Authors: James W. S. Jahng (1Stanford Cardiovascular Institute, Stanford University School of Medicine, Stanford, CA, USA.; 2Division of Cardiovascular Medicine, Department of Medicine, Stanford University School of Medicine, Stanford, CA, USA.), Mark P. Little (3Radiation Epidemiology Branch, National Cancer Institute, Bethesda, MD, USA.; 4Faculty of Health and Life Sciences, Oxford Brookes University, Headington Campus, Oxford, UK.), Hyunsoo J. No (5Department of Radiation Oncology, Southern California Permanente Medical Group, Los Angeles, CA, USA.; 6Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, USA.), Billy W. Loo, Jr (6Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, USA.; 7Stanford Cancer Institute, Stanford University School of Medicine, Stanford, CA, USA.), Joseph C. Wu (1Stanford Cardiovascular Institute, Stanford University School of Medicine, Stanford, CA, USA.; 2Division of Cardiovascular Medicine, Department of Medicine, Stanford University School of Medicine, Stanford, CA, USA.; 8Department of Radiology, Stanford University School of Medicine, Stanford, CA, USA.; 9Greenstone Biosciences, Palo Alto, CA, USA.)
Categories: Article
Source: Nature reviews. Cardiology
Authors: James W. S. Jahng, Mark P. Little, Hyunsoo J. No, Billy W. Loo, Joseph C. Wu
Ionizing radiation is widely used in various industrial and medical applications, resulting in increased exposure for certain populations. Lessons from radiation accidents and occupational exposure have highlighted the cardiovascular and cerebrovascular risks associated with radiation exposure. In addition, radiation therapy for cancer has been linked to numerous cardiovascular complications, depending on the distribution of the dose by volume in the heart and other relevant target tissues in the circulatory system. The manifestation of symptoms is influenced by numerous factors, and distinct cardiac complications have previously been observed in different groups of patients with cancer undergoing radiation therapy. However, in contemporary radiation therapy, advances in treatment planning with conformal radiation delivery have markedly reduced the mean heart dose and volume of exposure, and these variables are therefore no longer sole surrogates for predicting the risk of specific types of heart disease. Nevertheless, certain cardiac substructures remain vulnerable to radiation exposure, necessitating close monitoring. In this Review, we provide a comprehensive overview of the consequences of radiation exposure on the cardiovascular system, drawing insights from various cohorts exposed to uniform, whole-body radiation or to partial-body irradiation, and identify potential risk modifiers in the development of radiation-associated cardiovascular disease.
Ionizing radiation contains quanta with enough energy to displace electrons and break chemical bonds and includes electromagnetic radiation, such as X-rays and γ-rays, or high-energy α-particles and β-particles emitted from nuclear decay. Ionizing radiation has been used in various industrial and medical practices, with X-rays in particular having revolutionized the diagnosis and treatment of localized cancer^1^. When traversing tissue, ionizing radiation either directly interacts with DNA molecules by displacing electrons, which leads to ionization, or acts indirectly by transferring part of its energy to one of the electrons in a molecule that it traverses, which in turn ionizes the surrounding molecules, such as DNA. Given the high water content in cells, this ionization induces the production of reactive oxygen species or reactive nitrogen species, triggering various biological effects that result in DNA damage, protein or lipid modifications, and damage to the organelles^2^ (Fig. 1a). Incomplete clearance of radiation-damaged cells can lead to adverse cellular responses, including apoptosis and other non-apoptotic forms of cell death (such as necrosis or necroptosis, autophagy-dependent cell death and ferroptosis) and cell inactivation (including mitotic catastrophe and senescence)^3^. The overall effect of radiation damage on cell survival is not always linearly associated with the absorbed dose, and heterogeneous radiosensitivities across different cell types have been reported (Box 1).
Determining the health implications of ionizing radiation at the organ level is challenging. Historically, the biological effects of ionizing radiation were classified as either stochastic or tissue reaction effects (previously known as deterministic effects)^4,5^. Stochastic effects include cancer or heritable effects owing to acquired mutations in somatic or reproductive cells^5^. Tissue reactions encompass various degenerative or adverse tissue remodelling effects occurring after radiation exposure, which generally transpire only above a certain threshold dose, and with the severity of the effect increasing with dose (Fig. 1b). Tissue reactions were thought to be primarily caused by the depletion of functional parenchymal cells targeted by ionizing radiation, which is true in highly proliferative tissues, such as those in the haematopoietic, spermatogenic and digestive systems. These organ systems are often affected in patients with acute radiation syndrome, which is characterized by nausea, vomiting and myelosuppression, and occurs on a timescale of minutes to hours up to several months after whole-body or substantial partial-body exposure of >1 Gy (ref. 6). Conversely, chronic radiation syndrome develops over a timescale of months to years after an annual whole-body fractionated radiation exposure of >0.7 Gy and is characterized by structural and functional organ impairment accompanied by fibrosis, atrophy and vascular or neural damage^6,7^.
Numerous signalling mechanisms, and specifically non-targeted or radiation-induced bystander effects, are likely to have important roles in late-onset chronic radiation syndrome^7^. Bystander effects are biological responses to ionizing radiation observed in non-irradiated cells owing to communication with irradiated cells through gap junctions or via soluble factors^8^. Irradiation at sublethal doses results in epigenetic and metabolic remodelling that leads to senescence associated secretory phenotypes^9,10^. Soluble factors include reactive oxygen species or reactive nitrogen species, cytokines, growth factors, chemokines and exosomes from proximal or distal cells (Fig. 1b); these mechanisms have been discussed previously^11^. Bystander effects can be induced at very low doses, but the magnitude of the effects does not always correlate with radiation dose, highlighting their importance in the long-term health complications associated with ionizing radiation exposure <0.5 Gy (ref. 9).
Emerging evidence indicates that late-onset cardiovascular disease (CVD) and cataract also occur after protracted low-dose ionizing radiation exposure at doses >0.5 Gy (refs. 12,13). In 2012, the International Commission on Radiological Protection acknowledged CVD as a tissue reaction effect^6^. Radiation-associated CVD (RACVD) is a multifaceted syndrome with varying magnitudes of dose and response for different subtypes of CVD. In this Review, we discuss the risks associated with radiation exposure to the cardiovascular system and describe the various types of exposure that have been associated with CVD, their clinical relevance and possible modifying risk factors.
Radiation dose is expressed in absorbed dose (the amount of energy that radioactive sources deposit in materials) in grays; 1 Gy is equivalent to 1 J deposited per kilogram of mass. In terms of radiation effects, radiation is expressed as effective dose in sieverts, which accounts for radiation quality and tissue weighting factors to adjust for differences in biological effectiveness of radiation exposure. RACVDs are generally late-onset conditions associated with excess morbidity and mortality in various irradiated populations^14^ (Fig. 2a).
The Life Span Study^15^ (LSS) involved a cohort of ~120,000 survivors of the atomic bombings in Japan, who were prospectively followed up for >60 years from October 1950 to assess the long-term health risks of whole-body radiation exposure. This cohort included individuals who received uniform, whole-body doses ranging from 0 Gy to 4 Gy (predominantly γ-rays, with a small fraction of neutron radiation). A mortality report from the follow-up period 1950–2003 documented an excess relative risk (ERR) per gray of 0.09 (95% CI 0.01–0.17, P = 0.02) for cerebrovascular disease (CeVD) (indicating a 9% increase in mortality per 1 Gy increase) and an ERR/Gy of 0.14 (95% CI 0.05–0.23, P < 0.001) for heart disease^15^. The dose–response association became significant for exposures >2 Gy for CeVD and >0.5 Gy for heart disease^15^. In a subsequent mortality report that included patient follow-up data up to 2008, specific heart disease subtypes were assessed^16^. A significant ERR/Gy of 0.45 (95% CI 0.13–0.85) was documented for valvular heart disease (VHD), 0.36 (95% CI 0.10–0.68) for hypertensive organ damage and 0.21 (95% CI 0.07–0.37) for heart failure (HF). Of note, the ERR/Gy for ischaemic heart disease (IHD) was not significant^16^.
Nuclear and medical workers can be subject to moderate to high cumulative doses of radiation (sometimes receiving >1 Gy, but often much less), via highly fractionated low-dose rate exposure. The health risks of occupational exposure are of particular concern for workers who were employed in the mid-twentieth century, when the cumulative doses were generally much higher than those in the modern workforce^17^. For example, workers at the Mayak Production Association, the first and largest nuclear materials production complex in Russia, received a high cumulative dose both from external γ-rays (mean cumulative external dose ~0.5 Gy) and from α-particles from plutonium (mean cumulative liver dose ~0.3 Gy), particularly in the early period of plant operations (late 1940s to 1950s)^18–23^. After 70 years of follow-up, significant associations were identified between radiation dose and morbidity or mortality related to hypertension^24^, IHD^20^, CeVD^18,22^ and lower-extremity arterial disease^21^. Unlike the LSS cohort of survivors of the atomic bombings, the contribution of internal exposure from plutonium was high in the Mayak cohort, and this exposure contributed independently (via α-radiation) to the increased risk of CVD^18,20^. Similarly, significant associations between radiation exposure and morbidity or mortality related to CVD and CeVD were identified in clean-up workers at the site of the Chernobyl power plant^25–27^ and in residents living near to the Techa River, which had been contaminated with radioactive waste from the Mayak nuclear complex^28^ (Table 1). However, not all occupational or environmental radiation exposure has been linked with an increase in the incidence of, or death from, CVD or CeVD^12,29–35^. This discrepancy might be a reflection of the complexity of RACVD manifestations at low-dose ionizing radiation exposure, although low statistical power is another probable explanation for the absence of detectable risk in some groups.
The International Agency for Research on Cancer (IARC) conducted a study involving nuclear industry workers from 15 countries and found weak evidence of an increased risk of death related to CVD and CeVD and negative trends for IHD and other cardiac end points^36^. The International Nuclear Workers Study^37^ (INWORKS) was undertaken to follow up on the three largest national workforces in the IARC study (France, UK and USA) and found that occupational radiation exposure was significantly associated with mortality from CVD (ERR/Sv = 0.22, 90% CI 0.08–0.37), IHD (ERR/Sv = 0.180, 90% CI 0.004–0.360), acute myocardial infarction (AMI) (ERR/Sv = 0.26, 90% CI 0.03–0.51) and CeVD (ERR/Sv = 0.50, 90% CI 0.12–0.94). The exposures in these cohorts were accumulated at uniformly low dose rates, with low cumulative doses (mean 25.2 mSv, ranging from 0 Sv to 1.9 Sv)^37^. Both INWORKS and the IARC study^38,39^ conducted extensive analyses to adjust for dose errors; most analyses minimized the effect of these errors by focusing on radiation dose measured from external beams of high-energy photons (100–3,000 keV) and excluding individuals with substantial exposure to alternative radiation types, such as neutrons or internal emitters^38,39^. The results of these studies raise concerns about possible risks associated with low-dose occupational exposure to ionizing radiation, such as in medical workers performing cardiac catheterization procedures who receive protracted ionizing radiation from X-ray equipment^40^. The magnitude of dose varies depending on the type of the procedure performed, but a median annual dose of 5 mSv and a lifetime exposure of 50–200 mSv have been reported in workers in a cardiac catheterization laboratory^41^. Furthermore, blood samples from these workers show evidence of genotoxicity or altered redox status^30,42,43^, and one study reported a substantial increase in risk factors for CVD (including hypertension and hypercholesterolaemia)^41^. Exposure to low-dose ionizing radiation can induce both immediate and long-lasting adverse effects, regardless of the dose received, and two systematic reviews reported a linear trend in morbidity or mortality related to CVD or CeVD across various cohorts exposed to occupational or environmental radiation^12,44^. Of note, one systematic review reported substantial interstudy heterogeneity (Fig. 2b), possibly resulting from lifestyle and medical factors known to modify the risk of CVD^12^ (Table 1). This heterogeneity was reduced if only higher-quality studies or those with moderate or low dose rates were considered.
Generalizing RACVD risk estimates derived from the LSS cohort to other populations is challenging, given that the types of exposure between these cohorts might be very different. The atomic bomb survivors in the LSS cohort (single acute exposure, mean dose 0.1 Gy (range 0–4 Gy)) and the Mayak workers (chronic exposure, mean dose ~0.51 Sv (range 0 to >4.5 Sv)) both differ from the INWORKS occupational exposure cohort, who experienced chronic low-dose exposures over a working lifetime. However, the ERR/Gy values in these three groups are similar, at least within a factor of two. One possibility is that the dose rate effect for CVD might be influenced by the delivery time of radiation, which can influence the biological response and potentially mitigate the lethality of the radiation dose. A notable result from the aforementioned systematic review was the increase in ERR/Gy by a factor of approximately two when the radiation dose was delivered in a fractionated manner rather than in an acute manner^12^. Furthermore, a fourfold increase in ERR/Gy was observed when the maximum dose received was reduced from >5 Gy to <0.5 Gy (ref. 12). Intriguingly, a steeper dose–response slope for various types of circulatory diseases was observed in patients with tuberculosis who had been exposed to a fractionated fluoroscopic X-ray dose of <0.5 Gy (ref. 45). Consistent with this finding, inflammatory markers and adhesion molecules were found to be upregulated in humans and in experimental animals after exposure to moderate-to-high doses of radiation (0.5–5.0 Gy)^46–49^. However, after low or low-to-moderate exposures of <0.5 Gy, the balance shifts towards anti-inflammatory effects^48^. Radiobiological experiments in animals that assessed the effects of acute and chronic radiation exposure have demonstrated an upregulation in proteomic markers and atherosclerotic lesion development that showed marked non-linearity in dose–response^50,51^.
Various measures of organ dose have been used to estimate the risks of CVD associated with radiation exposure. The LSS^15,16^ and IARC study^36^ used colon dose, the Mayak analysis^52^ used external γ-ray dose and the INWORKS cohort^37^ used personal dose equivalent (Hp(10)), which was measured using a tissue-equivalent film badge worn at the surface of the body. No consensus exists on the most appropriate target organ that should be used to estimate the dose from internal emitters received by certain occupational groups^18,52^ (Box 2).
The radiation weighting factor that is generally used to estimate cancer risk is determined by linear energy transfer (LET), a measure of the energy density deposited along the radiation track. LET values typically vary from ~1 keV/μm for X-rays and γ-rays to ~10 keV/μm for protons, ~10–100 keV/μm for carbon ions and >100 keV/μm for heavier charged particles. Radiation with increased LET more readily ionizes along its tracks and generates more complex and clustered double-strand breaks^53^, which might underlie the greater efficacy in tumour control. Charged-particle therapy has been adopted clinically partly for this reason^54^. However, high-LET radiation can also damage normal tissue, and the weighting factor related to the risk of CVD that should be applied to the relevant dose in the target organ is unknown. Several experimental animal studies have assessed the changes in cardiac structure and diastolic function that are associated with high-energy heavy ion exposure at levels that are broadly consistent (using unweighted dose) with those resulting from the same doses of external γ-rays^55,56^.
Evidence from numerous cohorts has demonstrated an increasing relative risk of radiation-associated CVD with increasing time after exposure^44,57^. A systematic review reported that the increase in IHD-related mortality became significant in individuals with chronic occupational exposure to low-dose ionizing radiation after a latency period of ≥10 years (ref. 44). These findings mirror results from a study assessing individuals who experienced environmental radiation exposure, in whom CVD and IHD mortality became significantly associated with radiation dose after a lag time of ≥15 years, but not for a lag time of ≤10 years (ref. 28). However, in the Mayak cohort, only weak indications of changes in CVD and CeVD mortality and IHD morbidity or mortality were observed, with a latency period of up to 30 years (refs. 18,23). Of note, for the incidence of CeVD, there were stronger indications of increasing ERR with increasing latency^18^. Although changes in risk with varying lag times were minimal, a meta-analysis documented stronger trends for increased IHD mortality in subgroups with exposures that had a lag time of 10–20 years (ref. 44), suggesting that differences in exposure duration or latency among cohorts might contribute to variation in the relative risk of radiation-associated CVD.
A mortality analysis of the Japanese LSS cohort during 1950–2008 reported no elevated risk of IHD-related death^16^. However, when interpreting this finding, we need to bear in mind the relatively greater role of hypertension and the lesser role of elevated cholesterol levels as a cause of CVD or CeVD in the Japanese population compared with the Western population^57^. Occupational studies^37^, particularly older studies of national workforces^36^, tend not to include data on lifestyle factors (and often include only surrogates, for example, socioeconomic status, assessed via job title or level of education). However, several subsequent and smaller cohort studies^12,23,58^ have included data on major lifestyle and clinical risk factors for CVD, in particular blood pressure levels, smoking status, concomitant diabetes mellitus and body mass index, all of which are known to independently modify the risk of CVD or CeVD^59^. Although adjusting for non-radiation risk factors has not consistently affected the significant association between radiation and the incidence of CVD in many studies^12^, the interaction between total-body irradiation and these risk factors remains unclear.
Ionizing radiation is widely used for diagnostic imaging and radiation therapy, which often exposes the thorax, and sometimes the heart, to radiation (Fig. 3). For diagnostic imaging, the radiation dose varies depending on the imaging procedures (X-rays <0.5 mSv; coronary CT angiography ~20 mSv)^60^. Overall, no changes in the risk of CVD have been found with diagnostic imaging^12,57^, although there are indications of excess risks for individuals receiving fluoroscopy as a part of tuberculosis diagnosis^45^. Repeated diagnostic imaging can easily result in patients accumulating an effective radiation dose of >100 mSv, necessitating additional precautions to prevent potential health risks^61^. Radiation therapy-associated CVDs have been observed in numerous populations treated for cancer^12^ (Table 2 and Supplementary Table 1). A crucial aspect of assessing the population risk of radiation therapy-associated CVD is taking into account the competing risks of a second cancer. This assessment has generally not been performed in evaluations of population risk^12,14^, but would involve a straightforward calculation if the mortality for the group of interest (cancer survivors) could be used.
Radiation therapy is a cornerstone of cancer treatment and is used in >50% of patients with cancer. Radiation therapy can be delivered as a sole definitive treatment, as palliative therapy or as an adjuvant to surgery or chemotherapy^62^. However, the adverse effects of radiation therapy are a concern, particularly in children with cancers such as leukaemia, lymphoma, brain tumours and sarcoma, in whom cardiac morbidity can substantially affect quality of life^63^. The Childhood Cancer Survivor Study^64,65^ involved a large, US-based cohort (median age of 6.1 years at diagnosis and 27.7 years at the last follow-up) with mean heart doses (MHDs) ranging from 7.1 Gy to 7.8 Gy. The study reported an excess risk of HF, coronary artery disease (CAD) and VHD for MHD >15 Gy (ref. 65), but these findings were reliant on only partially individualized dosimetry and self-reported health outcomes^64–66^. Numerous European studies provided better individualized dosimetry (MHD range 7.5–11.9 Gy) and also reported significant excess cardiac mortality (standardized mortality rate/Gy 1.6, 95% CI 1.2–3.5) with increased risk of cardiac disease in all patients (ERR/Gy 0.6, 95% CI 0.20–2.5) and in patients not receiving concomitant anthracycline treatment (ERR/Gy 0.49, 95% CI 0.26–1.3)^67–69^. However, these European studies often lack information on risk factors, such as diabetes and hyperlipidaemia, both of which have been shown to interact with radiation therapy-associated CVD in the Childhood Cancer Survivor Study^64,65^. Furthermore, similar to the Childhood Cancer Survivor Study, the European studies often rely on self-assessed health outcome data, albeit with some medical record validation^67–69^.
Radiation therapy has been central in the treatment of Hodgkin lymphoma, which has a 5-year relative survival rate of >80% (ref. 70). During radiation therapy, patients generally receive 20–44 Gy of ionizing radiation to the neck and chest region^70^. An early retrospective analysis of patients with Hodgkin lymphoma treated with radiation therapy reported a significantly increased risk of CVD-related hospitalizations and mortality owing to AMI, congestive HF, pericardial effusion or VHD^71^. Patients with Hodgkin lymphoma undergoing mediastinal radiation therapy have a greater than fourfold increase in the risk of CVD compared with the general population or with patients undergoing non-mediastinal radiation therapy^72–75^. In particular, younger patients (aged <20 years) with Hodgkin lymphoma have the highest risk of RACVD, and the relative risk of RACVD increases with increasing time after treatment^71–74^. Concurrent chemotherapy, particularly anthracycline based regimens, such as adriamycin, bleomycin, dacarbazine and vinblastine, is an independent risk factor for the development of VHD and HF^72,74,76^. Using MHDs, many studies have shown similar (and significant) ERR/Gy for the incidence of major cardiovascular events^76^, adverse cardiac outcomes^77^ and CAD^78^. Of note, many of the studies reporting radiation therapy-related CVDs discussed earlier primarily involved patients treated in the mid-to-late 1900s with 2D mantle field irradiation, which resulted in a high MHD of 25–30 Gy for a prescribed radiation therapy dose of 30–40 Gy (ref. 79). By contrast, the advent of 3D involved-field radiation therapy led to a substantially reduced prescribed dose (to as low as 20 Gy), with MHDs reduced to <10 Gy and with the volume exposed being <35% (refs. 70,80–83). This change should lead to a considerable reduction in the excess risk of CVD associated with involved-field radiation therapy compared with the use of 2D mantle field irradiation^84^. However, the lifetime cardiovascular risk associated with an MHD of 10 Gy remains considerable^12^.
Radiation therapy is often prescribed as an adjuvant therapy after surgery or with chemotherapy for patients with breast cancer and has been effective in preventing cancer recurrence and reducing mortality^85^. Patients with breast cancer have typically been prescribed dose equivalents of 40–50 Gy, with an additional 10–16 Gy boost dose^86^. Notably, an increase in cardiac morbidity and mortality ratios by laterality (that is, depending on whether the cancer is in the right breast or in the left breast) has been observed in these patients, especially in the incidence of CAD^87–90^. Numerous studies of patients with breast cancer undergoing radiation therapy have reported an increase in IHD mortality^91–94^. Of note, with the exception of CAD, the risk of other cardiac diseases associated with radiation therapy among patients with breast cancer is often not significantly increased^95–97^ and might be complicated by the concomitant use of anthracycline^96,98,99^ or targeted treatments^100^. Similar to patients with Hodgkin lymphoma, the use of conformal delivery of irradiation at a tangential angle to spare the heart has reduced the MHD^70^. Studies in patients recruited between 2000 and 2010 have shown mixed results and generally have not reported any significant associations between radiation therapy and cardiac morbidity or mortality^101–104^.
Oesophageal cancer more often affects men, presents at an advanced stage and has a poor survival rate^105^. Radiation therapy is recommended for locally advanced cases and is prescribed with a dose of 50–60 Gy in conjunction with induction chemotherapy^106^. The prognosis of patients with oesophageal cancer after radiation therapy is often poorer than in patients with other types of cancer, given the higher recurrence and metastasis rate^107^ and the higher risk of cardiac mortality^108^. A more advanced age at the time of treatment (median age ≥65 years) contributes to the development of cardiac morbidity, primarily as pericardial effusion and atrial fibrillation, within 1 year (refs. 109–112). Pericardial effusion is classified as asymptomatic or symptomatic, with a significant association observed between symptomatic cases and an absorbed volume of ≥40 Gy by the heart or pericardium^110,111,113^. Two studies of patients with oesophageal cancer have provided quantitative informative regarding the ERR of adverse cardiac events and, notably, take other risk factors into consideration^114,115^. In the MD Anderson oesophageal cancer cohort^114^, MHD was significantly associated with adverse cardiac events (Common Terminology Criteria for Adverse Events (CTCAE)-defined grade 3+ events)^114^. Similarly, in the Shandong University oesophageal cancer cohort^115^, a higher MHD was significantly linked with an increased cardiac event rate and poor survival. Advances in treatment planning technology, such as 3D conformal radiotherapy, intensity-modulated radiation therapy (IMRT), volumetric-modulated arc therapy and proton beam therapy, can reduce the risk of cardiac events and improve overall survival^114–116^.
Lung cancer is the leading cause of cancer-related death worldwide^105^, and radiation therapy for locally advanced non-small-cell lung cancer (NSCLC) has been associated with substantial cardiotoxicity. Patients with various stages of central or peripheral NSCLC generally receive a radiation therapy dose of >60 Gy with concurrent chemotherapy^117^. A phase III randomized clinical trial evaluating dose escalation for NSCLC found a significant decrease in overall survival with a higher radiation dose (74 Gy) compared with the standard dose (60 Gy)^118^, and significant cardiac exposure was observed in the high radiation dose group^119^. However, a survey of the literature suggests that MHD or cardiac volume exposure is not always associated with the overall survival of patients with NSCLC^120–122^, reflecting a challenge in estimating the risks of RACVD for this cancer, given the variation in primary tumour locations and their proximity to the heart^117^. However, several studies have shown an association between MHD and symptomatic cardiac events^119,123,124^, major adverse cardiac events (a composite of non-fatal CeVD, non-fatal myocardial infarction (MI) and cardiovascular death)^125,126^, AMI^123^, CTCAE grade 2+ cardiac events^127^ and CTCAE grade 3+ cardiac events^128,129^. Symptomatic pericardial effusion tends to be prevalent when considering CTCAE grade 2+ cardiac events^127,130,131^, whereas congestive HF or AMI tends to be more common among the CTCAE grade 3+ cardiac events^123,124,130^. Similar to patients with oesophageal cancer, patients with lung cancer are typically older (median age 65 years) than patients with other cancer types, and the development of symptomatic cardiac events can occur within 2 years of treatment^119,121,126,131^.
Modern radiation therapy planning and delivery are continuously evolving, resulting in a decline in doses delivered to non-targeted tissues over time. The strongest supporting evidence for a link between radiation therapy and RACVD was derived from older studies (1970s onwards) on patients with Hodgkin lymphoma who received higher radiation doses and less conformal radiation therapy. Advances in radiation therapy modalities, such as IMRT, volumetric-modulated arc therapy and proton beam therapy, as well as more precise anatomical planning (Box 3), have led to a substantially reduced MHD^132^ and consequently improved CVD and survival outcomes^114,121,126,130,133–136^. However, certain cardiac substructures remain exposed to radiation, which might result in the continued presence of specific types of RACVD. Subsequently, we describe different subtypes of RACVD that are potentially associated with site-specific radiation exposure (Table 3 and Fig. 4).
Radiation-associated coronary complications, specifically IHD or MI, have been observed in numerous cohorts^78,93^. MHD has been the primary dose metric used in the analyses of CVD in various cohorts of patients with cancers and is associated with increased ERR of coronary artery complications^78,91–93,123,137^, albeit with some exceptions^104,138^. Given that MHD is a crude summary of an often-heterogeneous dose distribution in the heart resulting from modern types of radiation therapy with 3D planning^139^, its association with an increased risk of CVD is noteworthy. Several studies have assessed the relationships between coronary artery complications and radiation dose by focusing on the volume exposed^95,121,126,140^, the chambers of the heart^95,112^ or specific segments of the coronary artery^77,140,141^. In one study that assessed both MHD and volume of the left ventricle receiving ≥5 Gy (LV–V5) in patients receiving radiation therapy for breast cancer, the volume metric was a better predictor of acute coronary events than MHD^95^. Another study tested a refined dosimetry using five left ventricle segments and six coronary artery segments to compare the relative ratio of the incidence of MI and CAD among patients with left-sided versus right-sided breast cancer^140^. The apex of the left ventricle received the highest dose of radiation and showed the strongest correlation with the incidence of MI, whereas among the coronary segments, the mid or distal sections of the left anterior descending (LAD) coronary artery received the highest dose and showed the strongest correlation with the incidence of CAD. Although MHD can still adequately predict the risk of acute coronary events^77,97^, several studies have highlighted the improvement in using LAD dosimetry to predict coronary complications^77,95,121,126,142–144^. Specifically, numerous studies have reported a stronger association between coronary artery stenosis and LAD dosimetry compared with MHD^90,144–148^. The examination of baseline coronary artery health using coronary artery calcium score and the early detection of coronary artery stenosis using coronary CT angiography facilitated the early identification of coronary artery complications among asymptomatic patients undergoing radiation therapy^146,147^, depending on their age at exposure and the follow-up period^78,121,137,149^.
The mechanistic pathways underlying irradiation-mediated coronary artery complications have been extensively studied in animal models. A single or fractionated high dose of X-ray to the heart of rabbits has been shown to induce a rapid decline in capillary density and loss of endothelial viability before the onset of any symptoms or structural changes^150^. These findings are consistent with autopsy findings showing that high-dose radiation causes vascular lesions, primarily in coronary vessels^151^. Mechanistically, radiation exposure affects the endothelium by initiating an inflammatory cascade^152^, accompanied by both molecular changes (senescence and coagulation) and functional changes (angiogenesis and barrier dysfunction)^153^.
Radiation-associated valvular complications involve progressive valvular thickening and calcification with subsequent valve regurgitation and/or aortic stenosis. The prevalence of radiation-induced valvular complications in patients with Hodgkin lymphoma is estimated to range from 2.9% to 17% (ref. 154). VHD generally manifests after an extended period from radiation therapy (>20 years), and the use of anthracycline-based chemotherapy also significantly increases the risk of VHD^72^. Given the advent of technology that allows contouring of valvular structures with CT scanning, retrospective analyses using valvular substructure dosimetry in patients with Hodgkin lymphoma have reported significant variation and a poor correlation with MHD using modern radiation therapy modalities, such IMRT or proton beam therapy^139,155^. Two studies have used chamber or valvular dose to assess VHD incidence^83,156^. Asymptomatic valvular complications were found to be strongly correlated with chamber volume dose proximal to the corresponding valves^83^, whereas a significant correlation was also noted between valvular dose and the incidence of VHD^156^.
The mechanisms underlying radiation-induced valvular complications are not fully understood, but several studies suggest that direct irradiation might have an important role^154^. Primary human aortic valve interstitial cells exposed to 10 Gy of γ-radiation in vitro showed an increased production of osteogenic factors (bone morphogenetic protein 2, osteopontin and alkaline phosphatase) that could potentially contribute to osteogenic morphogenesis^157^. Furthermore, targeted irradiation of the aortic valve in a mouse model induced aortic valve remodelling, and an increase in peak aortic jet velocity and mineralization was observed 3 months after irradiation^158^. Of note, valvular complications were more evident in mice with a deficiency in apolipoprotein E (Apoe^–/–^ mouse model).
Radiation-associated cardiac conduction complications include atrioventricular block, pathological node syndrome, QTc prolongation, supraventricular arrhythmia and ventricular tachycardia. These conduction complications occur in up to 5% of patients who have received radiation therapy, generally ≥2 months after treatment^119,159–161^. Patients receiving radiation therapy for breast cancer are more likely to need permanent implantation of a pacemaker than patients with breast cancer who have not received radiation therapy; chemotherapy did not increase the need for a pacemaker in these patients^162^. Furthermore, several studies have suggested an effect of radiation on the sinoatrial or atrioventricular nodes^159,160,163^. The right atrium dose has been reported to be a stronger predictor of arrhythmias than MHD^159^, and when the sinoatrial or atrioventricular node radiation dose was manually contoured, sinoatrial node dose was found to be higher among patients with right-sided breast cancer than those with left-sided breast cancer^163^. A similar study conducted in patients with Hodgkin lymphoma reported a substantial sinoatrial node dose (average mean dose of 6.6 Gy and an average maximum dose of 11.0 Gy), which correlated poorly with MHD^160^. The patients in these studies were all treated with conformal radiation therapy with relatively low MHD (~4 Gy), underscoring the potential risk of conduction complications, particularly in patients with right-sided breast cancer, who are traditionally considered to have a lower risk of radiation therapy-associated CVDs^142,163^. For patients with a high MHD (>10 Gy), the association between conduction complications and substructure dose is clearer. One study reported that half of all patients with NSCLC who developed grade 3+ acute coronary events within 2 years of radiation therapy had conduction complications^121^, whereas another study involving patients with NSCLC demonstrated a significant association between the incidence of atrial fibrillation and overall survival rate with maximal sinoatrial node dose (>20 Gy) and maximal right atrial dose (>19.1 Gy)^122^. Similarly, 21.4% of patients with oesophageal cancer who received radiation therapy developed atrial fibrillation, which was significantly associated with overall survival rate and mean left atrial dose^112^.
Radiation therapy-associated arrhythmia has also been widely reported. Cardiomyocytes derived from human induced pluripotent stem cells (iPSCs) showed a significant decline in beating rate 48 h after exposure to 5 Gy or 10 Gy of X-ray radiation^164^. Irradiated cells that were exposed to higher doses of radiation were more prone to changes in electrophysiological spatial distribution. A similar study using human 3D cardiac spheroids that included diverse cardiac cell types (atrial, ventricular, sinoatrial and atrioventricular node and Purkinje cells) evaluated the cardiac effects of various doses of X-ray irradiation (0.1–2.0 Gy). X-ray-irradiated cardiac spheroids showed a transient increase in beating rate 7 days after irradiation and increased arrhythmia 28 days after irradiation^165^.
Patients with Hodgkin lymphoma or oesophageal cancer have an ~30% prevalence of pericardial complications (including thickened pericardium, acute and chronic pericarditis, pericardial fibrosis or delayed pericardial effusion)^166,167^, which presents within a few months after radiation therapy^109^. Several studies have highlighted an increased incidence of pericardial effusion (ERR/Gy ~5%) in relation to pericardial dose^131,168^. Pericardial complications often received less attention than other cardiovascular complications, especially in survivorship studies, given that most asymptomatic cases are self-resolving^167^. However, symptomatic pericardial effusion can develop when a substantial volume of the pericardium is exposed to irradiation^109,113,169^. Several studies have demonstrated that in patients with oesophageal cancer or NSCLC, symptomatic pericardial effusion (a CTCAE grade 3+ cardiac event) was significantly associated with the volume of the pericardium receiving 50 Gy (refs. 113,168,169). Furthermore, given that ~90% of patients with oesophageal cancer or NSCLC undergo concurrent chemotherapy, the risk of pericardial complications directly caused by radiation exposure is difficult to determine^113,131,168,169^. Unresolved pericardial complications can lead to constrictive cardiomyopathy and impaired diastolic function^170,171^.
Several animal studies have modelled radiation-associated pericardial effusion and cardiomyopathy^172^. A study involving Dahl salt-sensitive rats identified a sex-based disparity in the development of pericardial effusion with hypertrophic remodelling and reduced stroke volume^173^. However, male and female rats received different lung doses of radiation by volume, and the sex-specific disparity was resolved when lung doses were matched^173^. Other studies also suggest a cardiopulmonary interaction in survival rates and the development of HF after irradiation^174,175^.
Radiation-associated myocardial complications are uncommon and difficult to estimate owing to the substantial interactions with chemotherapy-induced damage. One study reported mildly reduced systolic function in patients with Hodgkin lymphoma undergoing radiation therapy^176^. Radiation-associated myocardial complications can include constrictive cardiomyopathy accompanied by pericardial complications and/or congestive cardiomyopathy with diastolic dysfunction^98,177^. In patients undergoing conformal radiation therapy for breast cancer, the relative risk of HF with preserved ejection fraction (but not HF with reduced ejection fraction) was found to increase with increasing MHD^98^. Furthermore, imaging studies have reported evidence of myocardial fibrosis in patients undergoing radiation therapy^178,179^; an increase in fibrosis owing to haemodynamic or inflammatory stress underlies the development of HF with preserved ejection fraction^180^ and is also a hallmark of late tissue response^6^.
Several preclinical studies have also assessed the relationship between radiation dose and myocardial complications. Acute wholeheart irradiation of mice (16 Gy dose) induced myocardial fibrosis, increased inflammation and reduced microvascular density^181,182^. Furthermore, a multiomics analysis of C57BL/6 mice that received 16 Gy identified changes in oxidative phosphorylation and lipid metabolism, which were indicative of mitochondrial dysfunction^183^. Apoe^–/–^ mice subjected to protracted total-body γ-ray irradiation (6 Gy dose) for more than 300 days showed myocardial mitochondrial dysfunction^51,184^; somewhat similar changes have been reported in a proteomic analysis of post-mortem heart samples from the Mayak workers^185^. Finally, in a study that sought to explore whether genetic factors could modify tissue sensitivity to radiation, the substitution of rat chromosome 3 from the Brown Norway rat strain into the parental Dahl salt-sensitive rat (a highly sensitized model of radiation-induced cardiotoxicity) significantly reduced myocardial complications that were induced by cardiac irradiation (24 Gy) and improved survival by alleviating systolic dysfunction, fibrosis, inflammation and mitochondrial dysfunction^186^. This effect was attributed to an increase in the expression of the oxidative stress-modulating transcriptional regulator nuclear factor erythroid 2-related factor 2 (NRF2), which is encoded on rat chromosome 3. The substantial energy demands of the myocardium suggest that myocardial complications associated with radiation exposure might be driven by mitochondrial dysfunction.
Radiation-induced tissue reactions are primarily caused by the dose and volume of radiation exposure. However, the propagation of tissue reactions varies among individuals owing to intrinsic factors that influence the biological response after radiation exposure. These factors include age at exposure, biological sex and genetics, all of which alter the radiation response and the overall magnitude of radiation associated health risks. Although the importance of these intrinsic risk modifiers has been recognized, incorporating these variables into risk assessments is challenging.
Exposure to radiation during childhood is often associated with a higher risk of RACVD, partly owing to the longer life expectancy and the fact that developing organs in a child are more sensitive to radiation than fully developed organs in adults^187^. Data from the LSS cohort suggest that radiation-associated ERR/Gy for CVD decreased with increasing age at exposure, although some variations were observed across different end points^14,16,188^. In patients with cancer, the evidence on the effect of age at exposure on RACVDs is limited because most of these individuals are exposed to radiation in later life, at a narrower age range than the LSS or occupational exposure cohorts. In patients with Hodgkin lymphoma, the likelihood of developing coronary diseases and cardiac death is higher in younger patients than in the overall population undergoing radiation therapy^189^. Likewise, the relative risk of IHDrelated death among patients with breast cancer was much higher in patients who were diagnosed at age <60 years than in those diagnosed at age ≥60 years (ref. 190). However, the risk of CVD was not assessed in relation to cardiac dose in either of these two cohorts. The effect of age at exposure on the risk of RACVD has been explored in controlled experiments using rat and mouse models exposed to a wide range of radiation doses and analysed for various end points^191–193^. Overall, the results were not consistent, although the degree of perivascular fibrosis and the increase in blood pressure after irradiation were greater in younger rats than in adults^191^.
One emerging and potentially relevant mechanism for the development of age-related RACVD is clonal haematopoiesis of indeterminate potential (CHIP), a type of somatic mosaicism resulting in the clonal expansion of blood cells that increases the risk of atherosclerotic CVD^194^. The prevalence of CHIP increases with age, accompanied by increased variations in CHIP driver genes (DNMT3A, TET2 and ASXL1) and haematological malignancies, a hallmark of age-dependent stochastic effects after irradiation^188,194^. As radiation exposure is known to cause haematopoietic stem cell senescence and to accelerate CHIP^195,196^ and numerous studies have highlighted an increase in the prevalence of CHIP in different cohorts exposed to radiation^197,198^, CHIP might be a potential mechanism for the development of age-related RACVDs.
Evidence from both clinical and animal studies indicates that the degree of radiation-associated toxicity in normal tissues is affected by sex, predominantly owing to hormonal changes^5^. The reproductive system is very radiosensitive and radiation exposure >1 Gy can result in sterility^199^. A substantial increase in infertility with endocrine dysfunction has been reported in residents exposed to nuclear waste from the Chernobyl power plant^200^. Of note, sex hormone receptor signalling is known to confer radioresistance in cancer cells^201,202^. Oestradiol is well known for its cardioprotective properties in premenopausal women^203^, and the replenishment of oestrogen via hormone replacement therapy has been shown to significantly reduce the risk of coronary complications in patients with breast cancer who have undergone radiation therapy^91,93^. Similarly, in animal studies, female mice seem to have greater protection against thoracic radiation-induced cardiotoxicity than male mice, as demonstrated by preserved systolic function and lower levels of inflammatory cytokines^204,205^. However, the expected sex-specific disparity in ERR/Gy for heart disease was either absent or contradictory across different studies^16,37,206^, a possible reflection of the limitations of the data collected in existing studies, which often lack detailed information on dosimetry and other risk factors for CVD.
The heterogeneous adverse health effects associated with radiation exposure are in part related to variations in genetic background. From the earliest days of radiation research, differential responses in the development of CVD after acute thoracic radiation have been observed between different strains of rats^207^ and mice^208^, as well as in tissue samples from humans^209^. Considering the substantial variation in baseline coronary health among patients with cancer treated with radiation therapy^121,149^, the presence of common genetic variants associated with underlying coronary symptoms (such as coronary artery calcification)^210^ can disproportionally increase the risk of RACVD in certain individuals. Rare genetic diseases that affect DNA damage response include ataxia-telangiectasia (caused by a defect in ATM), Nijmegen breakage syndrome (caused by a variant in NBS1) and Li–Fraumeni syndrome (linked to germline variants in TP53)^211^, but the effect of these genetic variants on the development of RACVD is unclear. Measuring the contribution of common or rare genetic variants on the heterogeneous cardiovascular consequences of radiation exposure is challenging, and the field thus far has focused on the variations in gene expression patterns associated with genotype, referred to as ‘expression quantitative trait loci analysis’^212^. To date, several loci have been associated with varying responses to radiation therapy^213^. However, caution must be exercised in interpreting these results owing to multiple ways in which transcription can be regulated and the limited clinical applicability in the management of radiation therapy-related toxicity^214^.
The sources of radiation and the differing levels of exposure pose varying degrees of risk to the cardiovascular system. Advances in radiation therapy modalities have led to a substantial reduction in MHD, but cardiac substructures remain vulnerable to radiation-induced damage. In the future, other types of radiation exposure might become relevant, including radiation from space travel^215^, ultra-high-dose radiation therapy^216^ and radioablation treatments for lung cancer and ventricular tachycardia^217,218^. However, the mechanisms underlying the biological effectiveness of different radiation modalities remain unclear, and preclinical studies thus far have provided only very limited data on the variations in intrinsic radiosensitivity associated with genetic polymorphisms.
In 2022, the FDA Modernization Act 2.0 was signed into law to permit the use of alternatives to animal testing, including cell-based assays, such as human iPSCs, organoids and organ-on-chip models^219^. iPSCs offer accessible and scalable sources of human specialized cell types, such as iPSC-derived cardiomyocytes, and the incorporation of CRISPR–Cas9 genome editing will further facilitate genotype–phenotype correlation^220^. Moreover, advances in tissue engineering technologies have enabled the creation of increasingly complex multicellular 3D structures, such as self-organizing organoids and engineered heart tissues^221,222^. A novel organ-on-a-chip model has been used to examine the effects of low and high LET radiation exposure on different organ structures derived from humans^223^. However, iPSC derivatives have different developmental origins, resulting in cells that are structurally, functionally and metabolically immature^224^. Although 3D tissue engineering partially resolves the immaturity observed in 2D monolayer systems, current 3D tissue models lack vascularization and optimal culture medium conditions for long-term maintenance^225^, which are crucial for recapitulating late-onset normal tissue effects.
Finally, personalized radiation risk assessment must account for normal tissue toxicity, given that a slight alteration in the prescribed total dose can lead to substantial differences in normal tissue toxicities^226^. Although no experimental model is perfect, the use of predictive models, such as iPSC-based models, is valuable for guiding dose optimization to minimize exposure to the most vulnerable areas. Understanding the tolerable range of radiation dose for an individual can increase the therapeutic index by reducing complications in normal tissues, including those in the heart (Fig. 5).
Ionizing radiation is widely utilized in various clinical and industrial applications, but high levels of exposure pose substantial cardiovascular risks. More than a century of research has documented the relationship between radiation exposure and the excess risk of CVDs, leading to improved understanding and subsequent reductions in radiation exposure, by adhering to the ‘as low as reasonably achievable’ principle. Given that health complications after radiation exposure are multifaceted, a precise documentation of dosimetry and risk factors and an extended follow-up period are required to accurately assess risk. In particular, in the context of radiation therapy for cancer, marginal dose variations in the radiation treatment plan can result in substantial differences in latent normal tissue toxicity. A greater understanding of the variations and mechanisms underlying the heterogeneous presentation of normal tissue toxicity will pave the way for adopting applications that involve radiation exposure in a much safer manner in our daily lives.