Abstract
Utilizing BRAFV600E mutation as a marker may reduce unnecessary prophylactic central neck dissection (pCND) in clinically nodal negative (cN0) neck for small (≤2 cm) classical papillary thyroid carcinoma (PTC). We aimed to assess whether BRAF is a significant independent predictor of occult central nodal metastasis (CNM) and its contribution to the overall prediction after adjusting for other significant preoperative clinical factors in small PTC. Primary tumor tissue (paraffin-embedded) from 845 patients with small classical cN0 PTC who underwent pCND was tested for BRAF mutation. Clinicopathologic factors were compared between those with and without BRAF. BRAF was evaluated to see if it was an independent factor for CNM. Prediction scores were generated using logistic regression models and their predictability was measured by the area under the ROC curve (AUC). The prevalence of BRAF was 628/845 (74.3%) while the rate of CNM was 285/845 (33.7%). Male sex (odds ratio (OR): 2.68, 95% CI: 1.71–4.20), large tumor size (OR: 2.68, 95% CI: 1.80–4.00), multifocality (OR: 1.49, 95% CI: 1.07–2.09), lymphovascular permeation (OR: 10.40, 95% CI: 5.18–20.88), and BRAF (OR: 1.65, 95% CI: 1.10–2.46) were significant independent predictors of CNM, while coexisting Hashimoto's thyroiditis (OR: 0.56, 95% CI: 0.40–0.80) was an independent protective factor. The AUC for prediction score based on tumor size and male sex was similar to that of prediction score based on tumor size, male sex, and BRAF status (0.68 vs 0.69, P=0.60). Although BRAF was an independent predictor of CNM, knowing its status did not substantially improve the overall prediction. A simpler prediction score based on male sex and tumor size might be sufficient.
Introduction
Papillary thyroid carcinoma (PTC) is the most common type of differentiated thyroid carcinoma with an adjusted incidence doubled over the last 20 years (Kilfoy et al. 2009, HKCR 2013, SEER 2013). Despite its relatively good prognosis, locoregional recurrence (LR) is common (Wong et al. 2012). With recognition of the concept of step-wise progression of lymph node metastasis originating from the central (level VI) to the lateral compartment (levels II–V) and the fact that preoperative ultrasonography (USG) only identifies approximately half of the central nodal metastasis (CNM), a growing number of surgeons have advocated routine prophylactic central neck dissection (pCND) at the time of the total thyroidectomy (TT; Machens et al. 2009, Roh et al. 2009, Hwang & Orloff 2011). However, this remains controversial particularly in low-risk PTC as the American Thyroid Association (ATA) only recommends CND in clinically involved (cN1) neck lymph nodes or in T3 and T4 tumors (Cooper et al. 2009). Although a recent meta-analysis has found that those with clinically nodal negative (cN0) neck who undergo pCND might have reduced risk of LR than those who undergo TT-alone in the short-term, the former group has higher risks for temporary hypoparathyroidism and overall morbidity (Lang et al. 2013). Therefore, identification of predictive factors for occult CNM is crucial to reduce unnecessary pCND (Koo et al. 2009, Hartl et al. 2012, Zhang et al. 2012, Zhao et al. 2013).
In recent years, a T1799A point mutation in the v-raf murine sarcoma viral oncogene homolog B1 (BRAF) resulting in a valine-to-glutamic acid switch at codon 600 (BRAFV600E) has emerged as a molecular marker for aggressive behavior in PTC (Xing et al. 2005, 2013a). Previous studies have found that BRAF-positive (BRAF+ve) tumors are significantly larger in size, more frequent lymph node metastasis and extrathyroidal extension and also higher tumor stage, risk of LR, and disease-related mortality than BRAF-negative (BRAF−ve) tumors (Frasca et al. 2008, Xing et al. 2009, 2013b, O'Neill et al. 2010, Kim et al. 2012, Li et al. 2012, Alzahrani & Xing 2013). Therefore, in addition to the existing prognostic staging systems (Lang et al. 2007a), BRAF mutation could be used as a potential marker for stratifying tumor risk (Xing et al. 2009, Yip et al. 2009, Howell et al. 2013). Previous studies have examined the utility of BRAF mutation testing in optimizing surgical management and suggested that BRAF+ve patients may benefit from more extensive initial surgery such as pCND (Xing et al. 2009, Yip et al. 2009, O'Neill et al. 2010, Joo et al. 2012). Joo et al. (2012) evaluated the utility of BRAF mutation by pyrosequencing on 148 preoperative fine-needle aspiration (FNA) specimens and concluded that preoperative BRAF analysis by FNA could help to predict occult CNM. However, most studies only evaluated the association of BRAF with overall presence of lymph node metastasis rather than occult CNM alone (Kim et al. 2006, 2012, Frasca et al. 2008, Xing et al. 2009, Yip et al. 2009, O'Neill et al. 2010, So et al. 2011, Nam et al. 2012). In addition, there have been few studies adopting the strict definition of a pCND when examining the association between BRAF mutation and lymph node metastasis (Lee et al. 2012, Paulson et al. 2012, Dutenhefner et al. 2013, Howell et al. 2013). Furthermore, in some studies (Xing et al. 2005, Kim et al. 2006, Frasca et al. 2008, So et al. 2011, Nam et al. 2012), after adjusting for other significant clinicopathologic factors such as age, sex, multifocality, tumor size, and extrathyroidal extension, BRAF became nonsignificant. Therefore, currently there is still insufficient data to support pCND on the basis of BRAF mutation status alone in low-risk PTC (Xing et al. 2013a). Given these controversies, our study aimed to assess whether BRAF mutation was a significant independent predictor of occult CNM in cN0 neck and also the role of BRAF mutation in contributing to the overall prediction after adjusting for other significant preoperative clinical factors in a large cohort of small (≤2 cm) PTC.
Subjects and methods
Patients
This study protocol was approved by the local institutional review board (IRB No:H-1305-020-486). All consecutive patients who underwent TT and CND at Seoul National University Hospital from December 2008 to November 2012 were retrospectively analyzed. All data were collected prospectively. Patients who were diagnosed preoperatively by FNA or intraoperatively on frozen section were included. Figure 1 shows the study flow chart. Altogether there were 1916 patients with small (≤2 cm) classic PTC, who underwent TT and CND. All tumors classified as histological variants of PTC (including follicular variant) (n=52) (see Table 1) or with pathologic size >2.0 cm were excluded. Of the 1916 patients, 168 (8.8%) were excluded because BRAF testing was not done or available while 457 (23.9%) were excluded because they were suspicious of or cytologically confirmed to have lymph node metastases detected on preoperative neck USG or intraoperative evaluation. Within this latter group, 363 patients subsequently underwent lateral selective neck dissection, while the other 94 underwent therapeutic CND. Therefore, there were 1291 cN0 PTC patients who underwent TT + prophylactic CND (pCND) and had their tumor tissue tested for BRAF mutation. To ensure an adequate pCND specimen, those patients with <3 central lymph nodes (CLNs) harvested by pCND were excluded (n=446). Therefore, 845 patients were eligible for analysis. However, since a substantial proportion of patients were excluded, patient/tumor characteristics were compared between the two groups to look for possible selection bias on the basis of CLN yield.
Prevalence of BRAF mutation in the classic papillary thyroid carcinoma (n=845) and the excluded histopathologic variants (n=52)
Variant of papillary thyroid carcinoma | BRAF mutation (%) |
---|---|
Classic/conventional (n=845) | 628 (74.3) |
Follicular variant (n=21) | 7 (33.3) |
Tall cell (n=15) | 14 (93.3) |
Oncocytic (n=11) | 4 (36.4) |
Diffuse sclerosing (n=2) | 1 (50.0) |
Solid cell (n=2) | 0 (0.0) |
Clear cell (n=1) | 0 (0.0) |
Methods
DNA isolation from surgical specimen and FNA samples
B-type Raf Kinase V600E (BRAFV600E) mutation analysis from surgical specimen was conducted prospectively and routinely for all patients with PTC after February 2009. From the surgical specimen, areas of tumor were identified on hematoxylin and eosin (H&E) stained slides, marked by pathologists and dissected using a fine needle from 10-μm-thick unstained sections. In patients with bilateral or multifocal tumors, only the largest focus was examined for the BRAFV600E mutation. Genomic DNA was isolated by incubation with extraction buffer (1 M Tris–HCl, pH 7.4; 0.5 M EDTA, pH 8.0, 5% Tween 20) and proteinase K at 60 °C for 12–15 h, followed by standard phenol–chloroform extraction and ethanol precipitation.
To see correlation of BRAF between surgical specimen and FNA sample, the results of BRAF test from the two materials were compared in 19 patients who had BRAF mutation analysis from FNA samples before surgery. All FNAs were carried out under ultrasound guidance. All aspirations (usually two passes for each lesion) were obtained with 25-gauge or 27-gauge needles. The aspirated material was fixed with a hemolytic and preservative solution (Cytolit; Hologic Cytyc Company, Marlborough, MA, USA) after rinsing the needle into this solution. The resulting slide was fixed in 95% ethanol and stained with Papanicolaou. DNA extraction was performed on FNA samples using the ThinPrep 2000 system (Hologic Cytyc Company) using the QIAamp tissue kit (Qiagen).
BRAFV600E mutation analysis
The BRAF exon 15, which contains the most common BRAF mutation, a T1799A transversion (BRAFV600E), was amplified by PCR with genomic DNA. The primers and PCR conditions were as follows: forward, 5′-TCATAATGCTTGCTCTGATAGGA-3′; reverse 5′-GGCCAAAAATTTAATCAGTGGA-3′; denaturation at 94 °C for 10 min, followed by 35 cycles of 94 °C for 1 min, 60 °C for 1 min, 72 °C for 1 min, and a final extension step at 72 °C for 10 min. After purification of the PCR products with the QIAGEN-QIAquick PCR purification kit (Qiagen), direct DNA bidirectional sequencing was done with an ABI 3130XL Genetic Analyzer BigDye Terminator (Applied Biosystems). Sequence data were analyzed manually by two independent pathologists.
Management of PTC
A preoperative USG was routinely performed to examine both central and lateral neck compartments with any suspicious nodes aspirated for cytology. TT was the preferred procedure for all patients with a preoperative diagnosis of PTC. Once the diagnosis of PTC had been confirmed by frozen section, regardless of the tumor size or local extent, an ipsilateral pCND was performed for unifocal tumors while a bilateral pCND was performed for bilateral or isthmic tumors. All pCND were carried out in accordance with anatomical landmarks described by the ATA (Carty et al. 2009) and were performed immediately after the completion of the TT. It comprised the removal of all nodes and fibro-fatty tissue extending vertically from the hyoid bone to the thoracic inlet and laterally from the medial border of the common carotid artery to the midline of the trachea. The ipsilateral recurrent laryngeal nerve (RLN) was mobilized and skeletonized along its entire cervical course.
Postoperative assessment
All post-surgical patients were followed up within 1–2 weeks and then 2–3 monthly for the first year. Those taking calcium with or without calcitriol supplements were followed more frequently with an aim of gradually weaning off these supplements while maintaining normocalcemia. By definition, those who discontinued all supplements in the presence of normocalcemia ≤6 months after surgery were regarded as temporary hypoparathyroidism whereas those who continued for >6 months were categorized as permanent hypoparathyroidism. Also both vocal cords were examined endoscopically 1–2 days before and within 2 weeks after thyroidectomy using flexible laryngoscope. Any reduction in cord movement was recorded as vocal cord palsy. Those with vocal cord palsy were examined every 3 months. The presence of cord palsy lasting >6 months was regarded as permanent.
Follow-up protocol
All post-surgical patients were followed up within 2 weeks in a specialized oncology clinic. A follow-up visit was conducted at 3-month, 6-month, and then annually thereafter. Clinical examination, neck USG, and nonstimulated thyroglobulin (Tg) level were done during follow-up visits. Stimulated thyroglobulin (sTg) was defined as a Tg level measured in the presence of thyrotrophin (TSH) >30 mIU/l either by thyroxine withdrawal or recombinant TSH injections. Radioiodine (RAI) ablation and pre-ablation sTg level were done approximately 3 months after surgery (because most patients would have had a contrast CT before they were referred to us for neck USG and surgery) while the post-ablation sTg level was taken approximately 9 months after surgery (6–7 months after RAI ablation). Tg autoantibodies were measured at the same time. The decision for RAI was based on presence of ≥1 risk factors such as tumor size >1.5 cm, lymph node metastasis, age >45 years old, extrathyroidal extension, macroscopic postoperative residual disease in the neck, and distant metastasis. Thirty millicuries I131 was the standard ablative dose for low-risk PTC. TSH suppression to <0.1 mIU/l was recommended for high- and intermediate-risk patients. All relevant clinical, laboratory, radiologic, and perioperative data were collected prospectively and follow-up data were regularly updated in a computerized database.
Statistical analysis
Continuous variables were expressed as mean±s.d. and groups were compared using the Mann–Whitney U test. χ2 tests were used to compare categorical variables. Any clinicopathologic features which were statistically significantly associated with occult CNM in the univariate analysis were entered into multivariate analysis by logistic regression to determine independent factors and to formulate combined prediction scores based on the regression coefficients. The area under a receiver characteristic (ROC) curve (AUC) was used to measure the relative predictability of independent factors and combined prediction scores. A bootstrap approach with 1000 resamples was used to compare AUCs and to estimate 95% CIs for each AUC. All statistical analyses were conducted using SPSS version 18.0 (SPSS, Inc.) and R version 2.14.0 (R Foundation for Statistical Computing, Vienna, Austria). P values below 0.05 were considered statistically significant.
Results
Our cohort was mostly females (86.7%). The mean (±s.d.) and median (range) age at operation were 45.7±11.9 and 46.0 (12.0–77.0) years old respectively. The mean (±s.d.) tumor size was 0.8±0.4 cm. The mean (±s.d.) number of CLNs and positives CLNs removed were 6.6±3.8 and 0.9±1.8, respectively. The overall rate of occult CNM was 285/845 (33.7%), while the rate of BRAF+ve mutation in primary tumors was 628/845 (74.3%).
Table 2 shows a comparison of patient characteristics between those with ≥3 CLNs and with <3 CLNs. There were no significant differences except for a higher concomitant Hashimoto's thyroiditis (HT) (P<0.001) and CNM (P<0.001) for those with ≥3 CLNs.
A comparison of patient/tumor characteristics between those with ≥3 central lymph nodes (CLNs) harvested and those with <3 CLNs harvested during prophylactic central neck dissection
Patients with ≥3 CLNs harvested (n=845) | Patients with <3 CLNs harvested (n=446) | P value | |
---|---|---|---|
Age at operation (years) | 45.7±11.9 | 46.5±11.7 | 0.218 |
Sex | 0.116 | ||
Male | 112 (13.3) | 76 (17.0) | |
Female | 733 (86.7) | 370 (83.0) | |
Tumor characteristics | |||
Tumor size (cm) | 0.8±0.4 | 0.8±0.4 | 0.546 |
Tumor bilaterality | 171 (20.2) | 73 (16.4) | 0.087 |
Tumor multifocality | 292 (34.6) | 133 (29.8) | 0.085 |
Extra-thyroidal extension | 477 (56.4) | 254 (57.0) | 0.828 |
LV permeation | 70 (8.3) | 30 (6.7) | 0.780 |
Coexisting HT | 331 (39.2) | 60 (13.5) | <0.001 |
Occult CNM (pN1a) | 285 (33.7) | 66 (21.4)a | <0.001 |
BRAF mutation | 628 (74.3) | 338 (75.8) | 0.564 |
LV, lymphovascular; HT, Hashimoto's thyroiditis; CNM, central nodal metastasis; bold signifies statistical significance, P<0.05.
Even after excluding those with no CLNs harvested (n=138).
Table 3 shows a comparison of patients' clinicopathological features, tumor characteristics, and TNM tumor stages between BRAF+ve and BRAF−ve groups. Age and sex ratio were similar between the two groups. The BRAF+ve group had significantly larger sized tumors (0.8 vs 0.7 cm, P<0.001) and higher incidence of extrathyroidal extension (61.0 vs 43.3%, P<0.001) and occult CNM (37.4 vs 23.0%, P<0.001), while the incidence of coexisting HT was significantly less (34.6 vs 52.5%, P<0.001) than the BRAF−ve group. The number of CLNs harvested was similar between the two groups regardless of the extent of pCND, but the overall number of metastatic CLNs excised and the CLN ratio (CLNR) in the BRAF+ve group were significantly higher than in the BRAF−ve group (1.0 vs 0.7, P<0.001 and 16.1 vs 10.6%, P<0.001). However, when stratified into unilateral and bilateral pCND, these significant differences were not observed with bilateral pCND. The BRAF+ve group had a significantly higher proportion of stage III tumors and a corresponding lower proportion of stage I tumors than the BRAF−ve group. As a result, RAI ablation was given more frequently in the BRAF+ve group (35.2 vs 22.6%, P=0.001). After excluding those with elevated anti-Tg antibody, the pre-ablation sTg level in the BRAF+ve group was significantly higher than the BRAF−ve group (2.4 vs 1.0 μg/l, P=0.032) while the post-ablation sTg was similar (0.6 vs 0.2 μg/l, P=0.473).
A comparison of patient clinicopathological features, tumor characteristics, and postoperative stimulated thyroglobulin levels between those with a BRAF mutation (BRAF+ve group) and without a BRAF mutation (BRAF−ve group)
BRAF+ve group (n=628) | BRAF−ve group (n=217) | P value | |
---|---|---|---|
Age at operation (years) | 45.8±11.9 | 45.6±11.8 | 0.802 |
Sex | 0.116 | ||
Male | 90 (14.3) | 22 (10.1) | |
Female | 538 (85.7) | 195 (89.9) | |
Tumor characteristics | |||
Tumor size (cm) | 0.8±0.4 | 0.7±0.4 | <0.001 |
Microcarcinoma (<1 cm) | 460 (73.2) | 177 (81.6) | 0.045 |
Tumor bilaterality | 135 (21.5) | 36 (16.6) | 0.107 |
Tumor multifocality | 225 (35.8) | 67 (30.9) | 0.186 |
Extra-thyroidal extension | 383 (61.0) | 94 (43.3) | <0.001 |
LV permeation | 53 (8.4) | 17 (7.8) | 0.780 |
Coexisting HT | 217 (34.6) | 114 (52.5) | <0.001 |
Occult CNM (pN1a) | 235 (37.4) | 50 (23.0) | <0.001 |
Extent of pCND | 0.063 | ||
Unilateral | 483 (76.9) | 179 (82.5) | |
Bilateral | 145 (23.1) | 38 (17.5) | |
No. of CLNs harvested | 6.5±3.6 | 6.9±4.3 | 0.144 |
Unilateral pCND (n=662) | 6.0±3.2 | 6.1±3.0 | 0.463 |
Bilateral pCND (n=183) | 8.3±4.3 | 10.8±6.7 | 0.105 |
No. of metastatic CLNs excised | 1.0±1.8 | 0.7±1.7 | <0.001 |
Unilateral pCND (n=662) | 0.8±1.6 | 0.5±1.4 | <0.001 |
Bilateral pCND (n=183) | 1.4±2.3 | 1.5±2.4 | 0.886 |
Central LNR (%) | 16.1±26.7 | 10.6±24.2 | <0.001 |
Unilateral pCND (n=662) | 14.7±25.6 | 8.8±22.4 | <0.001 |
Bilateral pCND (n=183) | 20.7±29.4 | 18.3±30.9 | 0.542 |
Stage of PTC by TNM | 0.008 | ||
Stage I | 374 (59.6) | 155 (71.4) | |
Stage II | 3 (0.5) | 1 (0.5) | |
Stage III | 251 (40.0) | 61 (28.1) | |
Postsurgical RAI ablation | 221 (35.2) | 49 (22.6) | 0.001 |
Pre-ablation | |||
TSH (mIU/l) | 99.3±92.2 | 91.3±59.1 | 0.539 |
sTg level (μg/l)a | 2.4±12.7 | 1.0±1.6 | 0.032 |
Post-ablation | |||
TSH (mIU/l) | 119.0±56.1 | 107.4±40.0 | 0.356 |
sTg level (μg/l)a | 0.6±1.8 | 0.2±0.1 | 0.473 |
Continuous variables are expressed as mean±s.d.; categorical variables are expressed as number (%). PTC, papillary thyroid carcinoma; HT, Hashimoto's thyroiditis; LV, lymphovascular; CLN, central lymph node; CNM, central nodal metastasis; pCND, prophylactic central neck dissection; LNR, lymph node ratio; TNM, 7th edition Tumor, Node and Metastasis staging system; RAI, radioactive iodine; TSH, thyroid-stimulating hormone; sTg, stimulated thyroglobulin; bold signifies statistical significance, P<0.05.
After excluding patients with elevated anti-thyroglobulin antibody.
Table 4 shows a comparison of patients' clinicopathologic features, tumor characteristics, and BRAF mutation status between those with (N1a group) and those without occult CNM (N0 group). Age was similar between the two groups, but the proportion of males was significantly higher in the N1a group (22.5 vs 8.6%, P=0.023). Also the N1a group had significantly larger sized tumors (0.8 vs 0.7 cm, P=0.001) and higher incidence of tumor multifocality (41.4 vs 31.1%, P=0.003), extrathyroidal extension (69.1 vs 50.0%, P<0.001), lymphovascular permeation (LVP) (20.4 vs 2.1%, P<0.001), and BRAF+ve mutation status (82.5 vs 70.2%, P<0.001). The CNM risk for tumors <1 and ≥1 cm was 175/633 (27.6%) and 110/212 (51.9%) respectively. However, the N1a group had significantly lower incidence of coexisting HT than the N0 group (26.7 vs 45.5%, P<0.001).
A comparison of patient clinicopathologic features and BRAF mutation status between those with occult central nodal metastases (N1a group) and those without occult central nodal metastases (N0 group)
N1a group (n=285) | N0 group (n=560) | P value | |
---|---|---|---|
Age at operation (years) | 45.8±11.9 | 45.6±11.8 | 0.285 |
Sex (Male:Female) | 64:221 | 48:512 | 0.023 |
Tumor characteristics | |||
Tumor size (cm) | 0.8±0.4 | 0.7±0.4 | 0.001 |
Tumor bilaterality | 66 (23.2) | 105 (18.8) | 0.099 |
Tumor multifocality | 118 (41.4) | 174 (31.1) | 0.003 |
Extra-thyroidal extension | 197 (69.1) | 280 (50.0) | <0.001 |
LV permeation | 58 (20.4) | 12 (2.1) | <0.001 |
Coexisting HT | 76 (26.7) | 255 (45.5) | <0.001 |
BRAF V600E mutation | 235 (82.5) | 393 (70.2) | <0.001 |
HT, Hashimoto's thyroiditis; LV, lymphovascular; bold signifies statistical significance, P<0.05.
Table 5 shows the multivariate analysis for occult CNM. Male sex (odds ratio (OR): 2.681, 95% CI: 1.709–4.202, P<0.001), large tumor size (OR: 2.684, 95% CI: 1.802–3.997, P<0.001), tumor multifocality (OR: 1.491, 95% CI: 1.065–2.087, P=0.020), LVP (OR: 10.395, 95% CI: 5.176–20.877, P<0.001), and BRAF+ve mutation (OR: 1.647, 95% CI: 1.101–2.463, P=0.015) were independent risk factors while coexisting HT (OR: 0.560, 95% CI: 0.396–0.792, P=0.001) was an independent protective factor for occult CNM.
A multivariable analysis of clinicopathological risk factors for occult central lymph node metastases (N1a)
Covariates | β-coefficient | Odds ratio (95% CI) | P value |
---|---|---|---|
Male sex | 0.986 | 2.681 (1.709–4.202) | <0.001 |
Tumor size | 0.987 | 2.684 (1.802–3.997) | <0.001 |
Tumor multifocality | 0.399 | 1.491 (1.065–2.087) | 0.020 |
Extrathyroidal extension | 0.248 | 1.282 (0.898–1.829) | 0.171 |
Lymphovascular permeation | 2.341 | 10.395 (5.176–20.877) | <0.001 |
Coexisting Hashimoto's thyroiditis | 0.580 | 0.560 (0.396–0.792) | 0.001 |
BRAF V600E mutation | 0.499 | 1.647 (1.101–2.463) | 0.015 |
Bold signifies statistical significance, P<0.05.
Because of only male sex, tumor size and BRAF+ve mutation are potentially known before operation (i.e., without histopathology), these three factors were used to formulate a preoperative prediction score by logistic regression. Table 6 shows a comparison of predictability as measured by area under the receiver operating characteristic curve (AUC) between tumor size and two combined prediction scores. Although the AUC of the three prediction scores was not significantly different, the most important was that the AUC for prediction score 3 (based on tumor size, male sex, and BRAF) was not significantly higher than that of prediction score 2 (based on tumor size and male sex) (0.69 vs 0.68, P=0.60). Therefore, despite being an independent predictor in the multivariate analysis (see Table 5), knowing the BRAF mutation status did not add substantially to the overall prediction of occult CNM. Table 7 shows a comparison of occult CNM rate between each quartile of prediction score 2 and 3. For both scores, the chance of occult CNM increased from <20 to 55% as the prediction score increased from the first to the fourth quartile.
A comparison of predictability of central nodal metastasis as measured by area under the receiver operating characteristic curve (AUC) between tumor size and combined preoperative prediction scores
AUC (95% CI) | P value score 1 vs 2 | P value score 2 vs 3 | P value score 1 vs 3 | |
---|---|---|---|---|
Prediction score 1 based on tumor size only | 0.65 (0.61–0.69) | 0.33 | – | – |
Prediction score 2 based on tumor size and male sex | 0.68 (0.64–0.72) | – | 0.60 | – |
Prediction score 3 based on tumor size, male sex and BRAF mutation | 0.69 (0.65–0.73) | – | – | 0.13 |
Calculated from logistic regression: Prediction score 1=−1.716+1.288×(tumor size in cm). Prediction score 2=−1.873+1.102 (male=1; female=0)+1.283×(tumor size in cm). Prediction score 3=−2.278+1.084 (male=1; female=0)+1.246×(tumor size in cm)+0.569 (BRAF+ve=1; BRAF−ve=0). The higher the prediction score corresponds to higher risk of occult central nodal metastasis.
A comparison of central nodal metastasis (CNM) rate for each quartile of prediction score 2 and 3
Prediction score 2a | CNM (%) | Prediction score 3a | CNM (%) | |
---|---|---|---|---|
1st quartile | 0.00–0.51 | 48/249 (19.3) | 0.00–0.94 | 40/228 (17.5) |
2nd quartile | 0.52–0.89 | 67/245 (27.3) | 0.95–1.31 | 65/233 (27.9) |
3rd quartile | 0.90–1.41 | 62/160 (38.8) | 1.32–1.81 | 64/177 (36.2) |
4th quartile | >1.42 | 108/191 (56.5) | >1.82 | 116/207 (56.0) |
Prediction score 2=−1.873+1.102 (male=1; female=0)+1.283×(tumor size in cm). Prediction score 3=−2.278+1.084 (male=1; female=0)+1.246×(tumor size in cm) +0.569 (BRAF+ve=1; BRAF−ve=0).
To avoid negative values and facilitate interpretation, +1.74 was added to each prediction score 2 while +2.15 was added to each prediction score 3. This makes no difference to the performance of the score.
Table 8 shows the correlation of BRAF mutation status between FNA and surgical specimen. Of the 19 patients, 17 had matched BRAF results while two had mismatched results. For these two mismatched cases, both were BRAF+ve on FNA but BRAF−ve on surgical specimen. The correlation rate between FNA and surgical specimen was 89.5%.
Correlation of BRAF mutation status between using fine-needle aspiration (FNA) materials and surgical specimen
Patient no. | Age at operation (years) | Sex (M/F) | Tumor size (cm) | Occult CNM (pN1a) | BRAF mutation | ||
---|---|---|---|---|---|---|---|
On FNA | On surgical specimen | Matching between FNA and surgical specimen | |||||
1 | 37 | F | 0.5 | Negative | Negative | Negative | Matched |
2 | 46 | F | 1.0 | Negative | Negative | Negative | Matched |
3 | 39 | F | 0.6 | Positive | Positive | Positive | Matched |
4 | 46 | F | 0.5 | Negative | Positive | Positive | Matched |
5 | 50 | F | 0.9 | Negative | Positive | Positive | Matched |
6 | 73 | F | 0.6 | Negative | Negative | Negative | Matched |
7 | 54 | F | 0.9 | Positive | Negative | Negative | Matched |
8 | 68 | F | 0.4 | Negative | Positive | Negative | Mismatched |
9 | 31 | F | 1.2 | Positive | Positive | Positive | Matched |
10 | 50 | M | 0.5 | Negative | Positive | Positive | Matched |
11 | 39 | F | 0.4 | Negative | Negative | Negative | Matched |
12 | 55 | F | 0.6 | Negative | Positive | Positive | Matched |
13 | 57 | F | 0.3 | Negative | Positive | Positive | Matched |
14 | 63 | F | 0.3 | Negative | Negative | Negative | Matched |
15 | 30 | F | 0.5 | Positive | Positive | Positive | Matched |
16 | 34 | F | 0.6 | Positive | Positive | Negative | Mismatched |
17 | 55 | F | 0.3 | Negative | Positive | Positive | Matched |
18 | 44 | F | 0.3 | Negative | Negative | Negative | Matched |
19 | 50 | F | 2.0 | Positive | Positive | Positive | Matched |
Table 9 shows a 2×2 table between BRAF mutation and CNM. The sensitivity and specificity of BRAF+ve mutation status in predicting occult CNM were 235/285 (82.5%) and 167/560 (29.8%), respectively, while the positive (PPV) and negative predictive values (NPV) were 235/628 (37.4%) and 167/217 (77.0%), respectively. When the BRAF prevalence was lowered to 40%, the specificity increased to 64.8% (see Supplementary Table 1, see section on supplementary data given at the end of this article).
A 2×2 table between BRAF mutation and central nodal metastasis (CNM)
CNM+ve | CNM−ve | Total | |
---|---|---|---|
BRAF+ve | 235 | 393 | 628 |
BRAF−ve | 50 | 167 | 217 |
Total | 285 | 560 | 845 |
Based on these data, the sensitivity, specificity, PPV, and NPV of BRAF were 82.5, 29.8, 37.4, and 77.0% respectively.
In terms of clinical outcomes, rates of temporary and permanent hypocalcemia were 32.7 and 1.9%, respectively, while temporary and permanent RLN injury were 8.9 and 1.4%. After a mean follow-up of 9.4±5.4 months, there was no LR detected.
Discussion
The optimal initial surgical management for PTC patients without preoperative or intraoperative evidence of nodal involvement (i.e., cN0 PTC) remains controversial as the ATA currently only recommends CND for those with cN1 PTC. However, as pCND may reduce LR in the short-term (Lang et al. 2013), a more selective approach to minimize overall surgical morbidity would seem sensible and perhaps, cost-saving in the long-term (Lang & Wong 2013, Lang et al. 2013). It is worth noting that although our cohort comprised patients with no evidence of clinical or ultrasound evidence of CNM, the presence of occult CNM was still 33.7%. This finding is of interest because of the recent discussions on whether pCND is justified and on whether RAI should be given more selectively (Cooper et al. 2009). In terms of surgical morbidity, our rates of hypocalcemia and RLN injury after pCND were not significantly higher or different from our previous series without pCND performed (Chung et al. 2007) and were comparable with the literature (Lang et al. 2013).
To our knowledge, this is one of the largest studies examining the association between BRAF mutation and occult CNM in cN0 PTC. To ensure that BRAF was truly a preoperative rather than a postoperative predictor, a small proof of principle series of 19 FNA cases was conducted and showed an 89.5% concordance of BRAF between FNA samples and surgical specimens. Similar to previous studies (Frasca et al. 2008, Xing et al. 2009, Li et al. 2012), our data confirmed that the BRAF+ve group had significantly larger, more advanced, and aggressive tumors than the BRAF−ve group. It was interesting to find that the BRAF+ve group had significantly less coexisting HT on histology (34.6 vs 52.5%, P<0.001). This finding appeared to concur with previous studies which found reduced peritumoral lymphocytic infiltration in BRAF+ve PTCs (Sargent et al. 2006, Virk et al. 2013). Although the precise reason for this remains unclear, a recent study demonstrated that tumors with coexisting HT behaved less aggressively and had a better prognosis than those without coexisting HT (Dvorkin et al. 2013). Therefore, this inverse association was in keeping with the concept that BRAF+ve tumor behaved more aggressively. Our data also showed that the pre-ablation sTg level was significantly higher in the BRAF+ve group, implying that the risk of microscopic residual disease after a TT with pCND might still have been higher in the BRAF+ve group. Nevertheless, the post-ablation sTg was similar and so, a longer follow-up was necessary to evaluate its true impact of BRAF on survival outcomes. However, unlike other studies, our study did not find significant association between age, sex, tumor bilaterality and multifocality with BRAF mutation (Kim et al. 2006, Li et al. 2012, Nam et al. 2012).
In terms of predicting occult CNM, male sex, tumor size, tumor multifocality, LVP, coexisting HT, and BRAF mutation were independent risk factors by multivariate analysis. Furthermore, a doubling of CNM risk was observed when tumor size increased from <1 to ≥1 cm, implying that a 1 cm cut-off for routine pCND might appear reasonable. Although two large previous studies also reported similar findings, neither examined the role of BRAF in the context of other significant clinicopathological factors (So et al. 2011, Zhang et al. 2012). Paulson et al. (2012) reported their experience of 175 classic cN0 PTC but found no association between BRAF mutation and occult CNM. Two similarly designed but smaller studies also did not find any significant association between BRAF mutation and occult CNM (Lee et al. 2012, Dutenhefner et al. 2013). In fact, in one of the studies, the authors went further and concluded that it was premature in utilizing BRAF mutation status to decide whether or not to perform pCND in cN0 PTC (Lee et al. 2012). In contrast to these previous studies, although we did find that BRAF mutation status (OR: 1.65, 95% CI: 1.101–2.463) was an independent predictor of occult CNM in cN0 PTC, it did not contribute significantly to the overall prediction. When formulating preoperative prediction scores using male sex, tumor size, and BRAF+ve mutation, although the predictability (as measured by AUC) improved with each additional factor entered into the prediction score (i.e., from prediction score 1–3), the improvement in predicting occult CNM was not statistically significant. Our data found that using a simpler prediction score of tumor size and male sex alone, the prediction (as measured AUC) was similar to a more complicated prediction score of tumor size, male sex, and BRAF mutation (0.68 vs 0.69, P=0.60). Given the fact that BRAF testing is associated with extra cost, perhaps a simpler prediction score based on male and tumor size might be sufficient. Therefore, although BRAF mutation was an independent predictor for occult CNM, it did not substantially or significantly improve the overall prediction of occult CNM in cN0 patients. Despite the high pre-test probability (74.3%) of BRAF positivity, both the specificity (29.8%) and PPV (37.4%) were relatively low and so these further emphasized the fact that BRAF mutation was not useful in predicting CNM in small cN0 PTC.
However, it is worth noting that based on the adjusted OR, the BRAF+ve tumor in our study only had a 1.6–1.7 times greater chance of harboring occult CNM than a BRAF−ve tumor whereas to date, two other studies which found significant association had almost twice as high adjusted OR values (Joo et al. 2012, Howell et al. 2013). Perhaps, in these studies, BRAF mutation might have a more significant impact on the overall prediction. Also we would like to acknowledge several shortcomings. Firstly, this was a retrospective analysis and so was prone to selection biases. Secondly, although our series of 19 FNA cases did show an 89.5% correlation between FNA samples and surgical specimens, our study was principally based on paraffin-embedded sections after thyroidectomy and so our results might be slightly different from studies which tested BRAF mutation primarily from FNA samples. Therefore, our study could not be strictly considered to be examining the association between preoperative BRAF mutation and occult CNM. Nevertheless, even assuming that our study was entirely based on FNA samples, our conclusion would not have changed because this would have further lowered the predictability of BRAF mutation due to some discordance between FNA and surgical specimen (see Table 8). Thirdly, due to the strict definition of pCND, over a third of patients with an inadequate number of CLNs had to be excluded from analysis. Although excluding such substantial numbers of patients may introduce selection bias, the comparison of patient/tumor characteristics between those with ≥3 CLNs (n=845) and with <3 CLNs (n=446) did not reveal significant differences (Table 2). The only differences were those with ≥3 CLNs had significantly higher percentages of coexisting HT and CNM than those with <3 CLNs. The former finding could be explained by the fact that HT tended to have larger-sized CLNs and that led to higher CLN yield (Hartl et al. 2012) while the latter finding was probably due to inadequate nodes sampled and nodal under-staging (Lang et al. 2007b, 2012). Lastly, we would like to highlight the fact that our overall prevalence of BRAF positivity was relatively high (74.3%) when compared with that of other studies when only classical PTC were considered (approximately 45%) (Lee et al. 2012, Xing et al. 2013b). This is particularly interesting given the fact that these patients had small cN0 PTC. Although by including only the classical subtype of PTC did increase the overall prevalence of BRAF positivity from 72.9 to 74.3%, this increase was small because these variants only accounted for 5.8% of the entire cohort (see Table 1). Therefore, the exact reason for such high prevalence of BRAF positivity in our cohort remains unclear and may be due to geographical, genetic, or diet-linked factors, as suggested previously (Frasca et al. 2008). However, it is worth noting that in our locality, the prevalence of BRAF positivity has been reported to be much higher (60–70%) than other parts of the world (Chung et al. 2006, So et al. 2011) and so this was unlikely due to a selection or institutional bias. When the prevalence of BRAF mutation was lowered, our data showed that only the sensitivity and specificity of BRAF reversed while PPV and NPV remained static (see Table 9 and Supplementary Table 1). Although the absolute risk predicted by our model (Table 7) may differ slightly with lower BRAF mutation prevalence, we think that the increased risk of occult CNM associated with BRAF should be generalizable. However, we would acknowledge the applicability of BRAF mutation as a marker to reduce unnecessary pCND could be weakened due to the high prevalence of BRAF positivity in our cohort. Nevertheless, this was one of the largest studies aimed at examining the association between BRAF mutation and occult CNM in small cN0 PTC.
Conclusion
Among the cN0 PTC patients who underwent pCND, the BRAF+ve tumors were significantly larger in size, had more extrathyroidal extension, occult CNM, higher CLNR, pre-ablation sTg level but less coexisting HT than the BRAF−ve tumors. Male sex, large tumor size, tumor multifocality, LV permeation, and BRAF mutation were significant independent predictors of occult CNM while coexisting HT was a significant independent protective factor. When BRAF mutation was entered into logistic regression to formulate a prediction score, that score was not significantly better than that of a prediction score based on male and tumor size only. Therefore, based on our analysis using primarily paraffin-embedded tissue, despite being an independent predictor of CNM, BRAF did not add substantially to the overall prediction of occult CNM. Given the extra cost associated with BRAF testing, a simpler prediction score based on male and tumor size might be sufficient.
Supplementary data
This is linked to the online version of the paper at http://dx.doi.org/10.1530/ERC-13-0291.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
Funding
This study was supported by the Research Grant Number CB-2011-03-01 of the Korean Foundation for Cancer Research.
Author contribution statement
B H-H Lang, Y J Chai, B J Cowling, K E Lee, H S Min were involved in the review of literature, acquisition of data, and drafting and completing the manuscript. B H-H Lang, Y J Chai, B J Cowling, K E Lee, H S Min were also involved in the review of literature and drafting the manuscript. B H-H Lang, Y J Chai, B J Cowling, K E Lee, H S Min conceived the study, participated in the co-ordination and the acquisition of data, and helped to draft the manuscript. All authors read and approved the final manuscript.
References
Alzahrani AS & Xing M 2013 Impact of lymph node metastases identified on central neck dissection (CND) on the recurrence of papillary thyroid cancer: potential role of BRAFV600E mutation in defining CND. Endocrine-Related Cancer 20 13–22. (doi:10.1530/ERC-12-0309).
Carty SE, Cooper DS, Doherty GM, Duh QY, Kloos RT, Mandel SJ, Randolph GW, Stack BC Jr, Steward DL & Terris DJ et al. 2009 Consensus statement on the terminology and classification of central neck dissection for thyroid cancer. Thyroid 19 1153–1158. (doi:10.1089/thy.2009.0159).
Chung KW, Yang SK, Lee GK, Kim EY, Kwon S, Lee SH, Park do J, Lee HS, Cho BY & Lee ES et al. 2006 Detection of BRAFV600E mutation on fine needle aspiration specimens of thyroid nodule refines cyto-pathology diagnosis, especially in BRAF600E mutation-prevalent area. Clinical Endocrinology 65 660–666. (doi:10.1111/j.1365-2265.2006.02646.x).
Chung YS, Choe JH, Kang KH, Kim SW, Chung KW, Park KS, Han W, Noh DY, Oh SK & Youn YK 2007 Endoscopic thyroidectomy for thyroid malignancies: comparison with conventional open thyroidectomy. World Journal of Surgery 31 2302–2306. (doi:10.1007/s00268-007-9117-0).
Cooper DS, Doherty GM, Haugen BR, Kloos RT, Lee SL, Mandel SJ, Mazzaferri EL, McIver B, Pacini F & Schlumberger M et al. 2009 Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 19 1167–1214. (doi:10REF7=10.1210/jc.2013-1309).
Dutenhefner SE, Marui S, Santos ABO, de Lima EU, Inoue M, Neto JS Brandão, Shiang C, Fukushima JT, Cernea CR & Friguglietti CUM 2013 BRAF, a tool in the decision to perform elective neck dissection? Thyroid 23 1541–1546. (doi:10.1089/thy.2012.0304).
Dvorkin S, Robenshtok E, Hirsch D, Strenov Y, Shimon I & Benbassat CA 2013 Differentiated thyroid cancer is associated with less aggressive disease and better outcome in patients with coexisting Hashimotos thyroiditis. Journal of Clinical Endocrinology and Metabolism 98 2409–2414. (doi:10.1210/jc.2013-1309).
Frasca F, Nucera C, Pellegriti G, Gangemi P, Attard M, Stella M, Loda M, Vella V, Giordano C & Trimarchi F et al. 2008 BRAF(V600E) mutation and the biology of papillary thyroid cancer. Endocrine-Related Cancer 15 191–205. (doi:10.1677/ERC-07-0212).
Hartl DM, Leboulleux S, Al Ghuzlan A, Baudin E, Chami L, Schlumberger M & Travagli JP 2012 Optimization of staging of the neck with prophylactic central and lateral neck dissection for papillary thyroid carcinoma. Annals of Surgery 255 777–783. (doi:10.1097/SLA.0b013e31824b7b68).
Hong Kong Cancer Registry. Cancer incidence and mortality in Hong Kong 1983–2010. Available: http://www3.ha.org.hk/cancereg/e_stat.asp [Accessed on 29th June 2013]
Howell GM, Nikiforova MN, Carty SE, Armstrong MJ, Hodak SP, Stang MT, McCoy KL, Nikiforov YE & Yip L 2013 BRAFV600E mutation independently predicts central compartment lymph node metastasis in patients with papillary thyroid cancer. Annals of Surgical Oncology 20 47–52. (doi:10.1245/s10434-012-2611-0).
Hwang HS & Orloff LA 2011 Efficacy of preoperative neck ultrasound in the detection of cervical lymph node metastasis from thyroid cancer. Laryngoscope 121 487–491. (doi:10.1002/lary.21227).
Joo JY, Park JY, Yoon YH, Choi B, Kim JM, Jo YS, Shong M & Koo BS 2012 Prediction of occult central lymph node metastasis in papillary thyroid carcinoma by preoperative BRAF analysis using fine-needle aspiration biopsy: a prospective study. Journal of Clinical Endocrinology and Metabolism 97 3996–4003. (doi:10.1210/jc.2012-2444).
Kilfoy BA, Zheng T, Holford TR, Han X, Ward MH, Sjodin A, Zhang Y, Bai Y, Zhu C & Guo GL et al. 2009 International patterns and trends in thyroid cancer incidence, 1973–2002. Cancer Causes & Control 20 525–531. (doi:10.1007/s10552-008-9260-4).
Kim TY, Kim WB, Rhee YS, Song JY, Kim JM, Gong G, Lee S, Kim SY, Kim SC & Hong SJ et al. 2006 The BRAF mutation is useful for prediction of clinical recurrence in low-risk patients with conventional papillary thyroid carcinoma. Clinical Endocrinology 65 364–368.
Kim SJ, Lee KE, Myong JP, Park JH, Jeon YK, Min HS, Park SY, Jung KC, Koo do H & Youn YK 2012 BRAF V600E mutation is associated with tumor aggressiveness in papillary thyroid cancer. World Journal of Surgery 36 310–317. (doi:10REF19=10.1097/01.sla.0000262785.46403.9b).
Koo BS, Choi EC, Yoon YH, Kim DH, Kim EH & Lim YC 2009 Predictive factors for ipsilateral or contralateral central lymph node metastasis in unilateral papillary thyroid carcinoma. Annals of Surgery 249 840–844. (doi:10.1097/SLA.0b013e3181a40919).
Lang BH & Wong CK A cost-minimization analysis comparing total thyroidectomy alone and total thyroidectomy with prophylactic central neck dissection in clinically nodal-negative papillary thyroid carcinoma Annals of Surgical Oncology 2013 August 24. [Epub ahead of print] doi:10.1245/s10434-013-3234-9).
Lang BH, Chow SM, Lo CY, Law SC & Lam KY 2007a Staging systems for papillary thyroid carcinoma: a study of 2 tertiary referral centers. Annals of Surgery 246 114–121. (doi:10.1097/01.sla.0000262785.46403.9b).
Lang B, Lo CY, Chan WF, Lam KY & Wan KY 2007b Restaging of differentiated thyroid carcinoma by the sixth edition AJCC/UICC TNM staging system: stage migration and predictability. Annals of Surgical Oncology 14 1551–1559. (doi:10.1245/s10434-006-9242-2).
Lang BH, Yih PC, Shek TW, Wan KY, Wong KP & Lo CY 2012 Factors affecting the adequacy of lymph node yield in prophylactic unilateral central neck dissection for papillary thyroid carcinoma. Journal of Surgical Oncology 106 966–971. (doi:10.1002/jso.23201).
Lang BH, Ng SH, Lau L, Cowling B, Wong KP & Wan KY 2013 A systematic review and meta-analysis of prophylactic central neck dissection on short-term locoregional recurrence in papillary thyroid carcinoma after total thyroidectomy. Thyroid 23 1087–1098. (doi:10.1089/thy.2012.0608).
Lee KC, Li C, Schneider EB, Wang Y, Somervell H, Krafft M, Umbricht CB & Zeiger MA 2012 Is BRAF mutation associated with lymph node metastasis in patients with papillary thyroid cancer? Surgery 152 977–983. (doi:10.1016/j.surg.2012.08.019).
Li C, Lee KC, Schneider EB & Zeiger MA 2012 BRAF V600E mutation and its association with clinicopathological features of papillary thyroid cancer: a meta-analysis. Journal of Clinical Endocrinology and Metabolism 97 4559–4570. (doi:10.1210/jc.2012-2104).
Machens A, Hauptmann S & Dralle H 2009 Lymph node dissection in the lateral neck for completion in central node-positive papillary thyroid cancer. Surgery 145 176–181. (doi:10.1016/j.surg.2008.09.003).
Nam JK, Jung CK, Song BJ, Lim DJ, Chae BJ, Lee NS, Park WC, Kim JS, Jung SS & Bae JS 2012 Is the BRAF(V600E) mutation useful as a predictor of preoperative risk in papillary thyroid cancer? American Journal of Surgery 203 436–441. (doi:10.1016/j.amjsurg.2011.02.013).
O'Neill CJ, Bullock M, Chou A, Sidhu SB, Delbridge LW, Robinson BG, Gill AJ, Learoyd DL, Clifton-Bligh R & Sywak MS 2010 BRAF(V600E) mutation is associated with an increased risk of nodal recurrence requiring reoperative surgery in patients with papillary thyroid cancer. Surgery 148 1139–1145. (doi:10.1016/j.surg.2010.09.005REF33=10.1038/modpathol.2012.152).
Paulson L, Shindo M, Schuff K & Corless C 2012 The role of molecular markers and tumor histological type in central lymph node metastasis of papillary thyroid carcinoma. Archives of Otolaryngology – Head & Neck Surgery 138 44–49. (doi:10.1001/archoto.2011.226).
Roh JL, Park JY, Kim JM & Song CJ 2009 Use of preoperative ultrasonography as guidance for neck dissection in patients with papillary thyroid carcinoma. Journal of Surgical Oncology 99 28–31. (doi:10.1002/jso.21164).
Sargent R, LiVolsi V, Murphy J, Mantha G & Hunt JL 2006 BRAF mutation is unusual in chronic lymphocytic thyroiditis-associated papillary thyroid carcinomas and absent in non-neoplastic nuclear atypia of thyroiditis. Endocrine Pathology 17 235–241. (doi:10.1385/EP:17:3:235).
SEER Cancer Statistics Review, 1975–2010. National Cancer Institute Surveillance Epidemiology and End Results [online]. Available: [Assessed on 29th June 2013]
So YK, Son YI, Park JY, Baek CH, Jeong HS & Chung MK 2011 Preoperative BRAF mutation has different predictive values for lymph node metastasis according to tumor size. Otolaryngology – Head and Neck Surgery 145 422–427. (doi:10.1177/0194599811404649).
Virk RK, Van Dyke AL, Finkelstein A, Prasad A, Gibson J, Hui P, Theoharis CG, Carling T, Roman SA & Sosa JA et al. 2013 BRAFV600E mutation in papillary thyroid microcarcinoma: a genotype–phenotype correlation. Modern Pathology 26 62–70. (doi:10.1038/modpathol.2012.152).
Wong H, Wong KP, Yau T, Tang V, Leung R, Chiu J & Lang BH 2012 Is there a role for unstimulated thyroglobulin velocity in predicting recurrence in papillary thyroid carcinoma patients with detectable thyroglobulin after radioiodine ablation? Annals of Surgical Oncology 19 3479–3485. (doi:10.1245/s10434-012-239REF41=10.1245/s10434-012-2654-2).
Xing M, Westra WH, Tufano RP, Cohen Y, Rosenbaum E, Rhoden KJ, Carson KA, Vasko V, Larin A & Tallini G et al. 2005 BRAF mutation predicts a poorer clinical prognosis for papillary thyroid cancer. Journal of Clinical Endocrinology and Metabolism 90 6373–6379. (doi:10.1210/jc.2005-0987).
Xing M, Clark D, Guan H, Ji M, Dackiw A, Carson KA, Kim M, Tufaro A, Ladenson P & Zeiger M et al. 2009 BRAF mutation testing of thyroid fine-needle aspiration biopsy specimens for preoperative risk stratification in papillary thyroid cancer. Journal of Clinical Oncology 27 2977–2982. (doi:10.1200/JCO.2008.20.1426).
Xing M, Haugen BR & Schlumberger M 2013a Progress in molecular-based management of differentiated thyroid cancer. Lancet 381 1058–1069. (doi:10.1016/S0140-6736(13)60109-9).
Xing M, Alzahrani AS, Carson KA, Viola D, Elisei R, Bendlova B, Yip L, Mian C, Vianello F & Tuttle RM et al. 2013b Association between BRAF V600E mutation and mortality in patients with papillary thyroid cancer. Journal of the American Medical Association 309 1493–1501. (doi:10.1001/jama.2013.3190).
Yip L, Nikiforova MN, Carty SE, Yim JH, Stang MT, Tublin MJ, Lebeau SO, Hodak SP, Ogilvie JB & Nikiforov YE 2009 Optimizing surgical treatment of papillary thyroid carcinoma associated with BRAF mutation. Surgery 146 1215–1223. (doi:10.1016/j.surg.2009.09.011).
Zhang L, Wei WJ, Ji QH, Zhu YX, Wang ZY, Wang Y, Huang CP, Shen Q, Li DS & Wu Y 2012 Risk factors for neck nodal metastasis in papillary thyroid microcarcinoma: a study of 1066 patients. Journal of Clinical Endocrinology and Metabolism 97 1250–1257. (doi:10.1210/jc.2011-1546).
Zhao Q, Ming J, Liu C, Shi L, Xu X, Nie X & Huang T 2013 Multifocality and total tumor diameter predict central neck lymph node metastases in papillary thyroid microcarcinoma. Annals of Surgical Oncology 20 746–752. (doi:10.1245/s10434-012-2654-2).