Patient and demographic data recruitment
From January 1, 2013 to December 31, 2014, HNC patients who had ever completed RT and received regular follow-up at radio-oncology and neurology departments of Linkou Chang Gung Memorial Hospital were initially prospectively recruited, but the data were retrospectively reviewed. In this study, we aimed to detect the impacts of hypothyroidism in the early phase of post-RT vasculopathy, patients with previous stroke or >50% CAS at enrolment were assumed to be vulnerable to vascular events were excluded in this study to improve the homogeneity of cerebrovascular risks of the study population. Demographic data and details of risk factors for common stroke, such as dyslipidemia, hypertension, diabetes mellitus, heart disease, and cigarette smoking, were obtained from all the recruited patients. Laboratory data at enrolment, including glycated hemoglobin and low-density lipoprotein cholesterol, were recorded. Furthermore, medication records, particularly on the use of antiplatelets, statins, or thyroxine, were collected. In addition, all patients received regular thyroid function monitoring at enrolment, including free T4 and thyroid stimulating hormone (TSH) (Figure 1). We obtained written informed consent from all the participants. The study was approved by the Ethics Institutional Review Board of Chang Gung Memorial Hospital (No. 100-4153B).
Cancer and RT data
The pathological types and locations of HNC, the extent of lymph node involvement, and cancer staging were recorded. All the patients received photon-beam intensity-modulated radiation therapy (IMRT). IMRT is inherently limited by the physical properties of the photon beam, resulting in unavoidable irradiation of normal tissues at low to moderate doses even at substantial distances from the tumor . In the present study, a minimum distance of 5 mm around the clinical target volume was required to define each planning target volume. Treatment consisted of 70 Grays in 33 fractions over 6 weeks and 3 days; 5 fractions were delivered per week. All targets were treated simultaneously. Total treatment times that were 5 days longer than the scheduled treatment period were considered a major violation . Information on the accumulated total doses of RT and the time interval from the latest RT to study enrolment was ascertained from all the enrolled patients.
Carotid duplex ultrasound studies
We used Philips HDI 5000 (Wesley Hills, NY, USA) or Acuson Sequoia (Siemens, Munich, Germany) 5–10 MHz real-time B-mode imaging system and a 3.0 MHz pulsed-wave color Doppler spectrum analyzer to follow up the carotid and vertebral arteries after RT. B-mode examinations in the sagittal (anterior–posterior, posterior–anterior, and lateral) and transverse views of the extracranial carotid arteries were used to detect stenotic features. The degree of CAS was defined according to the standard ultrasound criteria . Personnel from our carotid duplex ultrasound (CDU) laboratory diagnosed CAS with an overall accuracy of >90% . As total plaque score (TPS) is a known predictor of CAS after RT , we also assessed the presence and severity of plaques in each CDU study . In each side, we measured 5 segments, namely the proximal common carotid artery, distal common carotid artery, carotid bifurcation, internal carotid artery, and external carotid artery. In total, we assessed 10 segments bilaterally. The grading scores of the plaques at each site were defined as follows: Grade 0, normal or no plaques; Grade 1, all plaques occupying <30% of the vessel diameter; Grade 2, at least 1 plaque occupying 30%–49% of the vessel diameter; Grade 3, at least 1 plaque occupying 50%–69% of the vessel diameter; Grade 4, at least 1 plaque occupying 70%–99% of the vessel diameter; and Grade 5, total occlusion of the vessel. The TPS for each patient was defined as the sum of the plaque scores obtained from the 5 arterial segments in both carotid arteries .
Patients were grouped based on their thyroid function at enrolment. Patients with normal TSH (≤5 mIU/L) and free T4 level were categorized in the euthyroid (EU) group. Patients with elevated TSH (>5 mIU/L), including overt HT and subclinical HT, or patients who had already received thyroxine treatment for HT at enrolment were categorized into the HT group.
Follow-up strategy and outcomes
All the patients received serial thyroid function (TSH and free-T4) follow-up every 6 months. Furthermore, serial CDU studies were performed annually to monitor CAS progression. The main outcomes of interest in this study were CAS progression and upcoming ischemic stroke (IS). We defined CAS progression as >50% stenosis on the B-mode or peak systolic velocities ≥120 cm/s based on the hemodynamic criteria at any internal carotid artery or common carotid artery in the follow-up CDU study. The presence of upcoming IS was confirmed by clinical stroke symptoms with compatible brain image findings.
Power analysis was performed to estimate the participant numbers in each group. Alpha error with 0.05, statistical power with 0.9, and allocation ratio with 1 were set up. The estimated proportions of stroke events of the HT and EU groups between 18-64 years old were set up with 3.3% and 2.4% according to previous study . The estimated sample size was 90 in each group. We used SPSS 22.0 (SPSS, Chicago, IL, USA) to analyze clinical data. We used the Kolmogorov–Smirnov test to examine normality. Parameters were presented as means ± standard deviation or n (%). We used an independent two-sample t test to examine differences in continuous data between the study groups. In addition, categorical variables were compared using the chi-square test or Fisher’s exact test. Both event risk and time to event (CAS progression or IS) between the study groups were compared using a multivariable Cox proportional hazards model with adjustments for the selected variables (total doses of RT, time interval from the latest RT, antiplatelet use, age, gender, and smoking) that might confound outcomes. Furthermore, the adjusted curves of time to event for each group were depicted using the multivariable Cox model. Moreover, time to CAS progression and IS were further analyzed during the subgrouping of patients in the HT group based on the use or nonuse of a thyroxine supplement. Statistical significance was set at p < 0.05.