Predictive Value of Preoperative Computed Tomography in Sternotomy for Substernal Goiter

Da-Wei Zhao Southwest Hospital, Third Military Medical University (Army Medical University Tai-Hua Xiang Shanghai Hospital of Wanzhou District Yi-Zeng Sun University of Chinese Academy of Sciences Yao Li University of Chinese Academy of Sciences Xiang Cui Southwest Hospital, Third Military Medical University (Army Medical University) Yan Liang Southwest Hospital, Third Military Medical University (Army Medical University) Fan Zhang (  zhangfan316@163.com ) University of Chinese Academy of Sciences


Introduction
Substernal goiter (SG) was rst described by Haller in 1749 and rst surgically removed by Klein in 1982 1 . At present, the de nition of SG is not uniform. The various different criteria have been suggested in different literatures, including more than 50% of the volume of goiter located behind sternum 2 , or a thyroid gland extending 3 cm below the thoracic inlet, or extension of the gland below the fourth thoracic vertebra 1 . SG can compress and shift the large vessels, trachea and nerves in the neck, resulting in dyspnea and dysphagia, and even Honer syndrome due to compression of cervical sympathetic nerve or obstruction of superior vena cava re ux 3,4 . Surgery is an effective treatment for SG, regardless of sternotomy. Surgical strategy depends on the experience and comprehensive judgment of surgeon. Therefore, we collected and analyzed the preoperative computed tomography (CT) imaging parameters of SG patients to explore predictors for sternotomy.

Patients
The clinical and radiological data of all patients with substernal giant goiter were retrospectively obtained from our computerized database between January 2010 and February 2019, which was provided by the Breast Disease Center in the Southwest Hospital of the Army Medical University. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Committee of the Southwest Hospital of Army Medical University. Patients all granted written informed consent for surgery and the use of clinical records in this study.
In the 8-year interval from 2010 to 2019, a total of 37 patients with SG diagnosed by CT, including 8 males and 29 females with an average age of 54 ± 10 years (range from 33 to 70 years), were enrolled.
Giant substernal goiter was de ned as follows based on CT scan: (1) a goiter extending below the plane of the thoracic inlet; (2) a goitre with ≥ 50% of its mass located in the mediastinum; (3) a goiter with the maximum diameter exceeding 7 cm. For each patients, information on demographic data [age, sex, and body mass index (BMI)], medical history data, preoperative CT imaging data and pathological type of tumors were also collected. The patients with incomplete clinical data and CT data had been excluded.
Of these, 4 patients were converted to sternotomy because of the di culty in traditional cervical approach. Based on surgical approaches, these patients were divided into two groups, the nonsternotomy approach group (33 patients) and sternotomy approach group (4 patients).

CT Data Processing
All patients had CT plain and enhanced axial and coronal 3D reconstruction cervical CT images. Two certi ed thyroid surgeons with 7 years and 12 years experience in cervical CT analyzed tumor features with regard to location, number, size, and shape. The maximum length and width, as well as the maximum length behind the sternum (de ned as length sternum ) of tumors were measured on coronal or sagittal images. The area of each layer of tumors was measured on axial images. Then, according to the formula: volume = ∑ (area × thickness), the volume above sternum (de ned as V 1 ), the volume below sternum (de ned as V 2 ), and the total volume of tumor (de ned as Vt) was calculated. In cases of discrepancies in the interpretations of the two radiologists, a consensus was reached.

Statistical Analysis
Statistical analysis was performed using the SPSS statistical software package (version 25.0, Chicago, IL, USA). Quantitative variables were presented as mean ± standard deviation (SD), tested for normality and equality of variances using the Kolmogorov-Smirnov test, and compared using unpaired Student's t test or nonparametric test, whereas categorical variables were expressed as counts in percentages and compared using the chi-squared test or continuity correction test. We used logistic regression analysis to investigate the factors associated with sternotomy. Multi-collinearity was assessed by checking the Variance In ation Factor. Variables signi cant at P < 0.05 by univariate analysis were subjected to forward stepwise multivariate logistic regression model to identify independent predictor for sternotomy. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the diagnostic capability of CT parameters and to calculate the area under the ROC curve (AUC), optimal cut-off value, sensitivities and speci cities. All statistical tests of hypothesis were conducted at the 0.05 level of signi cance.

Characteristics of Patients with SG
Of total 37 patients with SG, there were 33 cases (89.2%,) in non-sternotomy group and 4 cases (10.8%) in sternotomy group. There were signi cant statistical differences in length, width, length sternum , V 2 and Vt (all P < 0.05) between the two groups of patients with SG. There were no signi cant differences in age, sex, BMI and pathological types (all P > 0.05) between the two groups. The baseline characteristics of the two groups of patients were shown in Table 1.

ROC curve analysis
ROC curve analysis suggested that the AUC in a decreasing order was V 2 > length sternum > length > width > Vt > V 1 , that is, V 2 (AUC: 0.976, 95% CI: 0.928-1.024, P ≤ 0.01) and the length sternum (AUC: 0.962, 95% CI: 0.896-1.028, P ≤ 0.01) were the most valuable factor in judging whether sternotomy was warranted ( Fig. 1). The cut-off value for V 2 of 68.96 cm 3 provided the optimal sensitivity and speci city of 100% and 93.6%, respectively. The cut off values of length sternum was 46.7 mm with the sensitivity of 100% and speci city of 87.9%, respectively (Table 3).

Discussion
Substernal goiter accounts for 1%-20% of patients undergoing thyroidectomy. Most of the patients are middle-aged women, and it is four times more common in women than in men 5 . Due to the speci c location of tumor, some patients have no obvious symptoms in the early stage. With the enlargement of the mass, the tumor is limited by the surrounding bone structure and tends to oppress the adjacent important organs and structures, thus presenting various clinical symptoms, such as dyspnea mainly due to the involvement of the trachea, dysphagia resulting from the involvement of the esophagus, and hoarseness caused by compression of the recurrent laryngeal nerve 6 .
Surgical resection is an effective method for the treatment of SG [4][5][6][7] . At present, there is no uniform criteria to evaluate sternotomy. Because of the limitations of cervical ultrasound for the proper assessment of SG, preoperative CT examination is recommended for evaluating the extension of the thyroid, the presence of tracheal compression, to understand the relationship with surrounding organs, nerve and blood vessels, and also to develop the surgical strategy 5,8 . For patients with tumor extending into deep sternum, or intimately adhered to surrounding tissues, and those who have huge tumor mass or require a secondary surgery, to ensure the safety of the surgery, it is necessary to make routine preparations for thoracotomy.
Some studies have shown some related indicators of preoperative CT imaging in the prediction of surgical methods for substernal goiter. CT features, such as 70% of the volume of the thyroid mass located behind the sternum 9 , and the mass exceeding the aortic arch or tracheal carina 8 are the main indications for sternotomy. Malignant nodules, mass extending into the posterior mediastinum and the presence of ectopic solitary goiter were also the CT imaging indications of sternotomy 8,10−12 . In addition, reoperation for recurrent SG, non-recurrent laryngeal nerve, and abnormal vascular variation are also factors to be considered in thoracotomy.
Our study showed that the length, width, length behind sternum, volume below sternum, and total volume of tumor in patients with sternotomy approach were signi cantly greater than those with non-sternotomy approach. The length behind sternum was an independent risk factor for sternotomy. ROC analysis found that the volume behind sternum and the length behind sternum were the most valuable factor in judging whether sternotomy was performed. Our ndings provide a speci c reference standard for clinical decision-making based on the comprehensive judgment of surgeon, so as to develop a more bene cial surgical strategy for patients.
We admit that several limitations still exist in our study. First of all, due to the relatively small number of patients in this study, our conclusion requires further veri cation in large studies. Secondly, there is a potential for unintentional bias because of the relatively small number of subjects. Thirdly, the limitations of retrospective and single center study design would lead to a potential information bias of our study.

Conclusion
CT examination plays an important role in developing surgery strategy for SG. The length behind sternum of tumor is a convenient and independent predictor of sternotomy for SG. However, due to the relatively small number of subjects in this study, further larger sample clinical studies are still needed for veri cation.

Figure 1
Receiver operating characteristic curve of CT parameters of tumor.