As summarized in (Table 1), 343 women (69.6%) and 150 men (30.4%) (female: male was 2.3:1) were included in the study, with a mean age of 39.06 ± 10.63 years (rang, 16 to 72) at diagnosis of PTC and 449 (91.1%) were younger than 55 years. The BRAFV600E positivity rate among these patients was 80.7%. The mean tumor size was 13.26 ± 9.08 mm and 237 (48.1%) patients were with the tumor size ≤ 10 mm. In addition, HT, multifocality, bilateral tumor, capsular invasion, and ETE were found in 176 (35.7%), 84 (17.0%), 114 (23.1), 208 (42.2%), 138 (28.0%) patients respectively.
Table 1
Demographics and clinicopathological characteristics of 493 papillary thyroid carcinoma patients
Characteristic | Number (%) |
No. of patients | 493 (100) |
Sex | |
Male / Female | 150 (30.4) / 343 (69.6) |
Age (mean ± SD, years) | 39.06 ± 10.63 |
≥ 55 / < 55 | 44 (8.9) / 449 (91.1) |
Tumor size (mean ± SD, mm) | 13.26 ± 9.08 |
≤ 10 / > 10 | 237 (48.1) / 256 (51.9) |
Hashimoto's thyroiditis | |
Yes / No | 176 (35.7) / 317 (64.3) |
BRAFV600E mutation | |
Positive / Negative | 398 (80.7) / 95 (19.3) |
Multifocality | |
Yes / No | 84 (17.0) / 409 (83.0) |
Bilateral tumor | |
Yes / No | 114 (23.1) / 379 (76.9) |
Location of right tumor | |
Superior lobe | 112 (22.7) |
Middle lobe | 253 (51.3) |
Inferior lobe | 128 (26.0) |
Capsular invasion | |
Present / Absent | 208 (42.2) / 285 (57.8) |
ETE | |
Present / Absent | 138 (28.0) / 355 (72.0) |
cN0 / cN1 | 331 (67.1) / 162 (32.9) |
Con-CLNM (in right tumor only) | 40 (50.6) |
Right-CLNM | 280 (56.8) |
LN-prRLN metastasis | 158 (56.4) |
Only LN-arRLN metastasis | 122 (43.6) |
Number of right-CLNM (mean ± SD, range) | |
LN-arRLN | 3.22 ± 2.50 (1–16) |
LN-prRLN | 2.78 ± 2.10 (1–15) |
Right-LLNM | 144 (29.2%) |
Abbreviations: CLNM, central lymph node metastasis; Con-CLNM, contralateral central lymph node metastasis; cN0, clinical lymph node negative; cN1, clinical lymph node positive; ETE, extrathyroidal extension; LN-arRLN, lymph node anterior to the right recurrent laryngeal nerve; LN-prRLN, lymph nodes posterior to the right recurrent laryngeal nerve; Right-CLNM, right central lymph node metastasis; Right-LLNM, right lateral lymph node metastasis; SD, standard deviation. |
In this study, 331 (67.1%) cN0 patients and 162 (32.9) cN1 patients was recruited and with 280 (56.8%) patients were pathologically confirmed having right central lymph node metastasis (right-CLNM), of which 158 (56.4%) patients with LN-prRLN metastasis. Out of the 79 unilateral PTC patients that underwent bilateral CLND, 40 (50.6%) have contralateral CLNM (con-CLNM). Moreover, in this study, only 144 (29.2%) patients were found with pathologically confirmed right lateral lymph node metastasis (right-LLNM).
Univariate analysis revealed that male patient (p < 0.001), oval (p = 0.002), microcalcifications (p = 0.001), Capsular invasion (p < 0.001), ETE (p < 0.001), tumor size (p < 0.001), positive LN-arRLN (p < 0.001), and right-LLNM (p < 0.001) were significantly associated with positive LN-prRLN. However, age, HT, BRAFV600E mutation, multifocality, bilateral tumor, location of right tumor, and sonographic features including solid composition, hypoechogenicity, irregular shape, and poorly marginal were found with not significantly associated with positive LN-prRLN (p > 0.05) (Table 2).
Table 2
Univariate and multivariate analysis of clinicopathological characteristics correlated with LN-prRLN metastasis in papillary thyroid carcinoma patients
Characteristics | Univariate | Multivariate |
Total (n = 493) (%) | LN-prRLN metastasis (+) (n = 158) (%) | LN-prRLN metastasis (-) (n = 335) (%) | p | OR (95%CI) | p |
Age < 55 | 449 (91.1) | 145 (91.8) | 304 (90.7) | 0.709a | | |
Male | 150 (30.4) | 74 (46.8) | 76 (22.7) | < 0.001a | 1.978 (1.103–3.549) | 0.022 |
Hashimoto's thyroiditis | 176 (35.7) | 53 (33.5) | 123 (36.7) | 0.493a | | |
BRAFV600E mutation | 398 (80.7) | 129 (81.6) | 269 (80.3) | 0.723a | | |
Sonographic Characteristics | |
Solid composition | 491 (99.6) | 157 (99.4) | 334 (99.7) | 0.999c | | |
Hypoechogenic | 455 (92.3) | 141 (89.2) | 314 (93.7) | 0.081a | | |
Irregular shape | 385 (78.1) | 122 (77.2) | 263 (78.5) | 0.746a | | |
Poorly marginal | 327 (66.3) | 110 (69.6) | 217 (64.8) | 0.288a | | |
Oval | 138 (28.0) | 30 (19.0) | 108 (32.2) | 0.002a | 1.321 (0.661–2.640) | 0.431 |
Microcalcifications | 395 (80.1) | 140 (88.6) | 255 (76.1) | 0.001a | 0.702 (0.328–1.502) | 0.361 |
Pathological characteristics | |
Multifocality | 84 (17.0) | 31 (19.6) | 53 (15.8) | 0.295a | | |
Bilateral tumor | 114 (23.1) | 44 (27.8) | 70 (20.9) | 0.088a | | |
Location of right tumor | |
Inferior lobe | 128 (26.0) | 31 (19.6) | 97 (29.0) | 0.054a | | |
Middle lobe | 253 (51.3) | 84 (53.2) | 169 (50.4) | | | |
Superior lobe | 112 (22.7) | 43 (27.2) | 69 (20.6) | | | |
Capsular invasion | 208 (42.2) | 90 (57.0) | 118 (35.2) | < 0.001a | 2.119 (0.861–5.216) | 0.102 |
ETE | 138 (28.0) | 77 (48.7) | 61 (18.2) | < 0.001a | 3.558 (1.366–9.265) | 0.009 |
Tumor size (mm) | 13.26 ± 9.08 | 19.03 ± 10.45 | 10.54 ± 6.85 | < 0.001b | | |
Diameter > 10 | 256 (51.9) | 130 (82.3) | 126 (37.6) | < 0.001a | 2.375 (1.223–4.614) | 0.011 |
Pathologically confirmed LN metastasis | |
LN-arRLN | 272 (55.2) | 150 (94.9) | 122 (36.4) | < 0.001a | | |
Right-LLNM | 144 (29.2) | 113 (71.5) | 31 (9.3) | < 0.001a | 8.780 (4.789–16.097) | < 0.001 |
LN-arRLN metastasis | |
The number ≥ 1 | 272 (55.2) | 150 (94.9) | 122 (36.4) | < 0.001a | 14.488 (6.402–32.788) | < 0.001 |
Note: Variables with statistical significance were shown in bold. aChi-square test, bStudent's t test and cFisher's exact test were adopted. |
Abbreviations: ETE, extrathyroidal extension; LN-arRLN, lymph node anterior to the right recurrent laryngeal nerve; LN-prRLN, lymph nodes posterior to the right recurrent laryngeal nerve; OR, odds ratio; Oval, taller than wide; Right-LLNM, right lateral lymph node metastasis; SD, standard deviation; 95% CI, 95% confidence interval. |
It is worth noting that both tumor size and the number of positive LN-arRLN were positively correlated with the risk of positive LN-prRLN in patients with PTC (R2 = 0.9652, p < 0.001; R2 = 0.9109, p < 0.001, respectively) (Fig. 1a and 1b). Based on the receiver operating characteristic curve (ROC) analysis, we found that the number of positive LN-arRLN and tumor size were the well predictors for positive LN-prRLN (Fig. 1c). In our results, the area under the receiver operating characteristic curve (AUC) of number of positive LN-arRLN in positive LN-prRLN group was 0.853 (the optimal cutoff value = 1, 95% CI: 0.818–0.887), indicating that the accuracy of the test was good, which means the number of positive LN-arRLN ≥ 1 was a well predictor of positive LN-prRLN (Fig. 1c). The optimal cutoff value of tumor size in positive LN-prRLN group was defined as 10.05 mm (AUC = 0.779, 95% CI: 0.635–0.824) and means PTC patients with tumor size > 10 mm was at high risk for positive LN-prRLN (Fig. 1c).
To investigate the independent risk factors for positive LN-prRLN, binary logistic regression analyasis was performed. Our results indicated that male sex (OR = 1.978, p < 0.05), ETE (OR = 3.558, p < 0.05), tumor size > 10 mm (OR = 2.375, p < 0.05), presence of right-LLNM (OR = 8.780, p < 0.001), and the number of metastatic LN-arRLN ≥ 1 (OR = 14.488, p < 0.001) were the independent risk factors for positive LN-prRLN. However, no significant relationship was observed among oval, microcalcifications, capsular invasion and positive LN-prRLN (p > 0.05) (Table 2).
Based on the independent risk factors by binary analyses, a predictive model for positive LN-prRLN was developed. In our study, the AUC of the present model was 0.915, with Youden index of 0.672, sensitivity of 79.7%, specificity of 87.5%, positive predictive value of 86.4% and a negative predictive value of 81.2%. As several predictive models for positive LN-prRLN were reported in previous studies, we summarized the characteristics of these different models in Supplemental Table 1. In order to verify the superiority of the present model, and to compare the predictive ability between different models, the concordance index (C-index) was assessed by ROC curve. Our results showed that the AUC of our present model was the largest (AUC = 0.915), which means the present model was more appropriate in predicting positive LN-prRLN. In addition, net reclassification improvement index (NRI) was performed to quantify improvements in predictive performance of present model relative to other models. We found that adopt the present model resulted in significant improvement in reclassification for predict positive LN-prRLN (NRI > 0, all) (Fig. 2). Other parameters related to the diagnostic value were summarized in Supplemental Table 1.
Considering specific anatomy of the neck, the paratrachea-esophageal lymph nodes in the right central compartment are divided into two sections and labeled separately as LN-prRLN and LN-arRLN. Wherefore, we analyzed the predictive performance for right-LLNM among the number of right-CLNM, positive LN-arRLN, and positive LN-prRLN separately, and then found that the AUC of the number of positive LN-prRLN was the largest (AUC = 0.818) (see supplemental Fig. 1a), which means the positive LN-prRLN was more accurate in predicting right-LLNM. In this cohort study, right-LLNM was found in 144 (29.2%) patients. In univariate analysis, we found male sex (p < 0.001), oval (p < 0.001), microcalcifications (p < 0.001), location of right tumor (p < 0.05), capsular invasion (p < 0.001), ETE (p < 0.001), tumor size > 10 mm (p < 0.001), and positive LN-prRLN (p < 0.001) were significantly associated with right-LLNM (Table 3). However, age, HT, BRAFV600E mutation, multifocality, bilateral tumor, and sonographic features such as solid composition, hypoechogenicity, irregular shape, and poorly marginal were found with not significantly associated with positive LN-prRLN (p > 0.05). In the multivariate analysis, male sex (OR = 2.011, p < 0.05), microcalcifications (OR = 4.174, p < 0.05), tumor size > 10 mm (OR = 3.527, p < 0.001), and positive LN-prRLN (OR = 14.345, p < 0.001) were independent predictors of right-LLNM.
Table 3
Associations between clinicopathological characteristics and right-LLNM metastasis in papillary thyroid carcinoma patients
Characteristics | Univariate | Multivariate |
Total (n = 493) (%) | Right-LLNM metastasis (+) (n = 144) (%) | Right-LLNM metastasis (-) (n = 349) (%) | p | OR (95%C) | p |
Age < 55 | 449 (91.1) | 134 (93.1) | 315 (90.3) | 0.322a | | |
Male | 150 (30.4) | 68 (47.2) | 82 (23.5) | < 0.001a | 2.011 (1.130–3.577) | 0.017 |
Hashimoto's thyroiditis | 176 (35.7) | 48 (33.3) | 128 (36.7) | 0.481a | | |
BRAFV600E mutation | 398 (80.7) | 113 (78.5) | 285 (81.7) | 0.414a | | |
Sonographic Characteristics | |
Solid composition | 491 (99.6) | 144 (100.0) | 347 (99.4) | 0.896c | | |
Hypoechogenic | 455 (92.3) | 128 (88.9) | 327 (93.7) | 0.069a | | |
Irregular shape | 385 (78.1) | 119 (82.6) | 266 (76.2) | 0.117a | | |
Poorly marginal | 327 (66.3) | 99 (68.8) | 228 (65.3) | 0.465a | | |
Oval | 138 (28.0) | 21 (14.6) | 117 (33.5) | < 0.001a | 0.518 (0.256–1.050) | 0.068 |
Microcalcifications | 395 (80.1) | 135 (93.8) | 260 (74.5) | < 0.001a | 4.174 (1.590-10.956) | 0.004 |
Pathological characteristics | |
Multifocality | 84 (17.0) | 28 (19.4) | 56 (16.0) | 0.361a | | |
Bilateral tumor | 114 (23.1) | 37 (25.7) | 77 (22.1) | 0.385a | | |
Location of right tumor | |
Inferior lobe | 128 (26.0) | 28 (19.4) | 100 (28.7) | 0.042a | Reference | 0.216 |
Middle lobe | 253 (51.3) | 75 (52.1) | 178 (51.0) | | 1.079 (0.534–2.180) | 0.832 |
Superior lobe | 112 (22.7) | 41 (28.5) | 71 (20.3) | | 1.836 (0.835–4.035) | 0.131 |
Capsular invasion | 208 (42.2) | 87 (60.4) | 121 (34.7) | < 0.001a | 1.357 (0.576–3.198) | 0.486 |
ETE | 138 (28.0) | 72 (50.0) | 66 (18.9) | < 0.001a | 1.250 (0.501–3.120) | 0.632 |
Tumor size (mm) | 13.26 ± 9.08 | 19.03 ± 10.45 | 10.54 ± 6.85 | < 0.001b | | |
Diameter > 10 | 256 (51.9) | 125 (86.8) | 131 (37.5) | < 0.001a | 3.527 (1.810–6.871) | < 0.001 |
Pathologically confirmed LNM | |
LN-prRLN | 158 (32.0) | 113 (78.5) | 45 (12.9) | < 0.001a | 14.345 (8.136–25.293) | < 0.001 |
Note: Variables with statistical significance were shown in bold. aChi-square test, bStudent's t test and cFisher's exact test were adopted. |
Abbreviations: ETE, extrathyroidal extension; LN-arRLN, lymph node anterior to the right recurrent laryngeal nerve; LN-prRLN, lymph nodes posterior to the right recurrent laryngeal nerve; OR, odds ratio; Oval, taller than wide; Right-LLNM, right lateral lymph node metastasis; 95% CI, 95% confidence interval. |
Again, according to specific anatomy of the neck, we analyzed the predictive performance for con-CLNM among the number of right-CLNM, positive LN-arRLN, and positive LN-prRLN separately, and then found that the AUC of the number of positive LN-prRLN was the largest (AUC = 0.828) (see supplemental Fig. 1b), which means the positive LN-prRLN was more accurate in predicting con-CLNM. In our study, 40 (50.6%) con-CLNM was found in 79 right-PTC patients with unilateral lobe lesion performed bilateral CLND because those unfavorable features of ETE, or right-LLNM. In the univariate analysis, male sex (p < 0.001), ETE (p < 0.05), tumor size > 10 mm (p < 0.05), positive LN-prRLN (p < 0.001) and right-LLNM (p < 0.001) were significantly associated with con-CLNM (Table 4). However, age, HT, BRAFV600E mutation, multifocality, location of right tumor, capsular invasion, and sonographic features like solid composition, hypoechogenic, irregular shape, poorly marginal, oval, and microcalcifications were found with not significantly associated with con-CLNM (p > 0.05). In the multivariate analysis, we determined male sex (OR = 4.367, p < 0.05), positive LN-prRLN (OR = 21.134, p < 0.001) were independent predictors of con-CLNM.
Table 4
Associations between clinicopathological characteristics and con-CLNM metastasis in papillary thyroid carcinoma patients
Characteristics | Univariate | Multivariate |
Total (n = 79) (%) | Con-CLNM metastasis (+) (n = 40) (%) | Con-CLNM metastasis (-) (n = 39) (%) | p | OR (95%C) | p |
Age < 55 | 78 (98.7) | 40 (100.0) | 38 (97.4) | 0.990c | | |
Male | 25 (31.6) | 20 (50.0) | 5 (12.8) | < 0.001a | 4.367 (1.115–17.100) | 0.034 |
Hashimoto's thyroiditis | 41 (51.9) | 17 (42.5) | 24 (61.5) | 0.090a | | |
BRAFV600E mutation | 52 (65.8) | 26 (65.0) | 26 (66.7) | 0.876a | | |
Sonographic Characteristics | |
Solid composition | 78 (98.7) | 40 (100.0) | 38 (97.4) | 0.990c | | |
Hypoechogenic | 73 (92.4) | 36 (90.0) | 37 (94.9) | 0.695c | | |
Irregular shape | 66 (83.5) | 36 (90.0) | 30 (76.9) | 0.117a | | |
Poorly marginal | 60 (75.9) | 30 (75.0) | 30 (76.9) | 0.842a | | |
Oval | 18 (22.8) | 7 (17.5) | 11 (28.2) | 0.257a | | |
Microcalcifications | 72 (91.1) | 37 (92.5) | 35 (89.7) | 0.972c | | |
Pathological characteristics | |
Multifocality | 15 (19.0) | 11 (27.5) | 4 (10.3) | 0.051a | | |
Location of right tumor | |
Superior lobe | 9 (11.4) | 6 (15.0) | 3 (7.7) | 0.460c | | |
Middle lobe | 54 (68.4) | 25 (62.5) | 29 (74.4) | | | |
Inferior lobe | 16 (20.3) | 9 (22.5) | 7 (17.9) | | | |
Capsular invasion | 48 (60.8) | 27 (67.5) | 21 (53.8) | 0.214a | | |
ETE | 37 (46.8 | 24 (60.0) | 13 (33.3) | 0.018a | 1.619 (0.455–5.764) | 0.457 |
Tumor size (mm) | 13.08 ± 9.00 | 20.40 ± 9.57 | 17.62 ± 10.82 | 0.231b | | |
Diameter > 10 | 67 (84.8) | 38 (95.0) | 29 (74.4) | 0.011a | 1.008 (0.455–5.764) | 0.457 |
Pathologically confirmed LN metastasis | |
LN-prRLN | 51 (64.6) | 38 (95.0) | 13 (33.3) | < 0.001a | 21.134 (3.397-131.491) | < 0.001 |
Right-LLNM | 51 (64.6) | 34 (85.0) | 17 (43.6) | < 0.001a | 1.566 (0.314–7.808) | 0.584 |
Note: Variables with statistical significance were shown in bold. aChi-square test, bStudent's t test and cFisher's exact test were adopted. |
Abbreviations: Con-CLNM, Contralateral central lymph node metastasis; ETE, extrathyroidal extension; LN-arRLN, lymph node anterior to the right recurrent laryngeal nerve; LN-prRLN, lymph nodes posterior to the right recurrent laryngeal nerve; OR, odds ratio; Oval, taller than wide; Right-LLNM, right lateral lymph node metastasis; 95% CI, 95% confidence interval. |