This study’s retrospective design was approved by the Institutional Review Board of St. Vincent’s Hospital. We retrospectively reviewed data from 2,585 patients who were diagnosed with thyroid cancer and treated at St. Vincent’s Hospital between February 1982 and December 2012. All patients were treated by total thyroidectomy or lobectomy with isthmectomy, and central lymph node dissection was performed when lymph node metastasis was suspected during the preoperative evaluation. The postoperative pathological reports confirmed that all patients had tumor-free margins. Thyroid-stimulating hormone suppression therapy was administered during the follow-up period, and nonstimulated serum thyroglobulin measurements and neck ultrasonography (with or without neck and chest computed tomography) were routinely performed every 6-12 months.

Among the eligible patients, we identified 1,659 patients with T1N0 PTC. However, to create a homogenous study population, we excluded patients with aggressive histological variants (solid, insular, columnar, tall cell, or diffuse sclerosing), lymphovascular invasion, multifocal malignancies, extrathyroidal extension, or capsular invasion, as well as those who had received postoperative radioactive iodine therapy. We also excluded patients who had a cystic tumor portion or poorly defined margins to ensure that the tumor volume could be accurately calculated. Thus, 1,081 patients were selected for inclusion in this study.

All 1,081 patients had undergone preoperative ultrasonography to examine the thyroid tumor. Three experienced radiologists, each with more than 5 years of thyroid ultrasound experience, had evaluated the tumor’s dimensions (using transverse and sagittal orthogonal scans) and cervical lymph node status. The recorded dimensions were transverse length (T), longitudinal length (L), and depth (D), which allowed us to calculate the tumor’s volume (V) using the ellipsoid formula (V=T [cm]×L [cm]×D [cm]×π/6).

The clinicopathological variables that we evaluated included age at diagnosis, sex, surgery type, histological subtype, and tumor location. All patients were categorized according to tumor diameter as T1a (≤ 1 cm) or T1b (> 1 cm but ≤ 2 cm). We next performed receiver operating characteristic (ROC) curve analysis to estimate the optimal cut-off tumor volume for predicting recurrence. Based on the result of the ROC analysis, we also categorized the patients according to tumor volume as V1a or V1b. To validate the results of the ROC analysis, an internal validation was performed using the 1,000 bootstrap resampling method.

The primary outcome of interest in the study was recurrence-free survival (RFS). RFS was defined as the period between the date of the initial surgery and the first instance of contralateral lobe, locoregional (the operative bed, trachea, strap muscle, or cervical lymph nodes), or distant recurrence or the last follow-up. The presence of a contralateral lobe or locoregional recurrence was based on cytological or histopathological evidence of disease in specimens obtained by fine-needle aspiration biopsy or surgical excision. We did not use serum thyroglobulin levels to identify recurrent disease, as they do not definitively diagnose recurrent or persistent disease in patients who have not undergone total thyroidectomy.

Independent t tests and chi-square tests were used to evaluate inter-group differences in the clinicopathological variables. Survival curves were created using the Kaplan-Meier method and compared using the log-rank test. Multivariate analyses were performed using Cox’s proportional hazard regression model to assess the independent prognostic values of tumor volume and tumor diameter, and hazard ratios and 95% confidence intervals (CIs) were estimated for each variable. Because age is a known prognostic factor in PTC, we also conducted subgroup analyses using a cut-off age of 45 years. All statistical tests were two-sided, and p-values of < 0.05 were considered to indicate significance. All statistical analyses were performed using the SPSS ver. 12.0 (SPSS Inc., Chicago, IL).

The ROC curve analysis (Fig. 1) revealed that the optimal cut-off volume for predicting tumor recurrence was 0.545 cm³, which provided an area under the curve of 0.712 (95% CI, 0.610 to 0.814; p < 0.001; sensitivity 50.0%; specificity 85.3%). Therefore, patients were categorized as having a tumor volume of V1a (≤ 0.545 cm³) or V1b (> 0.545 cm³). The cutoff value obtained from the ROC analysis was internally validated using the bootstrap method. The estimated concordance probability determined by 1,000 repetitions of the bootstrap sample corrections was 0.713 (95% CI, 0.602 to 0.806).

The baseline characteristics of the study cohorts are shown in Table 1. The mean age of the entire cohort was 48.14±10.74 years. Eighty-nine patients were moved from the T1b group to the V1a group (26 were aged < 45 years, and 63 were aged ≥ 45 years). Total thyroidectomy was more performed more often in the T1b group than in the T1a group (55.4% vs. 28.3%, respectively; p < 0.001). Moreover, comparison of groups V1b and V1a revealed significant differences in the number of right lobe tumors (55.0% vs. 53.2%, respectively; p=0.005), isthmus tumors (7.1% vs. 2.5%, respectively; p=0.005), and total thyroidectomies (58.6% vs. 30.4%, respectively; p < 0.001).

Age-based subgroup analyses (Table 2) showed that total thyroidectomy was significantly more frequent in < 45-yearold patients in group T1b than group T1a (44.8% vs. 20.6%, respectively; p < 0.001) and group V1b than group V1a (45.7% vs. 22.2%, respectively; p < 0.001). There were also significantly more histological variants in younger patients in group T1b than group T1a (9.4% vs. 3.4%, respectively; p=0.015). When we compared the ≥ 45-year-old patients in groups T1b and T1a, we found significant differences in the number of right lobe tumors (56.8% vs. 53.0%, respectively; p=0.043), isthmus tumors (5.6% vs. 2.2%, respectively; p=0.043), and total thyroidectomies (61.7% vs. 33.3%, respectively; p < 0.001). When we compared the ≥ 45-year-old patients in groups V1b and V1a, we found similar differences in the number of right lobe tumors (58.6% vs. 53.1%, respectively; p=0.002), isthmus tumors (8.1% vs. 2.1%, respectively; p=0.002), and total thyroidectomies (67.7% vs. 35.5%, respectively; p < 0.001).

The mean follow-up time was 66.12±28.75 months, and 97.2% of the entire cohort experienced RFS. Table 3 provides detailed information regarding tumor recurrence in the study population. Thirty patients experienced tumor recurrence, with 17 cases of contralateral lobe recurrence and 12 cases of locoregional recurrence. Only one patient experienced distant recurrence (lung metastasis), which was detected via chest computed tomography and confirmed by excisional biopsy. None of the patients who were moved from the T1b group to the V1a group experienced tumor recurrence.

There was no significant difference in RFS rates between the T1a and T1b groups (98.2% vs. 94.2%, respectively; p=0.997) or the V1a and V1b groups (98.4% vs. 91.1%, respectively; p=0.253) (Fig. 2A). Multivariate analysis provided similar findings (Table 4).

When we conducted subgroup analyses using a cut-off age of 45 years, no significant differences in RFS rates were observed between the < 45-year-old patients in either the T1a/T1b groups (98.2% vs. 95.8%, respectively; p=0.137) or the V1a/V1b groups (98.3% vs. 94.4%, respectively; p=0.236) (Fig. 2B and C). However, ≥ 45-year-old patients in the V1b group showed significantly poorer RFS rates than those in the V1a group (88.9% vs. 98.4%, respectively; p=0.029) (Fig. 2C). Multivariate analyses revealed no significant agerelated differences in RFS rates between the T1a and T1b groups. However, poorer RFS rates were associated with older patients in group V1b than V1a upon multivariate analysis (hazard ratio, 2.676; 95% CI, 1.075 to 6.661; p=0.034) (Table 4).