Reviewing the epidemiological literature related to CVD and low-dose ionizing radiation is complicated by the different classifications of CVD. Moreover, all types of CVD are often pooled in one diagnosis in epidemiological studies. This hampers the thorough understanding of radiation-related CVD risk, and distinction should be made between the different clinical manifestations. In addition, many epidemiological studies face the problem of misclassification of the cause of death, except for stroke, for which the diagnosis tends to be reasonably good (9). In fact, stroke is not considered to be a CVD, but a circulatory disease since it involves the blood circulation in the brain and is unrelated to the heart. It is defined by brain injury, which occurs when a blood vessel in the brain ruptures, leading to haemorrhage, or when a blood vessel is blocked, leading to ischemia following the loss of blood supply in the brain area of concern.

External beam RT for breast cancer, Hodgkin's lymphoma, or even peptic ulcer disease in the early days often involves some incidental exposure of the heart. There are studies pointing to late secondary cardiovascular effects due to this scattered radiation exposure. Long-term follow-up was shown to be essential, as the cardiovascular complications may manifest years after the completion of RT.

Most peptic ulcers are caused by an infection with a type of bacteria known as Helicobacter pylori and are nowadays treated, at least partly, with antibiotics. However, mid-last century peptic ulcer disease patients were irradiated. Peptic ulcer disease patients treated with RT (n=1859) or by other means (n=1860) at the University of Chicago Medical Center between 1936 and 1965, were followed through 1997 by Carr et al (8). The irradiated patients received volume-weighted cardiac doses ranging from 1.6 to 3.9 Gy and the portion of the heart directly in the radiation field received doses of 7.6–18.4 Gy. The observed numbers of cause-specific deaths were compared with the expected numbers from the general population rates. Greater than expected coronary heart disease-related mortality was observed among the irradiated patients. The excess coronary heart disease risk in patients who received RT for peptic ulcer disease decades before indicates the need for long-term follow-up of CVD after chest RT.

Over the last half century, RT has evolved to become one of the cornerstones of treatment for various types of cancer. It is estimated that >50% of patients with cancer are treated with radiotherapy. Along with the development of novel chemotherapeutic agents, RT has revolutionized the prognosis of patients with various types of cancer. Cancers during childhood and adolescence are now more and more successfully treated and these patients go on to live an active and normal adult life, as evident by an increasing number of cancer survivors. Late cardiovascular effects are often observed in cancer survivors (10). Hodgkin's lymphoma was the earliest paradigm for the study of radiation-induced vascular disease. A first case report was published in 1924 about histological changes in the human heart after irradiation for Hodgkin's lymphoma (11). Amongst Hodgkin's lymphoma patients who received radiation, CVD is of the most common causes of death. Studies have shown that these patients have an increased risk of coronary artery disease, valvular heart disease, congestive heart failure, pericardial disease and sudden death. The risk is particularly high in patients treated before the age of 40 years (12–15).

In addition, RT for breast cancer often involves some incidental exposure of the heart to ionizing radiation. Darby et al (16) conducted a population-based case-control study of major coronary events (i.e., myocardial infarction, coronary revascularization, or death from IHD) in 2,168 women who underwent radiotherapy for breast cancer between 1958 and 2001 in Sweden and Denmark. The overall average of the mean doses to the whole heart was 4.9 Gy (range, 0.03–27.72). The rates of major coronary events increased linearly with the mean dose to the heart by 7.4%/gray [95% confidence interval (CI), 2.9–14.5; P<0.001], with no apparent threshold. The increase began within the first 5 years after radiotherapy and continued into the third decade after radiotherapy.

Due to improvements in radiation techniques (e.g., breathing-adapted RT, CyberKnife), the risk of cardiovascular complications in relation to radiation are uncertain, but may be expected to decline. However, patients with classical risk factors, such as hypertension, smoking and hyperlipidaemia may be at an increased risk of radiation-related cardiovascular complications, and these risk factors should be treated aggressively (10). Younger patients should be screened, as this patient population at risk usually has a considerable life expectancy.

The most informative cohort is the Life Span Study (LSS), consisting of 120,321 exposed and non-exposed individuals selected from respondents to the national census of Japan in 1950 calling for survivors exposed to the bombings in Hiroshima and Nagasaki, and from residential surveys in the cities after the national census. Mortality in this population has been investigated since the 1950s by collecting information through the national population registry (koseki) and death certificates obtained throughout Japan. Cancer incidence data was available from population-based cancer registries since 1957 in Hiroshima and since 1958 in Nagasaki (17). Next to the availability of these data, the large size, the presence of both genders and all ages, and well-characterized individual dose estimates makes this cohort a valuable source for risk estimation. Another cohort, the Adult Health Study (AHS), was established in 1958 and consists of 19,961 subjects from the LSS cohort. These survivors underwent biennial health examinations, which provided additional clinical and sub-clinical information to the death and cancer registries data. In this way, disease morbidity for a variety of conditions can be investigated (18).

Preston et al evaluated non-cancer mortality based on the LSS report 13 published by the Radiation Effects Research Foundation (RERF), which covers the time period between 1950–1997 (19). In their study, the weighted colon doses from the DS86 dosimetry system were used for individual dose estimates. Only the period between 1968–1997 was included to account for the 'healthy survivor' selection effect. Individuals had to be alive in 1950 to enter the LSS cohort and have thus survived the difficult conditions after the bombing, which means that the health experience of this cohort may not be typical for a normal population. This is reflected as a decrease in non-cancer mortality during 1950–1960 in the LSS members that received doses below 2 Sv, as shown by Shimizu et al (20). This 'healthy survivor' selection effect had largely disappeared by the mid-1960s. To exclude this confounding effect, Preston et al advised to restrict the analyses to proximal survivors who were within 3 km of the hypocenter of the bombing, and to a follow-up period starting from 1968 (19). Based on the linear no-threshold model (LNT) model, excess relative risk (ERR) estimates were calculated to be 0.17 with 90% CI (0.08–0.26) for heart disease and 0.12 (90% CI, 0.02–0.22) for stroke, for the period between 1968–1997 (19).

Shimizu et al evaluated ERR of mortality from heart disease and stroke in the LSS cohort with a follow-up of 53 years (1950–2003) (21). For individual dose estimates, weighted colon doses (Gy) from the DS02 dosimetry system were used. In addition, the authors obtained, by a mail survey, information regarding sociodemographic (education, occupation type), lifestyle (smoking, alcohol intake) and health variables (obesity, diabetes mellitus) from 36,468 members of the LSS cohort. This allowed them to evaluate the effect of these confounding factors on ERR estimates. It should be noted that they included the full follow-up period from 1950–2003 and all survivors, thus not taking into account the 'healthy survivor' selection effect. For the detrimental health outcomes of heart diseases and stroke there is, however, no healthy survivor selection effect. This has recently been confirmed by Schöllnberger et al (23) following the comments of Little et al (22). Shimizu et al (21) found an ERR of 0.14 (95% CI, 0.06–0.23) for heart disease and an ERR of 0.09 (95% CI, 0.01–0.17) for stroke based on the LNT model. Whereas the LNT model fitted best the data for heart disease, the quadratic model was best to fit the data for stroke (Fig. 1). The latter model implies relatively little risk at lower doses. Indeed, the calculation of ERR for stroke over restricted dose ranges revealed an ERR of 0.03 (95% CI, −0.10–0.16) for 0–1 Gy and −0.07 (95% CI, −0.28–0.16) for 0–0.5 Gy. Furthermore, they showed that the association of dose with CVD risk in the LSS cohort is unlikely to be an artefact from confounding by sociodemographic, lifestyle or disease risk factors.

The above-mentioned studies have used the LSS cohort for CVD risk estimations. Takahashi et al examined the association with dose and the incidence of stroke in the AHS cohort (18). For their study, information of health examinations from the follow-up from 1980 onwards has been used, resulting in 9,515 AHS participants. For individual dose estimates, weighted colon doses (Gy) from the DS02 dosimetry system were used. In this study population, risk for haemorrhagic stroke was observed to increase with dose. This was across the full range of doses for men, while in women there seems to be a threshold of approximately 1.3 Gy.

Studies on radiation workers are of interest since they generally involve relatively low doses received over repeated exposures, although in some cases, accumulated doses may be high. Various studies have been performed, of which the most important will be discussed. The largest studied cohort consists of 275,000 nuclear industry workers from 15 countries, referred to as the 15-country study (24). The average cumulative dose received was 20.7 mSv. An overall increasing trend, although not significant, for circulatory disease mortality was observed. It was concluded that their findings are compatible with both no increased risk and with an increased risk comparable to that observed in A-bomb survivors. The Chernobyl liquidator cohort consisted of 61,017 individuals with an average cumulative dose of 0.109 Gy. An ERR/Gy of 0.41 (95% CI, 0.05–0.78) was found for IHD morbidity and 0.45 (95% CI, 0.11–0.80) for the morbidity of cerebrovascular diseases, though the outcomes were not adjusted for recognized risk factors such as excessive weight, hypercholesterolemia, smoking, alcohol consumption and others (25). A study by Muirhead et al revealed an increasing circulatory disease mortality risk with dose, which was borderline significant, in the UK National Registry of Radiation Workers in the industrial and medical field (26). The average cumulative dose received was 24.9 mSv. This finding should, however, be interpreted with caution due to the lack of information on confounding factors. Another large cohort consisted of 206,620 radiation workers in the industrial and medical field, registered in the National Dose Registry of Canada (27). The average exposure of all workers was 6.3 mSv, with large differences between males (10.6 mSv) and females (1.7 mSv). A significant increasing trend of circulatory disease mortality with dose was observed in males. Again, there is a lack of information on confounding factors and there is also incompleteness of dose records. Finally, a recent publication studied a 34% increase in stroke incidence after a survey during the years 1994–2008 of a cohort of 90,957 radiologic technologists who worked with fluoroscopically guided interventional procedures (28). In addition, mortality from stroke was also modestly elevated, although not statistically significant. No statistically significant excess risks of incidence or mortality were observed from any other cardiovascular disorders evaluated.

The Mayak cohort is of particular interest since it includes information both on mortality and morbidity, and information on confounding factors (29). In 1948, the first nuclear energy enterprise in Russia, Mayak Plutonium Association, became operational. Since 1948 the Mayak personnel undergo regular routine medical examinations. In addition, every 3–5 years a more detailed examination is carried out in a specialized hospital. This examination system led to a unique archive of medical data, which was used to create the 'Clinic' medical-dosimetric database. In addition, from a dosimetric point of view, the database is sound. Individual dosimetry for external gamma exposure was introduced at the beginning of 1948 and for internal exposure during the 1960s (29). Complete data are available for 12,585 Mayak workers employed during the years 1948–1958 and followed-up until December 2000. The mean cumulated external dose was 0.91±0.95 Gy (99% percentile 3.9 Gy) for men and 0.65±0.75 Gy (99% percentile 2.99 Gy) for women. In this cohort, a significant increasing trend in IHD morbidity was observed with the increasing total external dose [ERR/Gy=0.11 (95% CI, 0.049–0.168)]. The influence of confounding factors on this trend was minimal (30). A follow-up study involved the analysis of a cohort including 18,763 Mayak workers with an additional follow-up of 5 years (31). Overall, risk estimates for IHD were similar to the earlier study [ERR/Gy=0.10 (95% CI, 0.045–0.153)]. Remarkable though, a statistically significant decrease in IHD incidence was found among workers exposed to external doses of 0.2–0.5 Gy compared to workers exposed to external doses below 0.2 Gy. This decreased risk is heavily influenced by the observations in female workers. The authors further noted that this finding should be interpreted with caution since it has never been reported in other studies. The latter analysis was further updated and extended by looking at the lag-time to progression of IHD and by using the updated dosimetry system MWDS-2008 (32). In that study, it was observed that the main detrimental effects of external radiation exposure occurred after >30 years. In addition, a statistically significant risk was observed in men for mortality caused by IHD [ERR/Gy=0.09 (95% CI, 0.02–0.16)] while the risk was not significant for women. Recently, Azizova et al published a study regarding incidence and mortality from IHD in an extended cohort of 22,377 Mayak workers first employed during the years 1948–1982 and followed-up to the end of 2008 (33). Risk analysis demonstrated a significant increasing trend in IHD incidence, but not mortality, with total dose from external gamma-rays after having adjusted for non-radiation factors and dose from internal radiation. ERR/Gy for IHD incidence in males was 6-fold higher than in females. In addition, a significant increasing linear trend was observed in IHD mortality, but not incidence, with total absorbed dose from internal alpha radiation to the liver after having adjusted for non-radiation factors and doses from external gamma-rays.

In the same Mayak population, the incidence of and mortality from cerebrovascular disease was studied (34). The cohort consisted of 22,377 workers from the extended Mayak worker cohort that was followed-up to the end of 2008. In this study, the workers were exposed to a mean cumulated external dose of 0.54±0.76 Gy (95% percentile 2.21 Gy) for men and 0.44±0.65 Gy (95% percentile 1.87 Gy) for women. After correction for confounding factors, a significant increasing trend in cerebrovascular disease incidence was observed with increasing total external dose [ERR/Gy=0.46 (95% CI, 0.37–0.57)]. In addition, the authors showed that the cerebrovascular disease incidence was significantly higher in workers with a total external dose >0.1 Gy when compared to those exposed to lower doses. Restricting the analysis to a subcohort with negligible internal exposure for incidence of cerebrovascular disease supports a dose response sub-linear for low doses for incidence of cerebrovascular disease (35). In that study, the excess relative risk/dose was confirmed to be significantly higher for the incidence of cerebrovascular disease in comparison to cerebrovascular disease mortality and the incidence of stroke. The authors hypothesized that this difference was based on the complex nature of cerebrovascular diseases. The incidence was mainly related to chronic forms of cerebrovascular disease, while the mortality was mostly caused by the acute forms. Finally, having a young age during exposure was observed to be an important, aggravating modifier of radiation risk for incidence of cerebrovascular disease and stroke.

It should be noted that apart from the classical vascular-related confounding factors, occupational studies have to deal with the 'healthy worker' selection effect, similar to the 'healthy survivor' selection effect in A-bomb survivors. The 'healthy worker' selection effect occurs when workers who are healthier and have lower mortality and morbidity rates are selectively retained in the workplace, as such accumulating higher doses. One can adjust for this confounding factor by considering the duration of employment as a confounding factor in the analysis, as done in the 15-country study (24).

The Advisory Group on Ionizing Radiation (AGIR) from the Health Protection Agency reviewed the available epidemiological data for low- and moderate-dose exposure in 2010. Taking all the studies together, they reported a small, but statistically significant overall ERR/Gy of 0.09 (95% CI, 0.07–0.12). AGIR noticed, however, that there was a lot of heterogeneity in risk estimates of the different studies included in their meta-analysis (9). Little et al recently extended this meta-analysis (36). They estimated excess risks for four subgroups of circulatory disease, classified according to the 'International Classification of Diseases, 10th revision': IHD, non-IHD, cerebrovascular disease and all other circulatory diseases. A significant effect of heterogeneity between the different studies was found for cerebrovascular disease and other circulatory diseases, but not for IHD and non-IHD. ERR were calculated based on the LNT model, which implicitly assumes a linear association of CVD risk at low doses and dose rates. They noted that this assumption is reasonable since there is little evidence for non-linearity in the Japanese atomic bomb survivors and Mayak workers data. Furthermore, at least for IHD and non-IHD, the ERR/Sv was consistent between the Japanese atomic bomb survivors, Mayak workers and other occupational cohorts (36). Although it should be noted that Schöllnberger et al and others advocate for the consideration and testing of other dose-response models for non-cancer effects (32,37). To conclude, the overall consensus of the above-mentioned studies is that there is a significant elevated CVD risk for doses >0.5 Gy (9,38).

CVD is the leading cause of mortality and morbidity and accounts for 30–50% of all deaths in most developed countries. It is a multifactorial disease with many risk factors, such as lifestyle and other personal factors (39). The most established risk factors include the male gender, elevated low-density lipoprotein (LDL) levels, smoking, hypertension, a family history of premature coronary disease and diabetes mellitus (40). Epidemiological studies, as presented above, have limited statistical power to detect a possible excess risk of CVD following low-dose exposure (<0.5 Gy), due to the high background level of CVD in the population as a whole and many potentially confounding risk factors (9). For example, it has been calculated that, if the excess risk is in proportion to dose, a cohort of 5 million individuals would be needed to quantify the excess risk of a 10 mSv dose (41). Other factors that have an influence on epidemiological results are the distribution of the dose range, the accuracy of dosimetry, the duration of follow-up after exposure and correct assignment of cause of mortality, as reviewed by Borghini et al (42).

Although epidemiological studies have led to a better insight in radiation-related CVD risk, there are still many uncertainties that need to be clarified. These include whether there is a threshold dose; whether the latency of CVD development is dependent on the dose; the identification of the sensitive targets in the heart and vasculature; whether exposure has an impact on CVD incidence or progression, or both; and the exact impact of acute, fractionated or chronic exposure on risk estimates. For an accurate dose risk assessment, these questions need to be answered.

Classical epidemiological studies, as described above, do not provide all the needed insight to answer these questions. A more targeted approach, such as the integration of epidemiology and biology is required. For example, the assessment of subclinical endpoints and other cardiovascular biomarkers by functional imaging in patients receiving radiotherapy may provide insight into the development and progression of CVD following radiation exposure (39,42). Single-photon emission computed tomography (SPECT) or positron emission tomography (PET) imaging of micro-vascular perfusion has already been applied in breast cancer studies. The outcome differed between studies. For instance, whereas in one study perfusion defects were observed within 6–12 months after radiotherapy (43), no significant differences in perfusion defects were found in another study (44). In addition, the evaluation of cardiovascular biomarkers in patients receiving radiotherapy may be useful. For instance, elevated levels of N-terminal pro-B-type natriuretic peptide (NT-proBNP) in the blood (45,46) have been shown to be predictive for heart failure and/or CVD mortality across a broad range of individuals (47). Higher values of NT-proBNP were found in patients treated with radiotherapy for left-sided breast cancer compared to patients treated with other means (48).

Next to epidemiology, radiobiological research is essential for understanding CVD risk specifically in the low-dose region. Since epidemiological findings for low and moderate doses are suggestive and not persuasive, their use in dose risk assessment is limited. A thorough understanding of the biological and cellular mechanisms gathered through experimental studies is thus needed to complement the epidemiological findings. Once we have a comprehensive understanding of the underlying biological mechanisms, biologically-based dose-response models can be included in the dose risk assessment. This may prove to be beneficial for an accurate risk estimation in the low-dose region (49).

The possible excess risk of CVD following exposure to low doses is of great societal concern. According to the ICRP, a dose of 0.5 Sv may lead to approximately 1% of exposed individuals developing cardiovascular or cerebrovascular disease >10 years following exposure, in addition to the 30–50% suffering from disease without being exposed to ionizing radiation (50). Although the assumed risk is rather small, it may have serious implications for public health. Indeed, seeing the high background rate of CVD, the absolute number of excess cases would be substantial (42).

Various issues, such as occupational radiation exposure, the future of nuclear power, manned space flights and the threat of radiological terrorism, call for a thorough understanding of low-dose health risks (41). The main concern is, however, the increasing use of ionizing radiation for diagnostic medical purposes (Fig. 2) (51). For instance, since 1993, the number of computed tomography (CT) scans has quadrupled in the US and similar trends are observed in Europe (52).

In particular, the increased use of non-invasive cardiovascular imaging techniques, such as cardiac CT scans and myocardial perfusion imaging with radionuclides, are of importance (53). Indeed, effective doses range from 1 to 20 mSv depending on the procedure (Table I) (58). Although one cannot deny the huge health benefits of these improved diagnostic procedures, concerns are raised regarding the 'overuse' and potential associated health risks (54). For example, it has been observed that 14–22% of cardiac imaging tests are inappropriate in the US (55,56).

Radiation protection of patients is not based on dose limits, but on the principle of justification that states that the benefits and risks from the use of ionizing radiation should be carefully evaluated. However, this risk/benefit balance is highly patient-dependent and the decision for the use of a specific imaging test relies on the physician's judgment. Several guidelines have been published by various societies, such as the European Society of Cardiology and the American College of Cardiology Foundation, to aid in this decision (53,57). These guidelines provide information regarding the accuracy of the tests, the usefulness of the information obtained from the test, and also regarding the risks of the tests including those related to adverse radiation health effects. In addition, the implementation of informed consent, in which the possible risks are communicated to the patient and in view of his or her consent, will stimulate physicians to more carefully balance benefits and risks of a specific imaging procedure (54,58). Furthermore, the development and implementation of dose-lowering techniques may be beneficial not only for the patient (59), but also for the physician. In addition, the identification of biomarkers of susceptibility allows screening for sensitive patients, aiding in the evaluation of the risk/benefit balance (60).

The endothelium is a single layer of cells that lines the interior of the vascular system and has thus a strategic position between the blood and the surrounding tissues. Endothelial cells are involved in a wide range of physiological processes, such as the regulation of vascular tone, vascular permeability, blood coagulation/fibrinolysis and inflammation, which are required to maintain proper vascular functioning (Fig. 7) (103). Endothelial dysfunction has been observed in patients with atherosclerosis and in patients that exhibit CVD risk factors such as smoking, dyslipidaemia, obesity and diabetes mellitus (104), and is considered one of the first indicators of future cardiovascular morbidity and mortality (105–108). It should be noted that the endothelium is regarded as a critical target for radiation-induced CVD.

The endothelium displays phenotypical and functional heterogeneity depending on the vascular bed and tissue it is situated in (109,110). The major function of arteries and veins is the circulation of blood through the body. Capillaries, on the other hand, are the major exchange vessels and the capillary endothelium is thus very thin and usually fenestrated to ensure the optimal diffusion of oxygen and nutrients between the blood and underlying tissue (110). Arterial endothelial cells are sensitive to disrupted flow at branching points and curvatures in the arterial system, which are, consequently, 'hotspots' for inflammation, coagulation and atherosclerosis (110). A healthy arterial endothelium mediates vasodilatation and actively suppresses thrombosis, vascular inflammation and inhibits vascular smooth muscle cell proliferation and migration (111).

CVD risk factors associated with endothelial dysfunction include hypertension, smoking, dyslipidaemia and aging. A common underlying cellular mechanism leading to endothelial dysfunction is oxidative stress (112). Oxidative stress is defined as an imbalance between the generation of reactive oxygen species (ROS) (Fig. 8) and the activity of enzymatic and non-enzymatic antioxidant systems (113). At physiological levels, ROS are important signalling molecules; however, at higher concentrations, ROS cause cellular injury by inducing oxidative damage to DNA, lipids and proteins, and may result in cell death (114). Apart from cellular damage, oxidative stress leads to a decrease in nitric oxide bioavailability, premature senescence and mitochondrial dysfunction in endothelial cells.

The recognition of the endothelium as a central regulator of the cardiovascular system significantly enhanced endothelium-related research. One of the milestones was the first successful isolation and subsequent cultivation and characterization of endothelial cells in vitro, in the 1970s (109). Jaffe et al (115) and Gimbrone et al (116) reported independently the isolation of endothelial cells from human umbilical veins [human umbilical vein endothelial cells (HUVECs)]. Since then, many studies have relied on the use of HUVECs as they are relatively easy to obtain. Although originating from large vessels, HUVECs are unique since they exhibit endothelial properties that are intermediate between those of large vessels (e.g., the aorta) and those of the microvasculature (117). EA.hy926 cells, which are an immortalized derivate of HUVECs, are commonly used as well (118). However, it should be taken into account when using these cells as models for studying the endothelium radiation response, that there is a differential response to ionizing radiation in primary and immortalized endothelial cells. For example, EA.hy926 cells are more sensitive to the induction of apoptosis and double-strand breaks (DSB) in comparison to the same dose of ionizing radiation administered to HUVECs (119).

It should be noted that in vitro endothelial cell models are not limited to HUVECs and EA.hy926 cells. Endothelial cells may be isolated from other sources in the human body, such as the dermal microvasculature, coronary arteries and the pulmonary vasculature. However, these are usually not easy to obtain and at best only a small amount of material is available. In addition, these primary cell cultures cannot be kept in long-term cultures as they begin to lose their endothelial cell characteristics. Therefore, immortalized derivatives of these primary cells have been established; for example, EA.hy926 cells are a derivate of HUVECs. Immortalized cells are characterized by their ability to overcome senescence and can be kept in culture for long periods of time (120). These immortalized derivatives, however, also exhibit tumour cell characteristics (121). For instance, immortalized human coronary artery endothelial cells [transfected with human telomerase reverse transcriptase (hTERT)] showed 40% of aneuploidy at a low passage number and 100% of aneuploidy at a high passage number (121).

Of course, in vitro endothelial models, although useful, are not fully representative for the in vivo situation. Yet, advances have been made, such as the development of co-culture models where endothelial cells are cultured with vascular smooth muscle cells to study the atherosclerotic process (122,123). In addition, 3-D cultures, where endothelial cells are grown in a matrix allowing the formation of tubule-like structures are increasingly used to study angiogenesis (124,125). A recent review described in detail the use of these co-culture and 3-D culture models in radiation research (126).

Gaining insight into the endothelial response to ionizing radiation exposure is not only of importance for understanding the development of radiation-related CVD, but also for optimizing cancer treatment. For instance, adverse reactions in the surrounding healthy tissue of the tumour are related to the radiation-response of the microvasculature in the tissue of concern. In addition, tumour growth is highly dependent on an abundant blood supply, which is maintained by a rich vascular network (127). The tumour vasculature thus became a potential target in radiotherapy (128). Understanding the effects of radiotherapy on the tumour endothelium can improve treatment regimes. Since the tumour endothelium differs in many aspects to the normal endothelium (129), this is not in the scope of this review.

Below, an overview is given of classical cellular radiation effects and how these may affect endothelial functioning. Next, the impact of ionizing radiation on mitochondrial function and on senescence is discussed. Finally, attention is paid to the contribution of new technologies, such as high throughput transcriptomic profiling, allowing a better understanding of the underlying molecular signalling pathways: