A total of 823 consecutive patients with PTC who underwent total thyroidectomy or lobectomy at Shandong Cancer Hospital & Institute between 2010 and 2015 were retrospectively reviewed in this study. Only patients with a first primary PTC were included, because their mechanism of cancer development may be different from patients with secondary PTC. The patients with no data of thyroid-stimulating hormone (TSH), serum thyroglobulin, total cholesterol, triglyceride concentrations, and/or fasting glucose level that were used as confounding variables before surgery will be excluded; finally, 796 subjects were eligible for analysis in this study. We reviewed the medical records and pathology reports of each patient to define initial clinic-pathological features, including primary tumor size, gross extra-thyroidal invasion, cervical LN metastasis, multifocality, vascular invasion, and advanced TNM stage after surgery and disease recurrences. Besides that, the generic information such as sex, age, weight, and height were recorded.

BMI was calculated using measurements of weight and height reported on the anesthesia summary as weight in kilograms divided by height in meters squared (kg/m²). Every patient will visit every 3 months during the first year after operation. We recorded BMI and other measurements for every visit, but we only used weight and height of the patients at their first visit. TSH and serum thyroglobulin was measured using an immunoradiometric assay. Total cholesterol, triglyceride concentrations, and fasting glucose were measured with an enzymatic colorimetric method and a nonenzymatic method, respectively.

During each operation, central compartment dissection was performed at the surgeons’ discretion. When lateral neck LN metastasis was suspected from the preoperative or intraoperative findings, lateral neck LN dissection was performed. The preoperative assessments for LN metastasis included a high-resolution ultrasound scan, fine-needle aspiration cytology (FNAC), and a computed tomography (CT) scan. The diagnosis, tumor characteristics, and/or the extent of disease were confirmed by cross-checking all pathology specimens. Tumor node metastasis (TNM) staging was based on the AJCC Cancer Staging Atlas (2012).^[19]

All patients were followed with a clinical examination including thyroid hormone and ultrasound scan of the neck every 3 months during the first year, and annually thereafter, stimulated serum thyroglobulin levels were evaluated at the same time. Tumor recurrence was defined as high-resolution ultrasound scan and then confirmed by cellular pathology of fine needle aspiration or recorrection in a patient who initially met the criteria for remission. All patients seen in the clinic were offered enrollment in the study and informed consent was obtained. Shandong Cancer Hospital & Institute approved this study.

We used logistic regression models to estimate odds ratios (ORs) with 95% confidence intervals (CIs) to analyze the relationship between the BMI and LN metastases. BMI is associated with an increased risk of common and less common malignancies with a 5 kg/m² increase in BMI.^[12] So, BMI with a 5 kg/m² increase was used as a continuous variable in the initial models. At the same time, the other clinicopathological features were analyzed. All of the following clinicopathological features of PTC patients were used as binary variables such as primary tumor size (≤1 or >1 cm), advanced TNM stage (stage I/II or stage III/IV), extrathyroidal invasion (yes or no), age (<45 or ≥45 years), and confounding variables (above reference range or no). All models were adjusted by TSH, serum thyroglobulin, total cholesterol, triglyceride concentrations, and/or fasting glucose level by the mechanism of forward. And, we also performed an analysis using 4 BMI groups in which the normal BMI group was used as a reference category. BMI groups defined by the WHO classification are as follows: underweight (<18.5 kg/m², n = 18), normal weight (18.5–24.9 kg/m², n = 403), overweight (25–29.9 kg/m², n = 311), and obese (>30 kg/m², n = 64). In the model, changes in BMI were defined as an independent variable, while changes in clinicopathological features including LN metastases were dependent variables. The numbers of LN metastases grouped by BMI category without normal distribution were tested using nonparametric test followed by the Mann–Whitney U test. Mean ± standard deviation was used to represent the quantitative parameters. Differences between categorical or continuous variables were analyzed using the Chi-squared test or Nonparametric test. Statistical differences among four groups were analyzed by ANOVA test. SPSS version 17.0 for Windows was used to conduct all statistical analyses. A P value < 0.05 was considered statistically significant.

The clinic-pathological characteristics of patients with papillary TC are summarized in Table 1. A total of 796 papillary TC patients (184 men and 612 women) were studied. The mean (SD) age was 46 (11.6) years (range, 12–79 years). The mean (SD) BMI was 25(3.40) kg/m² (range, 15.62–38.53 kg/m²) and the prevalence of underweight, overweight, and obese patients were 2.3%, 39.1%, and 8.0%, respectively. The mean (±SD) tumor size was 1.37 (±1.11) cm (range, 0.1–9 cm), and the papillary microcarcinoma was shown in 53.3% of tumors. Three hundred and eight patients (38.7%) of the tumors had extra-thyroidal invasion and 239 patients had nodular goiter and/or Hashimoto thyroiditis (HT) at the same time. The prevalence of central neck LN metastasis and lateral neck LN metastasis were 51.3% (n = 408) and 35.3% (n = 281), respectively; of these, 242 had both central cervical and lateral cervical LN metastasis, and the rest (39 cases) had skip metastasis, in which the lateral cervical LN had metastasis, while the central cervical LNs had no metastasis (Table 1). About 1.5% of patients had distant metastasis. Three hundred forty-eight patients (43.7%) had multifocal PTCs, and 255 patients (32%) showed advanced TNM stage (stage III or stage IV). Twenty-five patients (3.1%) had vascular invasion.

The 5-point increase in BMI was not statistically associated with cervical LN metastasis, as summarized in Table 2. But in the overweight group and the obese group, subjects had an increased OR for clinicopathological features of PTCs in central neck LN metastasis (OR = 1.09, CI: 0.78–1.53, P = 0.601 in overweight group, OR = 1.14, CI: 0.61–2.05, P = 0.605 in obese group) and lateral neck LN metastasis (OR = 1.35, CI: 0.98–1.84, P = 0.06 in overweight group, OR = 1.92, CI: 1.12–3.27, P = 0.017 in obese group) (Table 3). The sample size is quite small (only 18 patients) in underweight group, so we do not consider doing a statistical test. To further understand the unexpected relationship between BMI and LN metastases, the subjects were categorized into underweight, normal, overweight, and obese according to the WHO. The mean number of LNs removed was 4.09 in the central compartment and 10.88 in the lateral compartment. The mean number of positive nodes was 1.71 (range: 0–17) in the central compartment and 1.76 (range: 0–32) in the lateral compartment. The incidence of central and lateral LN metastases was 55.3% and 37.9% in overweight group, 55.9% and 45.3% in obese group, respectively. The mean number of positive central LNs and lateral nodes also increased with the increase of BMI when BMI ≥18.5 kg/m² (Table 4). It was not shown in underweight group. It is because the number of cases is too small in the underweight group. The incidence of central and lateral LN metastases was significantly higher in overweight group and obese group than in normal-weight group, but it was not significantly different between underweight group and normal-weight group. Statistical analysis failed to show any significant difference in the number of removed lateral nodes among the groups.

In addition, the 5-point increase in BMI was strongly associated with extra-thyroidal invasion (OR = 2.201, CI: 1.691–2.865, P < 0.001), primary tumor size larger than 1 cm (OR = 1.267, CI: 1.028–1.561, P = 0.027), advanced TNM staging (OR = 1.479, CI: 1.13–1.937, P = 0.004), and multifocality (OR = 1.31, CI: 1.066–1.61, P = 0.01) in multivariable-adjusted models (Table 2). However, there were no statistically significant results between BMI and, distant metastasis, vascular invasion and nodular goiter and/or HT. In our study, the sample size is quite small (only 13 patients) for cancer recurrence, so we do not consider doing a statistical test.