We retrospected the electronic medical records of DTC patients from July 2014 to June 2017 at the Department of Nuclear Medicine at Zhujiang Hospital, Southern Medical University, after getting of approval by the local Institutional Review Board.

The patients who underwent cervical LN dissection, near-total/total thyroidectomy and then conduct the first RAI ablation in our hospital (17) were ruled out if they meet the following standards (1): patients had distant metastases upon surgery or in initial RAI ablation, (2) patients had missing data or incomplete data on size of LNM, (3) patients had previous RAI ablation in other hospitals, (4) patients had positive thyroglobulin antibody (TgAb) (> 115 IU/mL), (5) patient follow-up time was < 24 months after the first RAI ablation ( Figure 1 ).

The data was reviewed including patient demographics, surgery, and pathology reports. The pathologic TNM classification was based on the American Joint Commission on Cancer (AJCC), 8th edition.

The size of LNM is defined as a focus full of thyroid cancer metastases in the LNM with the largest dimension observed in multiple serial sections using microscope. The pathology slides were analyzed and measured the size of metastatic foci in each LN by An experienced pathologist (7).

Roche Cobase 801 electrochemiluminescence instrument was utilized for measurement of Tg (measuring range: 0.04–5000 μg/L). All ps-Tg was measured underlevothyroxine withdrawal just earlier RAI ablation in the same laboratory in our hospital.

Before RAI ablation, patients were informed to obey a low-iodine dietary for 3–4 weeks and levothyroxine withdrawal for 3 weeks; this was prolonged for 1 week if TSH levels did not reach 30 μIU/mL. An empiric RAI administered activity was used to decide the activity of 131I, using 1.85–3.70 GBq for thyroid remnant ablation and 3.70–7.40 GBq for postoperative residual cervical LNM.

After the initial RAI ablation, follow up every 3-6 months for the first 2 years. When achieving stable disease, follow up every 6-12 months. Patients were regularly followed-up with via measurement of Tg, thyroglobulin antibody (TgAb), and neck ultrasound during thyroid hormone therapy, using the 2009 ATA guidelines (18). Additional imaging methods including 18F-FDG PET/CT, 131I whole-body scans(WBS), chest and mediastinal computed tomography were performed when suppressed Tg (rTG) was >1 μg/L or when suspicious loco-regional/distant metastases were detected.

Incomplete biochemical or structural response within one year of the initial RAI ablation was considered persistence disease. Meanwhile, patients who were earlier considered as having significant effect to therapy was defined as recurrence disease, when they show functional, structural, or biochemical evidence of disease. For the above patients, further treatment plans including repeated surgery, extra RAI therapy, and other therapies were taken by the attending physician.

A total of data locked in June 2022 were utilized to determine clinical outcomes. At final follow-up, patients were considered to have long-term remission if they had a rTG < 1 μg/L, no observable TgAb, and no structurally identifiable evidence. Patients were considered as having persistent/recurrent disease if they had a rTG > 1 μg/L, stimulated Tg > 10 μg/L, or any evidence of disease on functional imaging (18F-FDG PET/CT, 131I WBS), structural imaging (ultrasonic, CT, or MRI), or biopsy-proven disease. Patients with persistent/recurrent disease were classed as having either structural evidence of disease on imaging or biochemical evidence of disease (rising TgAb, rTG > 1 μg/L, or stimulated Tg > 10 μg/L without a structural evidence of disease). And patients with structural evidence of disease were fulfilled the undermentioned standards (1): highly suspicious focus on the neck ultrasound (short axis > 5 mm, number, rounded shape, irregular margins, loss of the fatty hilum, enlarged cortex, heterogeneous parenchymal echotexture, hypoechoic parenchymal echogenicity, calcifications, cystic change or necrosis, disorganized peripheral vascularity during follow-up), (2) positive cytology/histology, or (3) findings on 18F-FDG PET/CT, 131I WBS, or other imaging highly suspicious of metastatic evidence (11, 13, 16, 19–21). Disease free survival (DFS) was defined as the period of time among the discovery of persistent/recurrent disease and first operation.

Statistical analysis was performed using SPSS software (version 26.0). Categorical data were presented as numbers and percentages. Qualitative parameters were analyzed using chi square test or Fisher’s exact test and are expressed as frequencies and percentages. Continuous data were presented as the mean ± SD or the median (range). A P value < 0.05 was considered statistically significant. The X-tile software was used to acquire the optimum cut-off for the size of LNM to predict persistent/recurrent disease (22). Multivariate analysis was done for variables with a P value < 0.05 in the univariate analysis to determine the statistical risk factors associated with poor prognosis. Curves of DFS were fabricated using the Kaplan-Meier method, and log-rank tests were used to assess the differences in DFS among the risk stratification. Missing data were handled using complete case analysis.

Of the 1,452 consecutive DTC patients who underwent near-total/total thyroidectomy administered in our department during the study, 77 patients were excluded because they did not undergo neck LN dissection, 344 patients were excluded because they had <24 months of follow up, 121 patients were excluded because of distant metastases upon surgery or initial RAI ablation, and 43 patients were excluded because they had received previous 131I treatment at other hospitals. In addition, 150 patients were excluded due to missing data or incomplete data on size of LNM (missing pathology reports: N = 32, incomplete data on size of LNM: N = 118), and 174 patients were ruled out due to the positive TgAb (>115 IU/mL). Finally, 543 patients with DTC were involved in the final cohort ( Figure 1 ).

The median of the primary tumor size was 1.5 cm (range: 0–9.0 cm). Patients with positive LN was 494 (91.0%), with a median number of positive LNs of 5 (range: 0–79). The median size of LNM was 0.4 cm (range: 0–5.0 cm). The median 131I-administered activity was 5.55 GBq (range: 1.85–22.94 GBq). The number of patients with multiple RAI ablation was 101 (18.6%) ( Table 1 ).

The median of follow-up was 63 months (range: 24–122 months). Before initial RAI ablation, 356 patients had no functional or structural evidence of disease and 48 (13.5%) patients had poor prognosis at final follow-up. In addition, 71 patients had highly suspected structural evidence of disease upon ultrasonic examination, and 43 (60.6%) patients had persistent/recurrent disease. Functional evidence of disease was found in 83 patients upon 131I post-therapy WBS, only 16 (19.3%) patients had persistent/recurrent disease, 33 patients had structural and functional evidence of disease, and 22 (66.7%) patients had persistent/recurrent disease.

Further, 442 patients taken one RAI ablation, and 57 (12.9%) patients had poor prognosis. Ninety-two patients taken two RAI ablations, 62 (70.7%) patients had poor prognosis, 9 patients taken three RAI ablations, and 7 (77.8%) patients had poor prognosis.

Generally, long-term remission was obtained in 414 patients (76.2%). In total 129 patients (23.8%) had poor prognosis, 106 patients (19.6%) had a structural incomplete response, and 23 patients (4.2%) had a biochemical incomplete response. Twenty-one patients with positive cytology/histology underwent repeated surgery, 7 patients (33.3%) had a significant effect, 6 patients (28.6%) changed from structural incomplete response to biochemical incomplete response, 7 patients (33.3%) maintained structural incomplete response, and 1 patient (4.8%) had distant metastases. There were no death associated with DTC during study.

The median value of the size of LNM was 0.4 cm (range, 0–5.0 cm), with a median value of 0.3 cm in patients with long-term remission and 1.1 cm in patients with poor prognosis (P < 0.001) ( Figure 2 ). The median of time interval among first operation and initial RAI ablation was 2.0 months (range, 0–103 months).

The optimum cut-off value for the size of LNM was decided using the X-tile software (version 3.6.1, Yale University), and the result is shown in Figure 3A . All patients were divided into 3 risk stratification for the assessment of long-term remission. There were 282 patients with the size of LNM ≤ 0.4 cm (low-risk group), 199 patients with the size of LNM ranging from 0.4 cm to 1.4 cm (intermediate-risk group), and 62 patients with the size of LNM > 1.4 cm (high-risk group) ( Figure 3B ). Curves of DFS were shown in the Kaplan-Meier method, the low-risk group had 92.2% DFS rates, intermediate-risk group had 67.3% DFS rates, high-risk group had 32.3% DFS rates. The hazard ratio (95% CI) of the high-risk patients and intermediate-risk patients were 13.653 (8.135–22.913) and 4.674 (2.881–7.583), respectively. The low-risk patients was used as the reference ( Figure 3C ).

During the univariate analyses, various risk factors were identified to be significantly related to predictors of long-term remission ( Table 2 ). Meanwhile, these risk factors were further included in multivariate analysis. Cumulative 131I-administered activity, the size of LNM and ps-Tg were independent factors in predicting clinical outcome ( Table 3 ).

Recently, some studies report that the ps-Tg cut-off was 10.1 μg/L (13, 16). When we combined the size of LNM with ps-Tg ( Figure 4 ), we found that patients with a ps-Tg ≤ 10.1 μg/L had a considerably increased DFS rates (low-risk = 95.4%, intermediate-risk = 87.0%, and high-risk = 57.7%). In contrast, patients with ps-Tg > 10.1 μg/L had a considerably reduced DFS rates in each risk group (92.2–81.0% in low-risk patients, 67.3%–40.5% in intermediate risk patients, and 32.3%–13.9% in high-risk patients) ( Figure 4 ).