Article Text

Download PDFPDF

Clinical outcome of coronary stenting after thoracic radiotherapy: a case-control study
  1. Christophe L Dubois1,
  2. Christos Pappas1,
  3. Ann Belmans1,2,
  4. Katrien Erven3,
  5. Tom Adriaenssens1,
  6. Peter Sinnaeve1,
  7. Mark Coosemans1,
  8. Peter Kayaert1,
  9. Caroline Weltens3,
  10. Walter Desmet1
  1. 1Department of Cardiology, University Hospitals Leuven, Leuven, Belgium
  2. 2Department of Public Health, Biostatistical Centre, Catholic University Leuven, Leuven, Belgium
  3. 3Department of Radiotherapy, University Hospitals Leuven, Leuven, Belgium
  1. Correspondence to Dr Christophe L Dubois, Department of Cardiology, University Hospital Gasthuisberg Leuven, Herestraat 49, Leuven B 3000, Belgium; christophe.dubois{at}


Objective Patients with lymphoma, lung or breast neoplasia show significant improvement in their disease-specific survival after radiotherapy (RT), but these benefits may be offset by delayed effects of irradiation of the heart. We compared clinical outcome after coronary stenting in patients with neoplastic disease and previous thoracic RT with matched patients without previous RT.

Design Single-centre retrospective case-control study.

Patients and methods Each patient with former thoracic RT undergoing coronary stenting between June 1998 and June 2005 was matched to two control patients according to several known prognostic factors (gender, age, available follow-up, stented vessel, drug-eluting stent use, unstable coronary disease, renal insufficiency, diabetes, bifurcational disease, stent length and size and ejection fraction).

Main outcome measures Major adverse cardiac events (MACE) were defined as the composite of cardiac death, acute myocardial infarction (AMI) and target lesion revascularisation (TLR) and were assessed at latest follow-up and compared using Cox regression analyses.

Results 41 patients underwent coronary stenting at 6±4 years after RT. Clinical outcome at 5±2 years after stenting was compared with outcome in 82 matched patients. For all-cause mortality, the hazard ratio for RT versus no RT was 4.2 (95% CI 1.8 to 9.5; p=0.0006). For cardiac mortality, the estimated hazard ratio was 4.2 (95% CI 1.0 to 17.0; p=0.0451). No significant differences were detected in terms of AMI, TLR, MACE or stent thrombosis.

Conclusions Our findings suggest an increased risk of all-cause and cardiac mortality in patients who underwent coronary stent implantation after previous thoracic RT. Verification in larger patient populations is warranted.

  • Radiation therapy
  • coronary stenting
View Full Text

Statistics from


Radiation-induced heart diseases are increasingly recognised as more patients who received radiation therapy (RT) survive their diseases with improved management of various malignancies.1–5 Radiation affects every component of the heart, resulting in changes that range from subclinical histopathological alterations to overt clinical disease. Pericardial involvement is the most common and includes asymptomatic pericardial effusion and constrictive pericarditis. The diseases involving the myocardium, valvular apparatus and conduction system are often subclinical. When symptomatic, they are often the harbinger of radiation-induced coronary artery disease (CAD). Improvements in the modern radiation delivery systems have minimised irradiation of the heart.6 However, emerging indications have broadened the application of RT for various malignancies in the chest, as a part of bone marrow transplantations7 and as brachytherapy for advanced pre-existing CAD.8 RT may also have an impact on the outcome of coronary revascularisation. Although coronary artery bypass grafting (CABG) is often indicated after RT because of the ostial location of the coronary lesions, and has been associated with favourable long-term outcomes,9 mediastinal RT is often considered as a contraindication for CABG since it has been complicated by extensive substernal scarring, mediastinal fibrosis and pulmonary dysfunction. In addition, increased friability of potential conduit vessels, especially the internal mammary arteries, may render them difficult to use as grafts or to suture in damaged vessels.10 Interestingly, outcome of percutaneous coronary intervention (PCI) after thoracic RT has rarely been described. Increased rates of restenosis after PCI were reported in a small series of patients with lymphoma receiving mediastinal RT, presumably caused by radiation-induced intimal hyperplasia.11 Moreover, delayed endothelialisation after PCI with stenting, as described with brachytherapy for in-stent restenosis,8 might similarly affect healing after stenting in patients who underwent previous external RT, leading to an increased risk of stent thrombosis, acute myocardial infarction (AMI) and death.

We report 30-day, 1-year and late clinical outcome in a series of 41 patients undergoing coronary stenting several years after thoracic RT and compare these with matched patients without RT.


All patients with former thoracic RT involving the heart in the radiation field and undergoing coronary stenting in a single centre between June 1998 and June 2005 were matched to two control patients each according to several known prognostic factors for adverse cardiovascular outcomes, restenosis12 and stent thrombosis.13 Parameters were entered into the model in the following order: gender, stented vessel, drug-eluting stent use (DES), available duration of follow-up, unstable coronary disease, renal insufficiency, diabetes mellitus, bifurcational disease, stent length and size and ejection fraction (EF). As soon as fewer than five matches were retained, the two nearest age matches for each RT patient were selected.

For PCI, patients were adequately pretreated with aspirin and a thienopyridine whenever possible. Intravenous unfractionated heparin was administered to maintain the activated clotting time above 250 seconds (200 seconds in case of administration of glycoprotein 2b3a inhibitors). The use of abciximab was left to the discretion of the operator. If tirofiban was being administered before PCI, it was continued throughout the procedure and for an additional 12 hours. Dual antiplatelet therapy was mandatory in all patients for at least 1 month after the procedure. In patients treated with DES this period was prolonged to at least 6 months.

Major adverse cardiac events (MACE) were defined as the composite of cardiac death, AMI and target lesion revascularisation (TLR) and were assessed at 30 days, 1 year and at latest available follow-up. Deaths were considered cardiac if no clear non-cardiac origin could be proved. AMI was defined clinically as the occurrence of symptoms with typical electrocardiographic changes or elevation of cardiac biomarkers with creatine kinase (CK) >2× the upper limit of normal in the presence of elevated creatine kinase MB isoenzyme (CK-MB). TLR was defined as any type of clinically driven revascularisation of the target vessel because of narrowing of the stented segment or the 5 mm proximal or distal to the initially dilated lesion. Besides MACE the individual components of MACE were reported as well as all-cause mortality, non-target lesion revascularisation (non-TLR) and stent thrombosis. Stent thrombosis was classified according to the Academic Research Consortium definition as definite, probable or possible and as early (0–30 days), late (31–360 days) or very late (>360 days).14 The definition of definite stent thrombosis required the presence of an acute coronary syndrome with angiographic or autopsy evidence of thrombus or occlusion. Probable stent thrombosis included unexplained deaths within 30 days after the procedure or AMI involving the target-vessel territory without angiographic confirmation. Possible stent thrombosis included all unexplained deaths occurring at least 30 days after the procedure.

Statistical analysis

Continuous variables are presented as means and SD. Categorical variables are presented as observed frequencies and percentages. For the comparison of baseline characteristics, continuous variables were compared by means of a linear mixed model that included a factor for group (RT or control) and a random intercept for each triplet, thus accounting for the matching within triplets. Categorical variables were assessed by the Cochran-Mantel-Haenszel test, using the matched triplet as stratification factor. Differences in outcome data at 30 days and 1 year between the two groups were assessed by the Cochran-Mantel-Haenszel test, using the matched triplet as stratification. In addition, for all outcomes except stent thrombosis, a Cox regression analysis was performed whereby for endpoints other than all-cause mortality, patients who died without experiencing the outcome were censored at the time of death. To account for the clustered nature of the data owing to matching, the robust sandwich estimate of the standard error of the regression coefficients was used, as proposed by Lin and Wei.15 The effect of RT was evaluated by assessing the associated hazard ratio using a Wald test based on the sandwich-estimated standard errors. Kaplan-Meier curves were created for all-cause mortality, cardiac mortality and MACE. However, for cardiac mortality and MACE, these figures do not correctly account for the competing risk of other causes of death. Therefore, for these two outcomes, cumulative incidence functions were additionally calculated that do take into account competing risks. All statistical analyses were done using SAS V.9.1. All tests were two-sided and assessed at the 5% significance level. No adjustments were made to the significance level to account for multiple testing owing to the exploratory nature of this study.


Between June 1998 and June 2005, 9173 patients underwent PCI at our institution. After excluding patients with cardiogenic shock, we identified within this group 41 patients who underwent coronary stenting at 6±4 years after thoracic RT for lymphoma, lung and breast neoplasia (n=6, 7 and 28, respectively), involving the heart in the radiation field. On average, the total prescribed radiation dose was 39 Gy for patients with lymphoma (range 33–54), 51 Gy for patients with lung cancer (range 43–60) and 50 Gy for patients with breast cancer. All doses were recalculated in biological equivalent doses of 2 Gy per fraction. Each patient was matched to two control patients without previous RT according to several known prognostic factors, as described in the model before. Patient, lesion and procedural characteristics and comparisons between groups are presented in table 1.

Table 1

Patient, lesion and procedural characteristics

Mean age at the time of PCI was 66 years, and 34% of patients were male. In each triplet, patients were well matched for diabetes (30%), renal failure (creatinine level >1.5 mg/dl) (8%) and EF (56±9%). Of note, 76% of the patients in the RT group presented with an acute coronary syndrome, well matched with the control group (77%). Acute coronary syndromes (ACS) included patients with unstable angina (34%), non-ST-elevation AMI (24%) and ST-elevation AMI (19%), equally distributed between groups (data not shown). Owing to imperfect matching, PCI of coronary bifurcations was significantly more represented in the control group with a bifurcation involved in 40% of the cases. The use of DES in this study was limited to 19% of the patients, most of the patients being included in the study before or in the early years of DES use. Cumulative stent length was 21±11 mm, reflecting the presence of multiple lesions in some patients. Nevertheless, appropriate matching for both stent length and diameter was obtained. Length of follow-up after PCI was similar in both groups (61±23 months for RT, 62±26 months for controls).

Clinical outcome at 30 days, 1 year and at latest follow-up is presented in table 2. In the RT group, one patient on dual antiplatelet therapy died after developing ventricular fibrillation on the day of PCI and was categorised as a probable acute stent thrombosis. In the control group one patient had an AMI while on dual antiplatelet therapy 1 day after stenting. This was not considered as a probable stent thrombosis since the AMI did not involve the target-vessel territory. Five patients in the RT group died during the first year of follow-up, of whom two died from cardiac causes. Both all-cause and cardiac mortality at 1 year after PCI in this group differed significantly from the control group in which none of the patients died. Cardiac deaths included one probable (see above) and one possible stent thrombosis (patient with left main DES stenting and sudden death on dual antiplatelet therapy 4.5 months after PCI). Long-term follow-up at 62±25 months after PCI showed a marked increase in all-cause mortality to 39% in the RT patients compared to 12.2% in the control patients (p=0.0006). Similarly, cardiac mortality was significantly lower in the control patients (3.7% vs 12.2% in the RT patients, p=0.045). All three cardiac deaths (unspecified deaths) in the control group occurred after 3 years of follow-up. No information on antiplatelet treatment at the time of death was available in these subjects. In contrast, all cardiac deaths in the RT group occurred before completion of 3.5 years of follow-up, in all but one patient while on dual antiplatelet treatment at the time of death. These deaths included the two patients who died during the first year as well as two sudden deaths and one unspecified death. The other components of MACE (AMI and TLR) did not differ between groups, resulting in a non-significant excess of MACE in RT patients. Figure 1 shows Kaplan-Meier curves for all-cause mortality, illustrating an early and continuously diverging curve in favour of control patients. For cardiac mortality and MACE, Kaplan-Meier curves were also created (figure 2A and B). Since these do not correctly account for the competing risk of other causes of death, cumulative incidence functions were additionally calculated and presented in figure 2C and 2D. No differences between groups were seen in terms of stent thrombosis (supplemental table 1). At latest follow-up, the rate of definite or probable stent thrombosis was 3.7% in the control group and 2.4% in the RT group (p=NS).

Table 2

Clinical outcome at 30 days, 1 year and at latest follow-up

Figure 1

All-cause mortality. Kaplan-Meier curves for all-cause mortality.

Figure 2

Cardiac mortality and MACE. Kaplan-Meier curves for cardiac mortality (A) and MACE (B). Cumulative incidence functions for cardiac mortality (C) and MACE (D).


The precise incidence of radiation-induced accelerated atherosclerosis is difficult to confirm owing to the high prevalence of CAD in Western populations. Reported estimates of relative risk of fatal cardiovascular events after mediastinal RT for Hodgkin's disease range between 2.2 and 7.2, and from 1.0 to 2.2 after irradiation for left-sided breast cancer.16 17 Anatomopathological changes in radiation-induced atherosclerosis have been described and could, besides their impact on clinical outcome, influence the selection of revascularisation strategy in these patients.4 Mediastinal RT is often considered as a contraindication for subsequent CABG owing to extensive scarring and friability of tissues and problematic wound healing.10 18 PCI therefore offers an elegant, less invasive mode of revascularisation, not limited by overt sequelae induced by RT, and its use has been described in case reports or small series to treat radiation-induced CAD with favourable immediate results.19–24 However, healing after microtrauma, as caused by balloon angioplasty and coronary stenting, might similarly be affected by the altered structural and functional integrity of the coronary vasculature after RT. Delayed endothelialisation and stent thrombosis after coronary brachytherapy for in-stent restenosis have been reported.8 Similarly, exaggerated intimal hyperplasia after external beam RT for lymphoma has been described.11 However, long-term outcome of coronary stenting in a milieu formerly affected and modified by external RT has never been reported.

We therefore conducted this study, specifically looking at cardiovascular outcomes after stenting in a relatively large group of patients (n=41) previously treated with external beam RT involving the heart in the radiation field. The proportion of female patients in our study was higher than in most reports of CAD and PCI, since 68% of our patients were treated for breast malignancy. Clinical presentation was an ACS in 77% of patients, reflecting a population at high risk for subsequent adverse cardiac events. Clinical characteristics (table 1) were well balanced in RT and control patients, except for smoking which was more prevalent in the control group. Lesion and procedural characteristics (table 1) also were well balanced between groups, except for a higher incidence of bifurcational stenting in the control group. Taken together, any difference in baseline characteristics seemed in favour of the RT patients, with fewer smokers, bifurcated lesions and a lower number of stents used in the RT group.

However, clinical outcomes were consistently worse in RT patients compared with controls (table 2). All-cause mortality was significantly higher in RT patients, a finding that is at least in part inherent to underlying malignant disease (39% vs 12.2% at latest follow-up, p=0.0006). Notably, a significant excess cardiac mortality in RT patients was already apparent at 1 year after coronary stenting and persisted at latest follow-up (12.2% vs 3.7%, p=0.045). Despite the high initial rate of ACS before stenting in RT patients (77%), the overall rate of AMI in the RT patients during long-term follow-up was 4.9%, similar to the 3.7% in control patients. The rate of clinically driven TLR was similar in RT and control patients at 1 year after stenting (14.6% vs 15.9%) and increased further during long-term follow-up (22% vs 19.5%). These numbers are in line with usually reported TLR rates in studies using mainly bare-metal stents, as was the case in 80% of patients in the present study. Conversely, these findings contrast with an aggressive restenosis pattern (86% of patients) and a 67% revascularisation rate in a previously reported series of 15 patients treated with mediastinal RT for lymphoma.11 Part of this difference may be explained by the absence of routine control angiography in our patients, avoiding the ‘oculostenotic reflex’ leading to angiography driven rather than clinically driven TLR. Moreover, patients with lymphoma usually were exposed to higher radiation doses to the mediastinum with a larger volume of the heart included in the radiation field and are consequently at higher risk for adverse cardiovascular outcomes. Our series includes only a limited number of patients with lymphoma (n=6, 15%), as opposed to all 15 described by Schömig et al.11 Six per cent of patients in the control group underwent a non-TLR at 1 year compared with none in the RT group (p=0.08). No additional non-TLR was reported during further follow-up. This difference is hard to explain, unless a more conservative approach was followed in patients with a history of malignancy. Finally, RT patients showed a non-significant excess of MACE at latest follow-up (34.2% vs 24.4%), owing to a significantly higher cardiac mortality and a numeric excess of AMI and TLR. Possibly, part of the coronary disease progression in RT patients was concealed by simultaneous progressive malignant non-cardiac disease leading to death.

Vascular healing after stenting is affected by previous external beam RT and could therefore induce stent thrombosis leading to AMI and death, as was described for endovascular brachytherapy.8 However, sequence, time interval, dose and spatial distribution are different. Brachytherapy has most often been used months after stenting to treat in-stent restenosis and limit intimal hyperplasia by delivering endovascular RT to the entire vessel wall limited to the region of interest. In contrast, we studied patients undergoing stenting 6±4 years after external beam thoracic RT using different doses and repeated applications. In our study, the incidence of definite and probable stent thrombosis analysed according to the ARC criteria14 was not different between the two groups. However, the incidence of possible stent thrombosis was more than twice as high in the RT-group as in the control group (supplemental table 1). This difference, however, did not reach statistical significance and may therefore be a spurious finding. Alternatively, the lack of significance may be due to the very low number of observations. To confirm statistical significance of a difference of this magnitude, over 300 patients per group should be studied. If this difference would be true, it would undoubtedly be regarded as clinically significant.

Our study has several limitations. Although the database was large, the number of patients undergoing coronary stenting several years after thoracic RT is relatively small. Although the time interval between RT and stenting and between stenting and follow-up was relatively large, these intervals may not have been long enough to fully appreciate deleterious effects of RT on the heart, since risk appears to increase with length of time since exposure. Nevertheless, a follow-up of 5±2 years after stenting seems long enough to evaluate clinical events related to PCI. The retrospective and non-randomised design of the study is inherent to its nature, but meticulous matching was performed in order to mitigate this limitation. Although matching of patients, lesion and procedural characteristics was performed according to variables likely to impact on MACE, restenosis and stent thrombosis, ongoing accelerated atherosclerosis after RT may still have affected MACE, without being directly related to the stenting procedure. In contrast, fatal non-cardiac events may have obscured coronary disease progression in RT patients with ongoing malignancy. Similarly, the higher risk for pulmonary emboli owing to underlying malignancy in the RT group may have led to sudden deaths that have falsely been categorised as cardiac deaths. Finally, even though the problem of confounding was addressed by matching to several known prognostic factors and by using valid statistical models to compare groups, the possibility of bias when assessing the associations of interest cannot be excluded.

In conclusion, our findings suggest an increased risk of all-cause and cardiac mortality in patients who underwent coronary stent implantation after previous thoracic RT. Verification in larger patient populations is warranted.


We thank Katrien Branders for data management and manuscript preparation.


View Abstract

Supplementary materials

  • Web Only Data 96/9/678

    Files in this Data Supplement:


  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.