a 'priming dose' in adaptive response studies. Here a very low dose of a stressor (e.g. a chemical or IR) results in a small response which in turn prepares cells to an exposure of the same stressor at a higher dose 79 . Our results mimic the effects seen when applying such a 'priming dose' .
Although 8-oxo-dG was proposed as a marker for radiosensitivity, evidence is lacking or comes from radiotherapy patients, who receive doses that are a lot higher than the doses in our study population 80 .
We describe for the first time that salivary 8-oxo-dG levels are significantly increased in both boys and girls after CBCT examination. No significant gender differences in salivary 8-oxo-dG levels were observed. Previous measurements in urine and other cells showed similar results [81][82][83] . To the best of our knowledge, similar findings of 8-oxo-dG secretion in saliva in children were not reported before. Previous studies analysed oxidative stress markers in adults. These studies reported higher ROS production and oxidative stress biomarkers in men when compared to premenopausal women (reviewed by Kander et.al. 84 ). It is noteworthy that these studies are all related to cardiovascular diseases and not radiation exposure. However, there are studies that report higher oxidative status in females which contradicts the aforementioned studies 85 .
Finally, the authors want to state that these results should be interpreted cautiously, since multiple sources of 8-oxo-dG exist. Guanosine bases in the nucleotide pool can also be oxidized and detected in saliva 75,78,86,87 . Therefore, the 8-oxo-dG that was detected can also originate from the nucleotide pool, rather than from the DNA.
FRAP values give information about the total antioxidant capacity of biological samples. Our data shows on opposite response between children and adults 30 minutes after CBCT examination: salivary FRAP values increase significantly in children, whilst they decrease significantly in adults. Furthermore, the response in children is significantly different from that in adults, indicating that children react differently to CBCT-associated radiation exposure. Interpretation of the data needs to be done cautiously, since the data show that the time of sampling (in the morning or in the afternoon) significantly affected the baseline salivary FRAP values in children.
The highest values were measured in the afternoon. Similar circadian changes in FRAP values were observed before 88 . After correcting for time of sampling, no significant changes in salivary FRAP levels were observed, except for girls that were sampled in the afternoon. However, since pair-wise tests were used, this circadian influence is expected to be limited in this study.
Total antioxidant capacity has been used previously as a salivary biomarker related to periodontal disease and dental caries. Decreases in total antioxidant capacity have been linked to periodontal disease 89 .
The use of total antioxidant capacity as a biomarker has several limitations. Firstly, the total antioxidant capacity that is measured is the result of a complex mixture of antioxidants that is present in saliva. The major antioxidant in saliva has been reported to be uric acid, which accounts for more than 85% of the salivary antioxidant capacity. In addition, a wide array of other potent antioxidants are found in saliva, such as superoxide dismutase, catalase, glutathione peroxidase, ascorbic acid, several vitamins and albumin 90,91 . In this regard, future analysis into the enzymatic activity of specific antioxidant enzymes, e.g. superoxide dismutase might be interesting. Secondly, a lot of biological variability of salivary total antioxidant capacity exists. We report an average salivary FRAP value of 202.90 ± 21.28 in adults at baseline, whereas an average of 610.83 ± 4.52 was reported before in healthy adults 92 . It is noteworthy that this patient population was Asian, where ours is European, which may suggest ethnical differences in salivary FRAP values. Finally, several confounding factors have been described that affect the saliva composition and can thus affect the total antioxidant capacity. Confounding factors may include circadian rhythm, gender, age and diet 88,90 . This study also found an effect of circadian rhythm (see above), age and gender. Girls show a significant increase in salivary FRAP values, whereas women show a significant decrease. www.nature.com/scientificreports www.nature.com/scientificreports/ Both boys and men showed a change (an increase and decrease, respectively), but this was not significant. These findings indicate that females are more susceptible to changes in total antioxidant capacity following IR exposure and that the net effects depends on the age of the individual. However, it is important to note that our patient group is relatively small (N = 72 for girls and N = 13 for women). Increasing the sample size could therefore yield different results. These limitations could interfere with interpretation of the results. Therefore, it is important to take these confounding factors into account during the design of a study. As with 8-oxo-dG, no dose response relationship was observed for FRAP values.
In conclusion, our data provide evidence that CBCT examinations causes changes in the oxidation response in children. In adults, a slight increase in 8-oxo-dG levels and a significant decrease in the antioxidant response were observed. Despite this increase in oxidation response, no induction of DNA DSBs in BMCs was observed in children nor in adults.
Since no DNA DSBs are observed, the changes in 8-oxo-dG and FRAP levels can also be explained by an adaptive response 4,93-95 . Since 8-oxo-dG excretion and antioxidant capacity both increase in children, it could indicate that the intracellular defence mechanisms are being 'primed' , i.e. they are being prepared for a subsequent exposure to IR. Indeed, since no DNA DSBs are observed, the DNA repair mechanisms are working properly, if DNA damage was ever induced at all, and the increase in 8-oxo-dG excretion and antioxidant capacity could indicate an increase in the antioxidant defence system. However, these results are not observed in adults. Therefore, additional research into the oxidation response following CBCT examinations is required. Besides age-related differences, we observed some gender-related differences. Girls/women showed a significant increase/decrease in FRAP values after CBCT examination, whereas boys/men do not. Our data also demonstrate that saliva can be used for biomonitoring after IR exposure even if the radiation doses are very low (<1 mGy). However, no dose response relationship was found, neither for 8-oxo-dG levels nor for FRAP values.
Nonetheless, these results should raise awareness about radiation protection and the 'As-Low-as-Diagnostically Acceptable being indication-oriented and patient-specific' (ALADAIP) principle among clinicians and radiologists 9 . However, this should be investigated into more depth to gather more information about the potential link between possible biological effects and the CBCT settings that were used. Furthermore, the effects observed and described in this study are short-term effects, i.e. within 30 minutes after CBCT examination. We can conclude that molecular effects, although very small, occur and that further research is warranted. These findings are an incentive for continuing research into the short-and long-term biological effects after CBCT examination, especially the antioxidant response, since fully understanding them could lead to an optimal use of CBCT in a paediatric population as well as improved radiation protection guidelines.

Materials and Methods
EU OPERRA -DIMITRA study. The DIMITRA study is an non-interventional, prospective study that focusses on radiation-induced effects related to diagnostic CBCT exposure in children. It is a multicentre study carried out in three European centres: the Oral and MaxilloFacial Surgery -Imaging & Pathology department (Katholieke Universiteit Leuven, Leuven, Belgium), the Dental Medicine Department of the Bretonneau Hospital (Paris, France) and the Iuliu Hatieganu University of Medicine and Pharmacy (Cluj-Napoca, Romania) 55 . All experiments and methods were performed in accordance with relevant guidelines and regulations. All experimental protocols were approved by a named institutional/licencing committee. Ethical approval was obtained at the participating sites (Commissie Medische Ethiek KU Leuven, B322201525196, Belgium; Comité d'Evaluation de l'Ethique des projets de Reserche Biomédicale Paris Nord, N°16-021, France; Comisia de Etica UMF Iuliu Hatieganu Cluj-Napoca, 208/21.04.2015, Romania). In case of underage children, both parents needed to consent unless one parent has explicit permission from the other parent 55 .
Patient selection. Patients with various indications were referred to the clinic for CBCT examination. They were examined using CBCT device settings that match their individual needs. Thus the FOV, kV, mAs and resolution mode are adjusted to fit with each individual's indication and age, in agreement with the ALADAIP principle, as described in the DIMITRA position statement by Oenning et al. 9 . Throughout the three participating centres, three CBCT devices were used: Accuitomo 170 (Mortia, Osaka, Japan), NewTom VGi evo (Cefla S.C., Imola, Italy) and Promax 3D (Planmeca OY, Helsinki, Finland).
Eligible patients were children/adolescents from 3 to 18 years old, as well as adults (>18 years old), with good oral hygiene. Exclusion criteria were the presence of systemic diseases, the use of antibiotics or anti-inflammatory drugs, smoking and not giving informed consent prior to enrolment. Informed consent. Informed consent was obtained for each participating patient. In case of underage children, informed consent from a parent and/or legal guardian was obtained prior to inclusion in the study.
Buccal mucosal cell collection and immunocytological staining. The collection and staining method were described in detail by Belmans et al. 55 . Briefly, synthetic swabs were used to collect BMCs just before, 30 minutes and 24 hours after CBCT examination using a protocol modified from Thomas et al. 45 . Before each swabbing the patientrinsed his/her mouth twice with water. The swabs were put in Saccomanno's fixative (50% ethanol and 2% polyethylene glycol in milliQ water) and stored at 4 °C. Next, the BMCs were centrifuged at 580 g for 10 minutes. Then they were washed three times in buccal buffer ( www.nature.com/scientificreports www.nature.com/scientificreports/ on coverslips by cytocentrifugation (ThermoFisher, Waltham, MA, USA). The coverslips were placed in 4-well culture plates (Nunc, ThermoFisher, Roskilde, Denmark) so that the BMCs were facing up.
Finally, images were acquired with a Nikon Eclipse Ti fluorescence microscope using a 40x dry objective (Nikon, Tokyo, Japan). Images were analysed with open source Fiji software 96 , which analyses each nucleus based on the DAPI signal and within each nucleus the signals from Alexa Fluor ® 488 and −568 represent the γH2AX and 53BP1 foci, respectively. The number of co-localized foci per nuclei were determined using the Cellblocks toolbox 97 .
Saliva collection. The collection of saliva samples was described in detail by Belmans et al. 55 . In summary, saliva samples were collected right before and 30 minutes after CBCT examination using the passive drool method 98 , and sampling coincided with the BMC collection. Immediately after collection, the whole saliva was stored at −20 °C until shipment. After shipment to the lab, saliva samples were centrifuged at 10,000 g at 4 °C and the supernatant was stored at −80 °C until further analysis.
8-oxo-dG enzyme-linked immunosorbent assay. 8-oxo-dG was analysed using a 8-oxo-dG enzyme-linked immunosorbent assay (ELISA). Prior to this assay, 500 µl of saliva was purified twice on a C18 solid phase extraction column (Varian, Lake Forest, CA, USA) as described by Shakeri Manesh et al. 99 . The 8-oxo-dG ELISA (Health Biomarkers Sweden AB, Stockholm, Sweden) was performed as described by Haghdoost et al. 78 . In short, 270 µl of sample/standard was added to 165 µl of primary antibody and incubated for 2 hours at 37 °C on a shaker. The ELISA plate was washed with 1x PBS and 140 µl of sample/standard was loaded per well. The plate was incubated overnight at 4 °C on a shaker. Next, the plate was washed with 1x washing solution and 140 µl of secondary antibody was added per well. After a 2 hour incubation at RT, the plate was washed with 1x washing solution. Afterwards, 140 µl of chromogenic substrate 3,3′,5,5′-tetramethylbenzidine (One-step substrate system, Dako, Glostrup Municipality, Denmark) was added and the plate was incubated for 15 minutes at RT. The colour reaction was stopped by adding 2 M sulphuric acid. Finally, the absorbance was measured at 450 nm (signal) and 570 nm (background) using a microplate reader (ClarioStar, BMG Labtech, Ortenberg, Germany). 8-oxo-dG levels were interpolated based on a standard curve (range: 0.02-10 ng 8-oxo-dG/ml).
Total antioxidant capacity determination. The Ferric Reducing Antioxidant Power (FRAP) assay (Cell Biolabs, CA, USA) was performed on whole saliva according to the manufacturer's instructions. Briefly, 100 µl of sample/standard and 100 µl reaction reagent were added per well of a 96-well plate. Then the plate was incubated for 10 minutes at RT on a shaker. Finally, the absorbance was measured at 560 nm using a microplate reader (ClarioStar, BMG Labtech, Ortenberg, Germany).
Statistics. Statistical analysis was performed using GraphPad 7.02 (GraphPad Inc., CA, USA). The results of the DNA DSBs in BMCs were analysed using repeated measures one-way analysis of variance (ANOVA). 8-oxo-dG and FRAP assay results were analysed using two-tailed paired t-tests. To analyse differences between age groups and differences in radiation sensitivity, two-tailed unpaired t-tests were performed. While all tests listed above are parametric tests, non-parametric alternatives were used if conditions were not met. P values lower than 0.05 were considered as statistically significant. Results are shown as mean ± standard error of the mean (SEM).

Power analysis.
Power analysis was performed in R 100 . Input values for expected differences and standard deviations, values were taken from a previously published validation study 55 . The power for the t-test with significance level 0.05 was calculated with the R-package 'pwr' 101 . For one-way ANOVA, the power was calculated with a significance level of 0.05 using the R-package wp.rmanova 102 . www.nature.com/scientificreports www.nature.com/scientificreports/