Author + information
- Received April 24, 2015
- Revision received June 30, 2015
- Accepted July 15, 2015
- Published online June 1, 2016.
- Benoy N. Shah, BSc, MBBSa,b,
- Dorothy M. Gujral, MBChB, MScc,
- Navtej S. Chahal, PhDd,
- Kevin J. Harrington, PhDc,
- Christopher M. Nutting, MD, PhDc and
- Roxy Senior, MD, DMa,b,d,∗ ()
- aCardiovascular Biomedical Research Unit, Royal Brompton Hospital, London, United Kingdom
- bNational Heart and Lung Institute, Imperial College, London, United Kingdom
- cHead and Neck Unit, Royal Marsden Hospital, London, United Kingdom
- dDepartment of Cardiology, Royal Brompton Hospital, London, United Kingdom
- ↵∗Reprint requests and correspondence:
Prof. Roxy Senior, Royal Brompton Hospital, Department of Cardiology, Sydney Street, London SW3 6NP, United Kingdom.
Objectives The aim of this study was to determine the effect of radiotherapy (RT) on intraplaque neovascularization (IPN) in human carotid arteries.
Background Exposure of the carotid arteries to RT during treatment for head and neck cancer is associated with increased risk for stroke. However, the effect of RT on IPN, a precursor to intraplaque hemorrhage and thus associated with plaque vulnerability, is unknown.
Methods In this cross-sectional study, patients who had undergone unilateral RT for head and neck cancer ≥2 years previously underwent B-mode and contrast-enhanced ultrasound of both RT-side and non–RT-side carotid arteries. Presence of IPN during contrast-enhanced ultrasound was judged semiquantitatively as grade 0 (absent), grade 1 (present but limited to plaque base), or grade 2 (extensive and noted within plaque body).
Results Of 49 patients studied, 38 (78%) had plaques. The number of plaques was significantly greater in the RT than the non-RT arteries. Overall, 48 of 64 RT-side plaques (75%) had IPN compared with 9 of 23 non-RT-side (39%) plaques (p = 0.002). Among patients with plaques, IPN was present in 81% of patients with RT-side plaques and 41% of patients with non-RT-side plaques (p = 0.004). Grade 0 IPN was significantly more common in patients with non-RT-side plaques (25% vs. 61%; p = 0.002), whereas grade 2 plaques were more common on the RT side (31% vs. 9%; p = 0.03). The only clinical variable that predicted the presence or absence of IPN was RT laterality.
Conclusions This is the first study in humans to reveal a significant association between RT and the presence and extent of IPN. This may provide insights into the mechanisms underlying the increased stroke risk among survivors of head and neck cancer treated by RT.
Atherosclerosis is the underlying pathobiological substrate that accounts for most cardiovascular events. However, not all patients with atherosclerosis experience such outcomes. In recent years, much research has focused on predicting which atherosclerotic plaques will “rupture,” triggering events such as myocardial infarction or stroke, and which plaques will not rupture, leading to the concept of the “vulnerable” (or unstable) plaque (1).
An emerging key feature of such plaques is intraplaque neovessels. These neovessels are fragile, leaky, and prone to bleeding, leading to intraplaque hemorrhage, which contributes to the necrotic core of plaques and is believed to increase the risk for plaque rupture (2). Consequently, the presence of intraplaque neovascularization (IPN) has been postulated as a precursor to vulnerable plaque (3,4). A number of imaging and histological studies have revealed a clear association between the presence and extent of IPN and subsequent cerebrovascular events (CVEs), including mortality (5–7). Much of this research in humans has focused on the carotid arteries, because of their superficial location (favoring noninvasive imaging) and the association with stroke and because patients undergoing carotid endarterectomy provide a suitable model for histological comparisons.
IPN can be visualized by several imaging techniques, including contrast-enhanced ultrasound (CEUS). CEUS uses transpulmonary ultrasound contrast agents, which remain intravascular at all times, effectively acting as red cell “tracers.” Carotid ultrasonography performed following the administration of contrast permits the visualization of IPN (8), and comparisons with histological neovessel density have validated the technique’s accuracy (9).
Radiotherapy (RT) damages arterial walls and promotes atherosclerosis. The carotid arteries frequently receive significant incidental doses of radiation during RT treatment of head and neck cancers (HNCs). Radiation vasculopathy had until recently been considered quite a rare entity, but as cancer survival rates improve, patients are “outliving” their malignancies and presenting later with the long-term sequelae of cancer therapy (10).
Several studies have shown that RT of the carotid arteries is associated with increased intima-media thickness (IMT), increased carotid plaque formation, and overall an increased risk for stroke (11,12). However, the effect of RT on plaque composition, specifically IPN, has not been studied in humans. We thus performed this study to assess the effects of RT on IPN in survivors of HNC who had previously received RT.
This was a cross-sectional study of patients previously treated with RT for HNC. Ethical approval was obtained, and all patients provided informed written consent. The inclusion criteria were as follows: age >18 years, histologically confirmed cancer treated with hemineck RT to ≥50 Gy, RT administered >24 months previously, and patient able to provide written informed consent. The exclusion criteria were patients with active HNC, patients with histories of carotid endarterectomy or carotid angioplasty, patients with bilateral RT, and known allergy to sulfur or sulfur-containing drugs.
Patients with HNC who had received unilateral RT prior to December 2009 were identified via the RT database. The requirement for RT to have been administered ≥2 years previously was established for 2 reasons. First, such patients are likely to be cured of their cancer and therefore live long enough to develop atherosclerosis. Second, plaques need time to develop after RT, and thus recruiting patients after a shorter time period may have resulted in many more studies with normal results. Eligible patients agreeable to participation (only 3 patients declined) were invited to attend for a baseline questionnaire, physical examination, blood tests, and carotid ultrasonography.
Patient clinical variables
Presence of cardiovascular risk factors was defined as follows: diabetes mellitus (random serum glucose ≥11.1 mmol/l, glycated hemoglobin ≥5.8%, or current use of glucose-lowering agents or insulin), hypertension (systolic blood pressure ≥140 mm Hg and/or diastolic blood pressure ≥90 mm Hg or current use of antihypertensive agents), hyperlipidemia (fasting serum low-density lipoprotein ≥2.6 mmol/l, high-density lipoprotein <2.3 mmol/l, triglycerides ≥2.3 mmol/l, or current use of cholesterol-lowering agents), ever smoker, and family history of premature coronary artery disease (first-degree relative with myocardial infarction or stroke, age <55 years [men] or <65 years [women]). Height (in meters) and weight (in kilograms) were recorded to calculate body mass index.
B-mode, color Doppler, and contrast-enhanced carotid ultrasonography (Figure 1) were performed using a high-resolution ultrasound system (Vivid 7, GE Healthcare, Chalfont, United Kingdom) equipped with a broadband (3- to 11-MHz) transducer. All scans were performed by a cardiologist blinded to the patient’s history, including the laterality of RT (as RT had been delivered >2 years previously, potential markers of treatment such as erythema, telangiectasia, and skin tenderness were no longer present). Electrocardiographic monitoring was performed continuously, and arterial blood pressure was recorded using an automated sphygmomanometer. In summary, the proximal, mid, and distal common carotid artery (CCA), bifurcation of the CCA, and proximal portion of the internal and external carotid arteries were systematically interrogated in long-axis and short-axis views.
Plaque was defined as per the Mannheim consensus as a focal structure encroaching into the arterial lumen by >0.5 mm, a distinct area of IMT >50% greater than the adjacent wall or >1.5 mm in thickness (13). IMT measurements were taken at the far wall of the distal CCA at end-diastole using a semiautomated edge detection algorithm (EchoPAC version 8.0; GE Healthcare), and the mean value obtained was an average of 3 measurements. Plaque area was calculated by tracing around the plaque edges in end-diastolic long-axis still images and was also expressed as an average of 3 measurements (EchoPAC).
After B-mode imaging, an intravenous cannula was inserted, and 2 vials (8 ml) of SonoVue ultrasound contrast agent (Bracco Diagnostics, Milan, Italy) were administered as a continuous intravenous infusion at a standardized rate of 1.2 ml/min, providing approximately 6.5 min of contrast opacification. During this time, the right and left carotid arteries were reimaged using a specific low–mechanical index (0.20) contrast preset, with special focus on areas of abnormality (i.e., plaques) identified during the B-mode scan. All images were subsequently transferred to the EchoPAC database and also stored on disc for off-line analysis, which was performed in random order (after all patients had been scanned) blinded to the patient’s clinical details. IPN was graded semiquantitatively as absent (grade 0), limited to the adventitia or plaque base (grade 1), or extensive and/or extending into the plaque body (grade 2) by a doctor blinded to the side of RT (9).
Continuous variables are presented as mean ± SD and categorical variables as proportions. Data were analyzed on a per-patient basis and a per-plaque basis. The per-patient analyses represent paired data (2 datasets obtained from each patient, 1 from the RT-side artery and 1 from the non-RT-side artery), whereas the per-plaque analyses represent unpaired data, as there were different numbers of plaques between the RT and non-RT sides (thus these were unmatched groups). Therefore, the per-patient analyses were conducted using the paired t test (continuous data) and McNemar test (categorical data), and the per-plaque analyses were conducted with the Student t test (continuous data) and chi-square test (categorical data). All statistical calculations were performed using SPSS version 19.0 (SPSS, Chicago, Illinois). A p value <0.05 was taken to indicate statistical significance for all tests.
For determination of intraobserver and interobserver variability, Cohen’s kappa statistic was used to measure agreement between 2 different assessments, at least 1 month apart, by 1 reader and 2 different readers, respectively, of 15 randomly selected CEUS cine images. For interobserver assessment, we used 2 readers to compare against the first observer, 1 with experience in CEUS imaging and a second cardiologist with no prior experience in CEUS imaging.
Of 50 patients who consented to the study, 49 underwent B-mode and contrast-enhanced carotid ultrasound studies (in 1 patient, intravenous access proved impossible and thus contrast could not be administered). The baseline demographics of these 49 patients are detailed in Table 1. The mean age was 57 ± 8 years, 69% were male, and the mean body mass index was 26.3 ± 4.4 kg/m2.
The histological tumor type was squamous cell carcinoma in the majority of patients (38 of 49 [78%]). There were 42 patients with oropharyngeal (tonsillar) tumors and 7 patients with parotid tumors. All patients had 3-dimensional conformal RT to the section of the carotid artery we studied (i.e., from the bifurcation to the proximal CCA). Some patients would have received intensity-modulated RT to the cranial part of the internal carotid artery or external carotid artery, but we did not examine these segments in this study. The average maximum dose to the irradiated side was 53 ± 13 Gy and to the unirradiated side was 1.9 ± 3.7 Gy. The dose to the unirradiated side would be considered clinically negligible.
Almost one-half the cohort (22 of 49 [45%]) had also received platinum-based chemotherapy drugs for treatment of their HNC. The mean time duration from RT to carotid imaging was 5.4 ± 2.5 years (range 2.1 to 12.6 years). Routine full blood counts were normal in all patients. The mean serum creatinine value was 75 ± 19 mmol/l (range 46 to 128 mmol/l). The mean total, high-density lipoprotein, and low-density lipoprotein cholesterol levels were 5.3, 1.5, and 3.2 mmol/l, respectively. All patients had normal serum calcium concentrations (mean 2.2 ± 0.1 mmol/l).
Of the 49 patients examined, plaques were detected in 38 (78%), with bilateral plaques in 15 patients and unilateral plaques in 23 patients. Patients with bilateral plaques more frequently had histories of smoking than those with unilateral plaques, but all other variables were similar between the groups (Table 2). The data presented in Table 3 demonstrate the differences in plaque number, area, and prevalence of IPN from the RT and non-RT carotid arteries. There was a greater number of plaques on the RT side, although mean plaque area and total plaque burden by area were similar between RT and non-RT arteries. A total of 87 plaques were detected, with 41 plaques in the right carotid artery and 46 in the left carotid artery. On a per-plaque basis, there was no significant difference in the frequency of IPN between plaques from the left- and right-sided arteries (27 of 41 [66%] vs. 30 of 46 [65%]; p = 0.95).
IPN was analyzed on a per-patient as well as per-plaque basis, but in both cases, IPN was more commonly seen on the RT side than the non-RT side. Of all 49 patients, 29 (59%) had at least 1 plaque containing IPN on the RT side compared with just 7 patients (14%) who had ≥1 plaque containing IPN on the non-RT side (p < 0.001). Among the 38 patients with plaque, 36 (95%) had plaque on the RT side, whereas 17 patients (45%) had plaque on the non-RT side (p < 0.001). Thus, analyzing only those patients with IPN, 29 of 36 patients had at least 1 plaque containing IPN on the RT side compared with 7 of 17 patients who had ≥1 plaque containing IPN on the non-RT side (81% vs. 41%, p < 0.001). Similarly, at a patient level, the absence of IPN (i.e., grade 0) was significantly more frequent in non-RT plaques than RT plaques. Conversely, grade 2 IPN, the highest grade and indicative of the most extensive IPN, was significantly more commonly identified in RT plaques than in non-RT plaques. Figure 2 shows examples of grade 0, grade 1, and grade 2 IPN during CEUS imaging (Online Videos 1 and 2).
We examined the impact of patient variables on the presence or absence of IPN, both in all patients and in the patients with IPN on the RT-side plaques only (Table 4). In summary, there was no statistical relationship between presence or absence of IPN and any clinical variable. There was a trend toward greater IPN presence in those with smoking histories and those with RT >5 years previously, but these did not reach statistical significance (possibly because of the small number of patients).
Intraobserver agreement and interobserver agreement between the first observer and another cardiologist familiar with the CEUS technique were excellent (100% agreement, kappa = 1.0). Interobserver agreement between the first observer and a CEUS novice reader was good (80% agreement, kappa = 0.60). This level of agreement is comparable with reports from other CEUS research studies (7).
This is the first study to assess the impact of RT on plaque neovascularization. Our results show that IPN is significantly increased in plaques within arteries exposed to RT compared with arteries that did not receive RT. On a per-patient and per-plaque basis, IPN was more common in RT-side plaques, and extensive IPN (grade 2) was also significantly more frequent on the RT side than the non-RT side. These findings were independent of all clinical variables; indeed, the only variable associated with presence of IPN was RT laterality.
It is widely believed that the initial injury to the vasa vasorum is a key feature of RT-related arterial disease (radiation vasculopathy), though it has remained unclear whether this process is predominantly inflammatory or ischemic in nature (14). However, our results show that the vasa vasorum have proliferated markedly into the plaques (IPN), rather than being reduced, favoring an inflammatory process. In “conventional” (i.e., non-RT-related) atherosclerosis, proliferation of the adventitial vasa vasorum is triggered by increased production of hypoxia-inducible factor, a response to reduced local oxygen tension due to increased thickness of the intima-media complex. Because RT is known to cause increased IMT, as we also observed in this study, it is not surprising that a greater degree of IPN was observed on the RT side.
Limitations of current evidence
Many groups have investigated the effects of head and/or neck RT on the carotid arteries (imaging studies) and future CVEs (outcome studies), which have recently been comprehensively reviewed (15). However, for multiple reasons, the published research displays marked heterogeneity in this field, with a number of problems associated with these studies. First, studies that have reported an increased relative risk for CVEs after RT have either used nonmatched control groups or matched their patients with HNC to geographically distinct and distant population data (16,17). Studies have not always reported on stroke subtype (ischemic or hemorrhagic) (18), and information on CVEs has not always been judged by a clinician or by brain imaging but by patient questionnaires (19). Finally, and possibly most important, the majority of these studies did not provide data on the laterality of strokes. For example, if the right carotid artery has been exposed to RT, a clinically ischemic event would, in almost all but the rarest cases, be expected in the right cerebral hemisphere and thus produce left-sided signs. However, this level of detail is not available in most papers (16,18,19); thus, although one might assume that an ischemic CVE must be related to the RT-side carotid artery, the data confirming this are absent.
Consequently, several questions regarding radiation vasculopathy remain unanswered, most prominent of which is the mechanism by which RT affects the arterial wall and subsequently increases stroke risk. The increased frequency of IPN we observed strongly implicates an inflammatory reaction, in keeping with conventional models of atherosclerosis.
Finally, it should be acknowledged that an association has been made among IPN, intraplaque hemorrhage, and adverse cardiovascular outcomes. This association has arisen from observational data. However, association does not imply causation; we hypothesize that intraplaque hemorrhage is secondary to IPN rupture and the cause of increased stroke risk. However, there are in vivo no studies to date of carotid plaques in which increased IPN has been detected and with prospective longitudinal follow-up to determine the exact contribution of plaque hemorrhage to plaque vulnerability.
There is increasing evidence that plaque composition is clinically important, independent of stenosis severity, in relation to outcome (20,21). As a result of large international trials, carotid endarterectomy is reserved for patients with symptomatic arterial plaques with stenosis severity >70% to 80% as judged by Doppler ultrasound (22,23). However, it is not known whether superior risk stratification could be achieved by accounting for novel markers of plaque vulnerability, such as the presence or absence of IPN, rather than stenosis severity alone.
Our results demonstrate that patients with RT-related carotid disease have a significant increase in IPN, a putative surrogate marker of plaque instability. One could therefore hypothesize that these patients with increased IPN are at increased risk for CVEs. It is unknown whether surgical removal of these plaques, even if causing only mild or moderate stenosis, is preferable to medical therapy alone. Intuitively, one may conclude that the mere increased presence of plaques on the RT side could account for the higher risk for stroke. However, studies published to date have already demonstrated an independent association between IPN and symptomatic plaques and cardiovascular events, including stroke; thus, a randomized controlled trial examining whether surgical plaque removal should be based on plaque composition would be the logical next step on the basis of our findings.
Our data imply that radiation increases neoangiogenesis, though prior studies have shown RT to be antiangiogenic. However, the published research is heterogeneous on this issue. Prior research in animal models revealed that irradiation dose-dependently induces the activation of the proangiogenic nitric oxide pathway in endothelial cells through increases in endothelial nitric oxide synthase activity (24). Furthermore, the survival and subsequent recurrence of neovessels after RT has been postulated as a possible mechanism to explain tumor recurrence (25).
Multiple prior studies have shown that RT increases plaque formation, and thus one might have expected a greater plaque burden on the RT side. There was a nonsignificant trend toward greater plaque burden on the RT side (39.1 vs. 28.5 mm2, p = 0.25). A possible explanation for this concerns methodology; discrete plaques can have their circumferences traced to derive an area, but diffuse thickening of (the IMT of) an entire segment may preclude tracing an area. We were unable to calculate areas for 12 such segments, which were then excluded from the comparison between RT and non-RT sides, all of which were from the RT-side vessel. It is possible that had it been possible to measure plaque area in these cases, the difference between RT-side and non-RT-side plaque burdens may have increased.
Finally, the high levels of intraobserver and interobserver reproducibility are not surprising given the excellent spatial resolution achieved during CEUS. We believe that this is explained largely by our use of a continuous infusion of contrast, as opposed to bolus injections, as an infusion provides a steady-state concentration of contrast in the artery and thus reduces swirling artifacts, which can increase the difficulty of IPN assessment. The subsequent clarity of images obtained, together with the clear distinction between the moving white appearances of a contrast bubble against the black background of a plaque, is likely to explain the high reproducibility of the technique. However, this was a small study, and larger studies, using different ultrasound systems and/or alternative methods of contrast delivery, may yield different results.
Strengths include a well-defined patient population, blinding of the carotid scanner from the side of RT treatment, use of the contralateral artery as an “internal” self-control, and, uniquely, use of a continuous intravenous infusion of contrast, which, unlike multiple bolus injections, produces a constant concentration of microbubbles within the bloodstream.
Limitations include a small sample size, lack of another imaging modality for comparison, lack of histology for verification, and lack of baseline carotid data (prior to RT). Regarding imaging, CEUS has been shown to be an accurate technique, and carotid magnetic resonance imaging, for example, has not been shown to be superior for identification of IPN. Carotid magnetic resonance imaging can assess more features of a plaque’s composition than CEUS, including the size of the lipid core and thickness of the fibrous cap. However, these aspects of plaque composition were not the primary focus of this research. Further studies using multimodality imaging (e.g., both carotid magnetic resonance imaging and CEUS) would be of scientific benefit. Regarding histology, these patients had no clinical indications for carotid surgery. Thus, histology was impossible to obtain, though previous studies have verified the accuracy of CEUS for identification of the presence and extent of IPN. With regard to baseline carotid data, these were unavailable because patients were recruited following RT, and for many patients, RT had been administered several years previously. Thus, data on IPN (and plaque) prevalence prior to RT were not available, though prior to RT, one might have expected to detect far fewer plaques (and thus less IPN).
Plaque neovascularization, a putative surrogate marker for plaque instability, is significantly increased in arteries exposed to RT during treatment of HNC. This effect of RT on IPN is independent of all other clinical variables. These results suggest that the atherosclerotic plaques of radiation vasculopathy may demonstrate increased vulnerability and this may help explain the greater risk for CVEs in this patient population.
COMPETENCY IN MEDICAL KNOWLEDGE: IMT and plaque formation are both increased in carotid arteries exposed to RT compared with unirradiated arteries. Neovascularization is significantly more prevalent in plaques exposed to RT than those not exposed to RT. This may provide a mechanistic explanation for the increased stroke risk in this patient cohort.
TRANSLATIONAL OUTLOOK: Currently it is unknown whether surgical excision of plaques on the basis of plaque composition, rather than stenosis severity, would improve patient outcomes. To translate contrast carotid ultrasonography into routine clinical practice, a randomized controlled trial is required to examine the hypothesis that surgical plaque excision on the basis of the presence or absence of “high-risk” plaque characteristics, such as IPN, improves outcomes compared with current practice.
The authors thank Dr. Dhrubo J. Rakhit for assistance with the interobserver agreement studies.
For supplemental videos and their legends, please see the online version of this article.
Dr. Shah was supported by the Northwick Park Cardiac Research Fund and the Cardiovascular Biomedical Research Unit, Royal Brompton Hospital and Imperial College London. Drs. Gujral, Harrington, and Nutting received support from the National Institute for Health Research Biomedical Research Centre at The Institute of Cancer Research, Royal Marsden Hospital. Dr. Shah has received travel grants from Bracco Diagnostics. Prof. Senior has received honoraria from Bracco Diagnostics. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- common carotid artery
- contrast-enhanced ultrasound
- cerebrovascular event(s)
- head and neck cancer
- intima-media thickness
- intraplaque neovascularization
- Received April 24, 2015.
- Revision received June 30, 2015.
- Accepted July 15, 2015.
- American College of Cardiology Foundation
- Finn A.V.,
- Nakano M.,
- Narula J.,
- Kolodgie F.D.,
- Virmani R.
- Michel J.B.,
- Virmani R.,
- Arbustini E.,
- Pasterkamp G.
- Virmani R.,
- Kolodgie F.D.,
- Burke A.P.,
- et al.
- Hellings W.E.,
- Peeters W.,
- Moll F.L.,
- et al.
- Fleiner M.,
- Kummer M.,
- Mirlacher M.,
- et al.
- Staub D.,
- Patel M.B.,
- Tibrewala A.,
- et al.
- Shah F.,
- Balan P.,
- Weinberg M.,
- et al.
- Coli S.,
- Magnoni M.,
- Sangiorgi G.,
- et al.
- Touboul P.J.,
- Hennerici M.G.,
- Meairs S.,
- et al.
- Plummer C.,
- Henderson R.D.,
- O’Sullivan J.D.,
- Read S.J.
- Dorresteijn L.D.,
- Kappelle A.C.,
- Boogerd W.,
- et al.
- Smith G.L.,
- Smith B.D.,
- Buchholz T.A.,
- et al.
- Bowers D.C.,
- McNeil D.E.,
- Liu Y.,
- et al.
- Ambrose J.A.,
- Tannenbaum M.A.,
- Alexopoulos D.,
- et al.
- Little W.C.,
- Constantinescu M.,
- Applegate R.J.,
- et al.
- Sonveaux P.,
- Brouet A.,
- Havaux X.,
- et al.
- Kozin S.V.,
- Duda D.G.,
- Munn L.L.,
- Jain R.K.