Post-RT NPC patients treated with intensity-modulated radiation therapy were recruited in the Department of Clinical Oncology of Queen Mary Hospital when the patients attended the follow up clinics. The inclusion criteria of post-RT NPC patients were Chinese NPC patients, aged 18 years or above, received a single course of RT for the primary tumour and neck lymph nodes, and a post-RT duration of 4 years or more. The exclusion criteria were a known history of leukopenia, thrombocytopenia, severe hepatic or renal dysfunction, an evidence for inflammatory or other malignant diseases, and a known history of carotid atherosclerosis prior to RT, previous carotid endarterectomy or stenting.
Patients with cardiovascular risk factor (CVRF) were recruited by poster advertisement in the Hong Kong Polytechnic University. The inclusion criteria of CVRF subjects were Chinese, older than 40 years, and have at least one CVRF, namely smoking, diabetes mellitus, hypertension, hypercholesterolemia or heart disease. The criteria for identifying these CVRF were the same as described in our previous study(14). The exclusion criteria for the recruitment of patients were previous RT, carotid endarterectomy or stenting.
Study design and ultrasound examination
The Human Subject Ethics Subcommittee of the Hong Kong Polytechnic University (HSEARS20160930001) and the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster approved this study (HA RE001F3). Each participant was provided an information sheet with detailed information of the study and the rights of participants. Informed written consent form was obtained from all participants before the commencement of the examinations. Medical history of the post-RT NPC patients including post-RT duration, history of chemotherapy and atherosclerotic diseases was obtained from archived clinical records of the patients. Individual face-to-face interview was conducted with post-RT NPC patients and CVRF patients to obtain the demographical information and history of CVRF.
All recruited participants had a carotid ultrasound examination. All ultrasound examinations were performed using the Esaote MyLab™ Twice eHD CrystaLine ultrasound unit in conjunction with a 3–13 MHz linear transducer (Esaote, Genoa, Italy). In order to ensure measurements were obtained with the subjects at resting state, all subjects were allowed to rest for at least 10 minutes before the ultrasound examination. In the carotid ultrasound examination, the subject lay supine on the examination couch with the shoulders and neck supported by a pillow so that the neck was slightly extended and the head turned away from the side under examination. On each side of the neck, the CIMT was evaluated using the automated quantification programs of the ultrasound unit: radiofrequency-based quality intima-media thickness (RF-QIMT).
For each subject, the presence or absence of carotid plaque in the internal, external and common carotid arteries was assessed. Carotid plaque was identified as focal arterial wall thickening > 50% of the adjacent intima-media layer(15). All identified carotid plaque of each subject were included in the study, and the characteristics of each carotid plaque were evaluated for grey-scale median (GSM) and detailed plaque texture analysis (DPTA).
In the evaluation of GSM and DPTA, multiple longitudinal grey scale ultrasound images of the carotid plaque were obtained. Archived digital images were reviewed and analysed using image-processing software (Adobe Photoshop CS v.8.0, Adobe, San Jose, CA, United States; and Image Pro Plus v.6.0, Media Cybernetics, Rockville, MD, United States).
For each carotid plaque, five longitudinal sonograms which clearly demonstrating the borders and internal echotexture of the plaque were selected to evaluate the GSM and DPTA. Image normalization was performed before the evaluations. With the use of the histogram facility of the software (Adobe Photoshop CS, v.8.0, Adobe, San Jose, CA, United States), the grey scale value of two reference areas (blood and adventitia) in the image were standardised 0 for blood and 190 for adventitia. All the pixels of the image were adjusted accordingly based on the standard linear scale, and a normalized image was produced.
For the evaluation of GSM, the carotid plaque was outlined manually on the normalized grey scale ultrasound image using software Adobe Photoshop CS, v.8.0 (Adobe, San Jose, CA, United States) and then the software calculated the median of the grey scale value of pixels (i.e. GSM) within the region of interest (ROI).
For the DPTA, the software Image Pro Plus v.6.0 (Media Cybernetics, Rockville, MD, United States) was used to analyse the normalized grey scale ultrasound images. In this software, 4-connect pixels consist in an “object”, and numbers of objects with defined grey scale ranges consist in a component. In the image analysis, the different components of carotid plaque were color-coded by the software based on the different grey scale ranges: yellow for blood (grey scale ranges 0–9), orange for lipid (10–31), red for muscle (32–74), light blue for fibrous tissue (75–111) and dark blue for calcium (112–255). The blood, lipid and muscle appeared hypoechoic, while fibrous tissue and calcification were hyperechoic(16). For each carotid plaque component, the average pixel density, area percentage and integrated optical density (IOD) were evaluated. Average pixel density was the mean value of pixel densities of each component. Area percentage was the percentage that the area of each component relative to the area of the plaque. IOD was the product of area and average pixel density, and it represented the integration of pixel density of the plaque component.
Basic information of the subjects and plaque characteristics were expressed as mean ± SD or SEM for the continuous data; counts and percentage were presented for categorical data. Normal distribution of the data was evaluated by Shapiro-Wilk test. The difference between post-RT NPC group and CVRF group were evaluated using Mann Whitney U tests for nonparametric variables (plaque number per subject, CIMT and GSM), unpaired Student t-test for parametric variables (age), and chi-square tests for categorical data (CVRFs, number of CVRFs, plaque presence). Comparisons among groups were evaluated by one-way ANOVA. The effects of age, gender, number of CVRFs and exposure to radiation on carotid plaque characteristics were analysed by multiple linear regression models. The risk factors regarded as candidate variables were age, gender, number of CVRFs and radiation. We used multivariable model to analyse the effects of different risk factors on plaque characteristics (CIMT, plaque presence, number of plaques per subject, plaque components). All statistical analyses were performed using SPSS 22 (IBM Corporation, Armonk, NY, United States) and Prism 7.0 (GraphPad Software, San Diego, CA, United States). P value < 0.05 was considered to be significant.