Multifocality is associated with central neck lymph node metastases in papillary thyroid microcarcinoma.

BACKGROUND: This study aimed to assess the predictive factor of multifocality to identify patients at high risk of central lymph node metastasis (CLNM). PATIENTS AND METHODS: Papillary thyroid microcarcinoma patients who underwent total or hemi-thyroidectomy with effective unilateral or bilateral central lymph node dissection were enrolled.
RESULTS: Multifocality, age, sex, tumor size, extrathyroidal extension, and nodular goiter were significantly associated with CLNM. Multifocality was an independent predictor for CLNM in multivariate analysis. Compared with unifocal disease, the odds ratio for CLNM was 1.447 for patients with ≥2 tumor foci (P<0.001) and 2.978 for patients with ≥3 tumor foci (P<0.001). The significant association is at ≥3 foci diseases.
CONCLUSION: Multifocality with ≥3 tumor foci was an independent predictive factor for CLNM in papillary thyroid microcarcinoma. Multifocality should be assessed when selecting patients for prophylactic central neck lymph node dissection, and we speculate that patients with multifocality should undergo more radical treatment.

Introduction

The incidence of papillary thyroid cancer (PTC) has increased significantly worldwide in the past few decades.1–3 Papillary thyroid microcarcinoma (PTMC), a subset of PTCs, is defined as a thyroid carcinoma sized ≤1.0 cm along its greatest diameter.4 PTMC accounts for about 35%–70% of all PTC cases,5–7 and its incidence has dramatically increased in both males and females.8 PTMC in most cases is slow growing, and the majority of patients have a long survival time, although locoregional lymph node recurrence is common. The surgical treatment of PTMC remains controversial, especially in patients with multifocal disease.

Tumor multifocality is observed in approximately 20%–40% of patients with PTMC5,9–12 and can be bilateral or unilateral. However, most staging systems do not include the tumor multifocality. While some studies fail to show a significant association between multifocality and recurrent disease,5,13 multifocality is generally regarded as a risk factor of lymph node metastasis and disease recurrence in PTMC.9–12,14

Despite much attention having been paid to the increasing incidence of PTMC, there is still no consensus on the value of prophylactic central lymph node neck dissection (CLND) in PTMC. According to the 2015 ATA guidelines, thyroidectomy without CLND may be appropriate for small (T1 or T2), noninvasive, clinically node-negative PTC (cN0), but this does not address the issue of whether CLND is of value in PTMC.15 Despite having small tumors, patients with PTMC have high incidence of lymph node metastasis and extrathyroidal extension (ETE); the incidence of central lymph node metastasis (CLNM) ranges from 30% to 60% in these patients.5,16–21 Unfortunately, no preoperative clinical features can reliably identify CLNM, and preoperative sonography has an extremely low sensitivity for CLNM.22 Therefore, it is imperative to develop a suitable approach to identify PTMC patients with a high risk for CLNM who may benefit from CLND.

The relationship between multifocality and CLNM in PTMC remains unclear. Few studies have assessed the association between the number of foci and the demographic characteristics of PTMC. This study aimed to assess the predictive factor of multifocality for CLNM in PTMC, which may improve our ability to identify patients who may benefit from CLND.

Patients and methods

Patient identification

From January 2007 to December 2015, a total of 12,530 patients underwent thyroidectomy at Zhejiang Cancer Hospital. Of them, 6,826 patients were pathologically diagnosed with PTC (PTC >1 cm in diameter was noted in 2,238 patients) and 4,588 with PTMC. Patients who underwent thyroidectomy without effective central lymph node dissection (dissection of at least 1 lymph node) or lateral lymph node dissection with or without central lymph node dissection were excluded. Patients with PTMC and other types of thyroid cancer such as medullary carcinoma or anaplastic thyroid cancer were also excluded. Eventually, 3,543 patients with PTMC who underwent effective central lymph node dissection (dissection of ≥1 lymph nodes) were enrolled in this study. The 3,543 patients were divided into 2 groups: 2,458 patients with 1 lesion and 1,085 patients with ≥2 foci (Figure 1).

Figure 1

Flowchart of patients’ recruitment.

All 3,543 patients in this cohort underwent total or hemi-thyroidectomy with effective unilateral or bilateral central lymph node dissection. If unilateral malignant lesion was present, ipsilateral central neck dissection of lymph nodes was employed; if bilateral malignant lesions were observed, bilateral central neck dissection of lymph nodes was done. The CLND was performed superior to the hyoid bone, inferior to the suprasternal notch, and lateral to the carotid sheaths. This study was approved by the Ethics Committee of Zhejiang Cancer Hospital. Written informed consent was obtained from all patients before surgery, and all operations were performed by experienced surgeons at the same cancer hospital. All patients gave a written informed consent to have their medical records reviewed for this study.

Data collection

Demographic characteristics, including age, sex, tumor size, multifocality, and tumor laterality (unilateral/bilateral), were collected from electronic medical records. According to the TNM staging system, age was dichotomized at 45 years. Tumor size was classified as <5 or ≥5 mm using the largest diameter. Multifocality was defined as ≥2 tumor foci in the same lobe or different lobes including the thyroid isthmus. The presence of ETE, lymphovascular invasion (LVI), lymphocytic thyroiditis, and nodular hyperplasia was recorded according to the pathological findings. In addition, the postoperative complications and prognosis were also evaluated in these patients. The postoperative complications’ information and prognosis were recorded through review of the medical records.

Statistical analysis

Descriptive statistics were employed to summarize study data. Continuous data are expressed as mean ± standard deviation. Statistical analyses were performed using SPSS version 23.0 for Mac (IBM Corporation, Armonk, NY, USA). χ2 test was used to assess the differences between groups in univariate analysis. Logistic regression analysis was applied in multivariate analysis to calculate the odds ratios (ORs) for the associations between CLNM and various clinicopathologic factors. A value of 2-tailed P<0.05 was considered statistically significant. Software Prism 6.0c (Graph Pad Software, Inc., La Jolla, CA, USA) was used to generate figures.

Results

Patients’ characteristics

A total of 3,543 patients with PTMC were enrolled (Figure 1). There were 2,867 females (80.92%) and 676 males (19.08%) with an average age of 45.70±10.48 years (range, 12–81 years). The average size of the largest focus was 5.97±2.30 mm (range, 0.5–10 mm; median, 6 mm). Multifocality was suspected on ultrasonography in 806 patients, postoperative microscopy indicated multifocality in 1,085 patients, and the consistency rate of multifocality between ultrasonography and microscopy was 74.29% (806/1,085). Overall, from microscopic foci, 2,458/3,543 patients (69.38%) had unifocal PTMC and 1,085 (30.62%) had multifocal PTMC; 767/1,085 patients (70.69%) had 2 foci, 219 (20.18%) had 3 foci, and 99 (9.13%) had ≥4 foci.

The average number of central neck lymph nodes dissected was 4.29±3.23 (range, 1–29). CLNM was found in 1,057/3,543 patients (29.83%), of whom 781/1,057 (73.89%) had 1–2 CLNM, 238 (22.52%) had 3–5 CLNM, and 38 (3.59%) had ≥6 CLNM. ETE was present in 1,487/3,543 (41.97%) patients, LVI in 25/3,543 (0.71%) patients, lymphocytic thyroiditis in 337/3,543 (9.51%) patients, and nodular hyperplasia in 1,163/3,543 (32.83%) patients (Table 1).

Table 1

Demographic characteristics of the cohort of patients with PTMC (n=3,543)

Variable	Mean + SD (median: range) or value (%)
Age (years)	45.70±10.48 (46: 12–81)
<45	1,616/3,543 (45.61%)
≥45	1,927/3,543 (54.39%)
Sex
Male	676/3,543 (19.08%)
Female	2,867/3,543 (80.92%)
Size of largest focus (mm)	5.97±2.30 (6: 0.5–10)
<5	994/3,543 (28.06%)
≥5	2,549/3,543 (71.94%)
Focality	1.45±0.88 (1: 1–13)
Unifocal	2,458/3,543 (69.38%)
Multifocal	1,085/3,543 (30.62%)
2 foci	767/1,085 (70.69%)
3 foci	219/1,085 (20.18%)
≥4 foci	99/1,085 (9.13%)
Bilateral tumors	1,670/3,543 (47.14%)
Number of central neck lymph nodes dissected	4.29±3.23 (3: 1–29)
Central neck lymph node metastasis
Negative	2,486/3,543 (70.17%)
Positive	1,057/3,543 (29.83%)
1–2 nodes	781/1,057 (73.89%)
3–5 nodes	238/1,057 (22.52%)
≥6 nodes	38/1,057 (3.59%)
ETE	1,487/3,543 (41.97%)
LVI	25/3,543 (0.71%)
Lymphocytic thyroiditis	337/3,543 (9.51%)
Nodular hyperplasia	1,163/3,543 (32.83%)

Abbreviations: ETE, extrathyroidal extension; LVI, lymphovascular invasion; PTMC, papillary thyroid microcarcinoma.

Postoperative complications and prognosis

Hypocalcemia was found in 9.23% (327/3,543) of patients, hoarseness in 1.72% (61/3,543) of patients, and bleeding in 0.11% (4/3,543) of patients. Among these patients, 344 were lost to follow-up, recurrence was observed in 51 patients, and 5 patients died (1 died of cerebral hemorrhage).

Univariate analysis

The associations between CLNM and several demographic factors were further investigated. In univariate analysis, CLNM was significantly associated with patients younger than 45 years, males, and patients with a largest focus of ≥5 mm, multifocality, ETE, or nodular hyperplasia (all P<0.001). There were no significant associations of CLNM with bilateral tumors (P=0.813), LVI (P=0.120), and lymphocytic thyroiditis (P=0.187; Table 2).

Table 2

Factors associated with CLNM in PTMC (n=3,543) in univariate analysis using the χ2 test

Variable	Central lymph node metastasis		P-value
	Positive (n=1,057) n (%)	Negative (n=2,486) n (%)
Age (years)			<0.001
<45	554 (52.41)	1,062 (42.72)
≥45	503 (47.59)	1,424 (57.28)
Sex			<0.001
Male	252 (23.84)	424 (17.06)
Female	805 (76.16)	2,076 (82.94)
Size of largest focus (mm)			<0.001
<5	183 (17.31)	811 (32.62)
≥5	874 (82.69)	1,675 (67.38)
Focality			<0.001
Unifocal	658 (62.25)	1,800 (72.41)
Multifocal	399 (37.75)	686 (27.59)
2 foci	223 (55.89)	544 (79.30)
3 foci	105 (26.32)	114 (16.62)
≥4 foci	71 (17.79)	28 (4.08)
Bilateral tumors	495 (46.83)	1,175 (47.26)	0.813
ETE	557 (52.70)	918 (36.93)	<0.001
LVI	11 (1.04)	14 (0.56)	0.120
Lymphocytic thyroiditis	90 (8.51)	247 (9.94)	0.187
Nodular hyperplasia	292 (27.63)	871 (35.04)	<0.001

Abbreviations: CLNM, central lymph node metastasis; ETE, extrathyroidal extension; LVI, lymphovascular invasion; PTMC, papillary thyroid microcarcinoma.

Multivariate analysis

Binary logistic regression was employed for multivariate analysis to calculate the ORs for CLNM (Tables 3 and 4). Age, sex, size of largest focus, ETE, nodular hyperplasia, and multifocality (≥2 foci) were controlled in 3,543 patients. Multifocality (≥2 foci) was significantly associated with CLNM (OR =1.447, 95% confidence interval [CI] =1.236–1.693; P<0.001). Five other factors were also independent predictors for CLNM (Table 3): age (OR =1.462; P<0.001), female sex (OR =0.682; P<0.001), the largest focus ≥5 mm (OR =1.854; P<0.001), ETE (OR =1.552; P<0.001), and nodular hyperplasia (OR =0.826; P=0.023).

Table 3

Factors associated with CLNM in PTMC (n=3,543) in multivariate logistic regression analysis

Variable	Central lymph node metastasis		95% CI
	OR	P-value	Lower	Upper
Multifocality (≥2 foci)	1.447	<0.001	1.236	1.693
Age (<45 years)	1.462	<0.001	1.258	1.699
Sex (female)	0.682	<0.001	0.569	0.817
Size of largest focus (≥5 mm)	1.854	<0.001	1.532	2.243
ETE	1.552	<0.001	1.328	1.814
Nodular hyperplasia	0.826	0.023	0.701	0.974

Abbreviations: CI, confidence interval; CLNM, central lymph node metastasis; ETE, extrathyroidal extension; OR, odds ratio; PTMC, papillary thyroid microcarcinoma.

Table 4

Factors associated with CLNM in multifocal PTMC (n=1,085) in multivariate logistic regression analysis

Variable	Central lymph node metastasis		95% CI
	OR	P-value	Lower	Upper
Multifocality (≥3 foci)	2.978	<0.001	2.256	3.932
Sex (female)	1.500	0.020	1.065	2.112
Size of largest focus (≥5 mm)	2.432	<0.001	1.678	3.525
ETE	1.474	0.005	1.127	1.928

Abbreviations: CI, confidence interval; CLNM, central lymph node metastasis; ETE, extrathyroidal extension; OR, odds ratio; PTMC, papillary thyroid microcarcinoma.

In subsequent analysis, the association between CLNM and multifocality was evaluated with the same model; patients with ≥3 foci had a significantly higher OR for CLNM than patients with ≤2 foci (multifocal; OR =2.978, 95% CI =2.256–3.932; P<0.001; n=1,085). Similarly, among patients with multifocality, having the largest focus of ≥5 mm (OR =2.432; P<0.001) and ETE (OR =1.474; P=0.005) were significantly associated with CLNM. Being female was also a risk factor for CLNM in multifocal PTMC (OR =1.500; P=0.020). Age (P=0.897) and nodular hyperplasia (P=0.678) were not associated with CLNM in multifocal PTMC (Table 4; data not shown).

The association between the number of the foci and CLNM was also assessed. The incidence of CLNM was 26.77% in unifocal disease patients, 29.07% in patients with 2 foci, 47.95% in patients with 3 foci, and 71.72% in patients with ≥4 foci. A significant association existed between multifocality and CLNM in PTMC (P<0.001; Figure 2).

Figure 2

Association between CLNM and the number of foci in PTMC (n=3,543).

Abbreviations: CLNM, central lymph node metastasis; PTMC, papillary thyroid microcarcinoma.

Multifocality was observed in 30.62% of patients with PTMC, and the incidence of CLNM was 36.77% (399/1,085) in multifocal PTMC compared to 26.77% (658/2,458) in patients with unifocal disease. The incidence of CLNM increased with the number of PTMC foci. There was no significant difference in CLNM between unifocal PTMC (26.77%) and PTMC with 2 foci (29.07%). However, the incidence of CLNM was 1.79-fold higher for PTMC with 3 foci (47.95%) and 2.68-fold higher for PTMC with ≥4 foci (71.72%), compared to unifocal disease. Multivariate analysis confirmed that multifocality (≥2 foci) was significantly associated with CLNM (OR =1.447). Furthermore, patients with ≥3 foci had a higher risk of CLNM than patients with 2 foci (OR =2.978; P<0.001).

Discussion

This study investigated the relationship between multifocality and CLNM in PTMC. This study reveals that multifocality is significantly associated with CLNM in PTMC. The high incidence of CLNM for patients with multifocal PTMC may indicate higher tumor aggressiveness and a higher risk of regional recurrence.

Consensus regarding the association between multifocality and CLNM has not yet been reached. Similar to our study, Afif et al23 reported an association between the number of foci in PTC and CLNM (P<0.001; OR =2.355 for ≥3 tumor foci; P=0.026), although PTMC was not analyzed as a separate subgroup. A meta-analysis by Qu et al24 revealed that CLNM was associated with multifocality in PTMC (relative risk =1.40, P=0.001). Guo et al17 reported the presence of 2, 3, or ≥4 foci was associated with a significantly higher risk of CLNM in PTMC (OR =1.675, 2.360, and 2.703, respectively) compared to unifocal PTMC. Similar findings have been reported by other studies.19,20 In contrast, multifocality was significantly related to CLNM in univariate analysis (P=0.020) but not in multivariate analysis (P=0.138) in the study by Zhou et al,21 and Lee et al18 suggested multifocality was not associated with CLNM and did not significantly influence the predictive factor of CLNM. It is possible that the differences in sample size and selection criteria among studies may explain the discrepancies. Our study indicates that multifocality is significantly associated with CLNM in PTMC.

Prophylactic CLND in the management of PTC remains controversial, especially in PTMC.25 Chang et al16 reported 39% (239/613) of patients with PTMC had CLNM, and they recommended prophylactic CLND for all patients. In contrast, Wada et al26 stated only patients with palpable lymph nodes should undergo prophylactic CLND. In the last decade, patients with PTMC in our hospital have routinely undergone prophylactic CLND, regardless of the presence of palpable lymph nodes. However, it would be desirable to identify subgroups of patients with less aggressive PTMC at low risk of CLNM who could avoid prophylactic CLND.

The reported incidence of CLNM in PTMC ranges from 30% to 60%.5,16–21 In this study, 29.83% of patients had CLNM, which is lower than previously reported. This could be attributed to the fact that most patients had earlier stage disease and because patients who underwent lateral lymph node dissection were excluded from this study. Several studies have identified lymph node metastasis, including lateral lymph node metastasis, as a prognostic factor for poor outcome in PTMC.9–12,14,27 Mercante et al14 found that lymph node metastasis was an independent risk factor for recurrence or distant metastasis (P<0.05). Usluogullari et al12 reported that lymph node metastasis was an independent predictor of recurrence in multivariate analysis (OR =51.4; P=0.003). Agarwal et al27 stated that lymph node metastasis was associated with more aggressive disease, recurrence, and poorer prognosis. Multifocality has also been associated with aggressive disease and poor prognosis in PTMC. Chow et al9 found the risk of cervical lymph node recurrence was 6.2-fold (P<0.01) and 5.6-fold (P<0.02) higher if lymph node metastasis and multi-focal disease were present at diagnosis. Qu et al11 reported that the presence of >3 tumor foci was an independent predictor of recurrence in multivariate analysis (HR =2.60, P=0.001). However, Mehanna et al13 failed to find that tumor multifocality and lymph node involvement were significantly associated with recurrence in PTMC. Kim et al5 found that multifocality was an independent risk factor for disease recurrence in PTC but not in PTMC. Further studies are required to define the prognostic value of CLNM in PTMC in the absence of lateral lymph node metastasis as a confounding factor. Moreover, large, multicentered studies with longer follow-up period are warranted to draw definitive conclusions.

In this study, CLNM was significantly associated with age, sex, tumor size, multifocality, ETE, and nodular hyperplasia in univariate analysis. Multivariate analysis confirmed that all these factors are independent predictive factors for CLNM in PTMC, in agreement with previous findings.20,28–30 Moreover, bilateral tumors, LVI, and lymphocytic thyroiditis were not significantly associated with CLNM in this study, consistent with previously reported studies.31,32 Otherwise, coexistence of thyroiditis was previously identified as a negative independent predictor for CLNM.33,34 However, in the analysis of patients with multifocal PTMC, age and nodular hyperplasia were not found to be independent predictors for CLNM. It is difficult to explain the role of sex as a predictor for CLNM. In the entire cohort, being female (OR =0.682, P<0.001) was associated with reduced risk of CLNM, whereas being female was a risk factor for CLNM in multifocal PTMC (OR =1.500, P=0.020). Similar reports have not been published on the effect of sex on the risk of CLNM, and the predictive factor of sex for CLNM in multifocal PTMC still needs to be further elucidated in future studies.

Although a large number of patients were included, there were still limitations in this study. First, the retrospective nature precluded the identification of the mechanistic relationships between prognosis and multifocality or CLNM. Second, all patients were treated at a single hospital. Third, a number of patients did not undergo CLND due to the low rate of ultrasonography-guided preoperative fine-needle aspiration in PTMC; these patients were diagnosed using the final pathology as having incidental PTMC and excluded from this study. In general, incidental PTMC has features suggestive of less aggressive disease and a lower incidence of CLNM. Although this bias does not affect the association between multifocality and CLNM, the incidence of CLNM reported in this study may be lower than that in other studies of PTMC. Moreover, the relationship between lateral lymph node metastasis and multifocality was not assessed in this study. Lastly, the impact of molecular marker such as a determinant of sensitivity to proteasome inhibitors (BRAFV600E) on multifocality and CLNM requires further research.

Conclusion

This study indicates that multifocality is associated with an increased risk of CLNM in PTMC. Multifocality with ≥3 tumor foci was an independent predictor for CLNM. In addition to the size of the largest focus and ETE, multifocality should be considered when selecting subgroups of patients with PTMC for prophylactic CLND. We recommend that patients with multifocality should receive more radical treatment.