Comparison of Significant Carotid Stenosis for Nasopharyngeal Carcinoma between Intensity-Modulated Radiotherapy and Conventional Two-Dimensional Radiotherapy

Radiotherapy (RT) serves as the most efficient treatment for nasopharyngeal carcinoma (NPC) and can cause carotid stenosis. This work compared the incidence of significant carotid stenosis between intensity-modulated radiotherapy (IMRT) and two-dimensional conventional radiotherapy (2D-RT) for NPC and explored the risk factors. We retrospectively reviewed 233 cases with NPC who underwent carotid ultrasound post IMRT or 2D-RT from 2006 to 2015. The incidence of significant stenosis after RT was 19.3%. Significant stenosis was identified in 20 (14.6%) of 137 patients treated with IMRT and 25 (26.0%) of 96 patients with 2D-RT, respectively (p = 0.035). Multivariate logistic analysis indicated age (odds ratio = 1.054, 95% CI = 1.011–1.099, p = 0.014), radiation technique (IMRT) (odds ratio = 0.471, 95%CI = 0.241–0.919, p = 0.027) and time interval (odds ratio = 1.068, 95%CI = 1.033–1.105, p = 0.001) as independent predictors for significant carotid stenosis. Our study suggests that IMRT was associated with decreased incidence of significant carotid stenosis versus 2D-RT for NPC. Prevention and carotid ultrasound should be considered for older NPC survivors with longer interval from RT, especially those treated with 2D-RT.


Incidence of carotid stenosis (≥50%) after RT and risk factors. Significant carotid stenosis (≥50%)
was identified in 45 (19.3%) of 233 patients at a median interval of 76 months post RT. A representative carotid ultrasound color duplex scan is displayed in Fig. 1.

Comparison between IMRT and 2D-RT. Demographic information and clinical characteristics of both
2D-RT and IMRT groups are shown in Table 1. There was no significant difference between age, gender, and atherosclerosis hazard factors including age, smoking, hypertension, diabetes, vascular disease, and concurrent chemotherapy. Significant stenosis was identified in 20 of 137 patients treated with IMRT and 25 of 96 patients with 2D-RT, respectively (p = 0.035) ( Table 2). Stenosis after RT was more commonly seen in 2D-RT (26.0%) versus IMRT (14.6%). However, there was no distinction between time interval of the 2D-RT group (median 77, range 59 to 97 months) and the IMRT group (median 72, range 61 to 93 months) (p = 0.204). Furthermore, the distribution of the stenosis within the carotid was analyzed in terms of common carotid artery (CCA), internal carotid artery (ICA) and carotid bulb. External carotid artery (ECA) was not included since insufficient information was given from most of the ultrasound scan report. We found that the incidence of significant carotid artery, internal carotid artery and carotid bulb stenosis after IMRT (11.0%, 8.8%, 8.1%, respectively) were all lower than 2D-RT (21.9%, 17.8%, 17.8%, respectively) ( Table 4).

Discussion
The literature has demonstrated that RT might cause carotid stenosis 8,12,13 . As an alternative, IMRT is being widely adopted for treating NPC patients 14,15 . Here we showed that 14.6% of the patients treated with IMRT developed carotid stenosis, whereas the incidence rate was 26.0% for patients treated with 2D-RT, and 19.3% for all patients. For comparison, in a population-wide screening of 22,636 asymptomatic individuals 16 , the prevalence of carotid stenosis was 4.2%. A study of head and neck cancer, excluding symptomatic patients, found that the 4 year-incidence was 14% 14 . In another study, Lam et al. reported an incidence of 30% at 9 years after treatment in 71 cases 17 . Similar to many such studies published before, they included patients with cerebrovascular disease and carotid stenosis before RT 18 .
One study evaluated carotid stenosis of NPC patients by using contrast-enhanced MR angiography scanning. They found a rate of 37.5%, which is higher than our finding 19 . The difference may be attributed to different measurement methods as was discussed by the authors. Another study identified carotid stenosis (≥50%) in 20.9% of 129 NPC patients who underwent 2D-RT, similar with our result for all patients 20 . However, their published incidence rate was lower than ours for the 2D-RT group and higher for the IMRT group. Whether the difference is significant remains to be explored.
The present study illuminated that an older age might relate to a higher incidence of stenosis (≥50%) in NPC patients after RT treatment, consistent with the results of Zhou et al. 19 . They also found that the risk rose with time after RT. Post-radiation duration was judged to be the main cause of radiation-induced carotid stenosis. Previous studies evaluated intervals ranging from several months to over 20 years 21 . This indicates that the effect of irradiation was lasting; with the extension of time, the injurious effect on arteries would be more pronounced 8 . We found a difference of time interval in the significant stenosis group contrasted with the non-significant stenosis group, and this was statistically significant by multiple logistic regression analysis.
The hazard factors of stroke (smoking, diabetes, hypertension, vascular disease, atrial fibrillation, and carotid stenosis) have been widely demonstrated to give rise to carotid stenosis 22 . In our study, there was no significant difference between stenosis patients (≥50%) and non-stenosis patients (<50%). Even so, management of these risk factors is beneficial for NPC patients post-RT since it works for the general population 23 .
Our results suggest that IMRT significantly reduced the incidence of carotid stenosis after RT versus 2D-RT. Furthermore, CCA, ICA and carotid bulb in the IMRT group all showed a lower rate of significant stenosis than that in 2D-RT group. Lower incidence might be explained by the fact that the beams used by IMRT are composed of segments of different intensities 24 , which can provide more accurate dose distribution and save the surrounding tissues 25 .
Our results also suggest an RT dose effect for cerebrovascular events 26 . However, it is not feasible to eliminate dose to the carotids in clinical practice, even with IMRT, because the carotid arteries are often involved in the carotid nodal targeted volumes 27 .
No distinction was found for time interval from RT to the first color ultrasound duplex-detected significant carotid stenosis between patients treated with IMRT and those with 2D-RT. Interestingly, it has been reported that patients treated with 2D-RT had a longer time to develop radiation-induced temporal lobe injury compared with those treated with IMRT 9 . From our perspective, the time interval would be affected inevitably by the frequency and time of ultrasound examination. Since only a small proportion of patients with carotid stenosis will develop symptoms, most of them will not be diagnosed until the stenosis is detected by ultrasound or other examination during follow-up. Early detection of this complication can result in a better outcome. In our opinion, carotid color ultrasound should be a clinical routine during follow-ups for NPC patients after RT 28 .
The exact mechanism for carotid stenosis is not clear 29 . Injury to microvasculature may be the most important mechanism for injury in the large arteries 7 . Further research needs to be conducted in this area.

Materials and Methods
Patients. Our study reviewed all patients with NPC who underwent carotid ultrasound after 2D-RT or IMRT from 2006 to 2015 at Sun Yat-sen Memorial Hospital. This work was approved by the ethics committee of Sun Yat-sen Memorial Hospital and performed in accordance with the approved guidelines. All the participants provided informed consents. The exclusion criteria were history of transient ischemic attack (TIA), cerebrovascular disease, carotid stenosis, metastatic disease, and previous neck RT at diagnosis. No patients had NPC recurrence or a second primary cancer; those who were treated with palliative care, underwent carotid endarterectomy, or previous stenting were also excluded. The schedules of radiotherapy were collected from archived clinical records.   Baseline demographic and comorbidity disease, including gender, age, smoking, and hypertension, diabetes, vascular disease, and concurrent chemotherapy were also included.

Ultrasound technique.
A carotid color-flow duplex scan was used with all patients. The study employed a color-coded duplex ultrasonograph with ATL HDI 3000 (Bothell, Wash), which combines a real-time B-mode image (5-10 MHz) that is used to acquire sagittal (anterior-posterior, posterior-anterior, lateral) and transverse views of the extracranial carotid system, with a pulsed-wave color Doppler flowmetry (3.0 MHz).
Vascular stenosis diagnostic criteria. Carotid stenosis was diagnosed according to velocities (peak systolic, end diastolic) as well as artery ratios (internal carotid artery to common carotid artery). The reduction of 50% or greater in luminal diameter on either/both sides of the neck was regarded as significant stenosis based on previous publication 30 . The time interval was defined as interval from completion of RT to detection of carotid stenosis, or to the last ultrasound scan if no carotid stenosis was detected.
Statistical analysis. Our data were expressed as mean ± SD (standard deviation) unless noted. Comparisons between groups were performed using t-test or x 2 test. Multivariate logistic regression was adopted to estimate risk factors for carotid stenosis. SPSS 20.0 (SPSS Inc., IL, USA) was applied to perform statistical analysis. We considered p < 0.05 in two-tailed tests as significant.

Conclusions
We believe that the present study was the first to make a comparison between IMRT and 2D-RT in terms of carotid stenosis in NPC patients. Our study revealed that older age was an independent hazard factor for significant carotid stenosis in irradiated NPC survivors, whereas IMRT was associated with decreased incidence of significant stenosis versus 2D-RT for NPC. Therefore, prevention and carotid ultrasound examination should be considered for older NPC survivors with longer interval from RT, especially those treated with 2D-RT.

Data Availability
Due to ethical and legal restrictions, the data supporting findings presented in this manuscript are available from the corresponding author upon request.