As often as 30% of the time, FNA cytology displays limited ability to discriminate between benign and malignant thyroid lesions, and an indeterminate diagnosis is rendered [31
]. Some clinicians feel that total thyroidectomy is appropriate for patients with an indeterminate FNA cytology result [9
]. Proponents of this approach argue that it eliminates the probability of thyroid cancer recurrence [33
]. Alternatively, if the suspect nodule is small, some clinicians opt to perform a hemi-thyroidectomy procedure following an indeterminate cytology result [37
]. Considered the minimum extent of surgery for a thyroid nodule, this procedure reduces the risk of postoperative complications associated with total thyroidectomy, such as hypoparathyroidism and laryngeal nerve injury [38
]. However, depending on patient/tumor risk stratification, postoperative confirmation of malignancy usually results in a second-stage completion thyroidectomy, which is associated with higher morbidity than initial total thyroidectomy [39
]. Regardless of the choice of surgical procedure, the incidence of malignancy in patients with indeterminate cytology findings varies greatly. Consequently, a large percentage of these patients would benefit from a method that improves the diagnosis of preoperative thyroid aspirate material.
A number of molecular markers have been evaluated as possible adjunct tests for refining the diagnosis of PTC on FNA. However, the predictive value of these markers has been limited to date due to a lack of specificity or sensitivity [40
]. Expressed by malignant thyrocytes, galectin 3 is a β-galactoside-binding protein that was initially believed to be a marker specific for PTC [41
]. Further analysis has revealed, however, that identification of this protein may be less reliable in conditions involving lymphocytic infiltration, such as Hashimoto's thyroiditis [42
]. HBME-1 is a monoclonal antibody that recognizes an epitope expressed in malignant mesothelioma and other adenocarcinomas, as well as PTC and follicular thyroid tumors [43
]. Although benign thyroid lesions do not express immunoreactivity for HBME-1 [44
], positive staining has been found in malignant thyroid tumors besides those of papillary differentiation [45
]. Telomerase is a specialized reverse transcriptase enzyme that maintains chromosome ends. Detection of telomerase expression by reverse-transcriptase PCR originally showed promising sensitivity and specificity for PTC diagnosis [46
], but telomerase repeat amplification (TRAP) has identified high expression of this molecule in FNA specimens from benign nodules [47
]. Up to 95% of PTC demonstrate strong immunostaining with cytokeratin 19 (CK19) [48
]. However, CK19 immunoreactivity is not specific for PTC, as positive immunoreactivity has been identified in benign follicular adenomas [49
]. Aberrant expression of the RET proto-oncogene results from chromosomal rearrangements in which the tyrosine kinase domain of RET is fused to the 5'-terminal region of an unrelated gene, leading to the generation of fusion proteins known as ret rearrangements in PTC (RET/PTC). Although RET/PTC rearrangements have been identified in a large percentage of PTC in individuals exposed to external radiation [50
], a relatively high frequency of RET/PTC rearrangements have also been found in benign nodular thyroid diseases of patients exposed to nuclear fallout and in benign conditions such as trabecular adenomas and Hashimoto's thyroiditis [51
The mutation at V600E in the BRAF kinase gene appears to be an attractive molecular marker for thyroid cancer diagnosis as it has been found to be the most common genetic event in PTC, while being highly specific for this tumor (Table ). The goal of this study was to identify the BRAF V600E mutation in thyroid FNA samples in an attempt to determine if BRAF mutation analysis can serve as a useful adjunct technique in indeterminate cytologic diagnoses. The use of LCPCR was chosen for SNP detection because both gene amplification and allele analysis could be performed in a homogeneous, closed-tube system on the same instrument. Increased specificity for mutation detection is realized due to the hybridization of two independent probes and the fact that the probe melting temperature is sequence specific. Of the cases evaluated in this study, which included 24 matched pairs of FNA and FFPE surgical tissues, there were no false positive BRAF mutation results by LCPCR and LCPCR demonstrated 100% concordance with DNA sequencing results.
BRAF V600E mutation prevalence in various thyroid neoplasms.
Previous investigators have also reported on use of the LCPCR method for detecting the V600E activating point mutation in the BRAF
gene. In contrast to the present study, which analyzed archival FNA and FFPE surgical material for mutation detection, these groups utilized primarily cell lines or a combination of cell lines and FFPE surgical material. Nikiforova et al. evaluated thyroid tumors and anaplastic carcinoma cell lines to demonstrate that BRAF
mutations, which in thyroid tumors were originally thought to be restricted to papillary carcinomas, also occur in poorly differentiated and anaplastic carcinomas [23
]. Among 259 thyroid tumor samples screened, this group demonstrated a 100% correlation in BRAF
V600E detection rate between LCPCR and single strand conformational polymorphism. This finding is in agreement with the results of the present study, in which LCPCR assay demonstrated 100% concordance between melting curve and DNA sequencing results. In contrast to the present study, however, Nikiforova used laser capture microdissection (LCM) to obtain DNA either from a small focus of papillary microcarcinoma or to study well-differentiated and poorly differentiated or anaplastic areas within the same tumor. Ikenoue et al. analyzed 12 colon and 9 gastric cancer cell lines by LCPCR for presence of the BRAF
V600E mutation [52
]. Using a mixture of standard DNA, Ikenoue determined that as little as 10% V600E mutant DNA could be identified in a background of WT DNA. In the present study, results of the limit of detection experiments confirmed that the 1 base pair change in the BRAF
mutation was detectable down to the level of 25% tumor when a homozygous mutant cell line was used as a control. Consequently, it was determined that the results of LCPCR for detection of the heterozygous BRAF
V600E mutation in FNA or FFPE samples containing less than 50% tumor cells may not be accurate. It is s possible that the level of sensitivity might be increased through the use of a technique such as LCM of the archival FNA slide material. As demonstrated by Nikiforova, et al, using LCM can significantly enhance the sensitivity for identifying mutant DNA in the presence of WT DNA, as the captured sample contains almost exclusively tumor cells.
In the present study, the preoperative diagnosis of PTC was confirmed in 3/19 (15.8%) indeterminate FNA samples that could not be conclusively diagnosed by cytology alone. This finding is consistent with reported mutation prevalence rates in indeterminate thyroid FNA cases (Table ) [27
]. However, 9/19 (47.4%) corresponding FFPE surgical samples collected from the same patients were positive for the mutation, for a 69% rate of concordance between the sample types. Of the 6 discordant FNA cases in this study, 5/6 (83%) contained <50% atypical cells. Because tumor DNA from FNA samples is invariably contaminated with the WT allele of the gene in question, the somatically mutated allele can be difficult to distinguish. This results in reduced sensitivity for identifying mutant DNA in the presence of WT DNA. Results of the limit of detection experiments confirmed that the one base pair change in the BRAF
mutation was detectable by LCPCR down to the level of 25% tumor when a homozygous mutant cell line was used as a control. Because the BRAF
V600E mutation in PTC is heterozygous, a detection limit of 50% atypical cells was established for the LCPCR assay, below which the results of LCPCR might not be accurate. Because LCPCR and DNA sequencing results were in 100% agreement for all samples, it is likely that the percentage of BRAF
mutant DNA in the discordant FNA samples was below the limit of detection for both methods.
BRAF V600E mutation prevalence rates in indeterminate thyroid FNA cases
Other investigators have also experienced varying degrees of concordance for BRAF
mutation status between matched FNA and FFPE samples. Cohen et al. noted discordant results in 3/49 matched pair samples for a 94% rate of concordance between the sample types [53
]. Of the three discordant FNA samples in Cohen's study, the mutation was not detected in 2 FNAs while the resected tumors harbored the mutation. In these 2 cases, the FNA material was found to be sparsely cellular. These previous findings, combined with those of the present study, confirm that overall tumor cell content of the FNA sample is critical for mutation detection, whether by LCPCR or another method.
If the results of the limit of detection experiments in the present study were used to select samples for mutation analysis, thereby excluding from analysis any FNA sample that contained <50% atypical cells, 2/8 indeterminate FNA samples would have been found to harbor the BRAF mutation. The resulting BRAF positivity rate in the indeterminate FNA samples would, therefore, be 25% rather than 15.8%. In actual clinical practice, the pathologist would control pre-analytic probability by carefully selecting which indeterminate thyroid FNAs would be referred for mutation analysis, based upon the cellular composition of the smears and/or the needle rinse pellet.
It is also possible that the discordant aspirate material in the present study contained only the BRAF WT genotype, and that the mutant DNA identified in the resected tumor samples was the result of an early tumorigenic event. However, this is an unlikely scenario, as inclusion criteria allowed cases to be selected for mutation analysis only if the cytology was reported as abnormal within a six-month period preceding the histology report. This restriction was instituted in an attempt to reduce mutation status discordance between the sample types. Five of the six discordant FNA cases were collected within one month prior to the FFPE surgical tissue.
Because PTC is frequently multifocal, there has been speculation regarding whether noncontiguous tumor foci arise from intraglandular metastases from a single primary tumor or originate as unrelated clones derived from independent tumors. A recent study by Shattuck et al. [55
] used PCR to evaluate the patterns of X chromosome inactivation of multiple distinct foci of PTC from 17 women. Discordant patterns indicative of independent origins were identified in tumors from 5 patients, leading to the conclusion that individual tumor foci in patients with multifocal PTC often arise as independent tumors. This finding could explain why an indeterminate FNA containing >75% atypical cells in the present study failed to demonstrate the mutation, while the resected tumor was positive for the mutation. The possibility exists that the tumor identified in the resected tissue, and in the nodule sampled by FNA, did not share the same clonal origin.
While the hotspot mutation at V600E is the most common genetic event in PTC, recent studies have demonstrated that up to 9% of follicular variant of PTC (FVPTC) cases demonstrate a mutation in codon 601 of the BRAF
gene, resulting in the substitution of lysine with glutamate (K601E) [25
]. Although five of the indeterminate thyroid FNA cases in the present study were classified histologically as FVPTC, none were found to contain the K601E mutation. However, because the hybridization probes were designed as a perfect match to the BRAF
WT genotype, other mutations covered by the probe would likely lead to a different temperature profile and probable detection of the K601E mutation.
Because the number of samples analyzed in the present study was limited, and the tumor cell content of some of the archival samples was scant, we were unable to determine the full diagnostic utility of LCPCR detection of the BRAF activating point mutation on indeterminate thyroid aspirates. Although molecular techniques such as LCPCR may be useful for refining a diagnosis of PTC, the absence of a BRAF mutation does not exclude the possibility of a malignant condition. In view of the large number of palpable thyroid nodules that require evaluation by FNA, a search for molecular markers such as BRAF may have clinical utility.
The results suggest that detection of BRAF mutation in thyroid aspirates may enhance the accuracy of FNA and refine preoperative diagnosis of PTC.