Role of BRAFV600E mutation analysis and second cytologic review of fine-needle aspiration for evaluating thyroid nodule
Thyroid fine-needle aspiration (FNA) is the primary diagnostic tool used for the evaluation of thyroid nodules. Although most aspirates provide diagnostic cytology, some are classified as indeterminate. The aim of this study was to determine whether the second review of FNA cytology can improve the diagnostic values and to assess the role of proto-oncogene B-Raf (BRAF) mutation testing in the diagnosis of papillary carcinoma (PC).
Thyroid aspirates from 1060 patients were submitted for cytologic evaluation and BRAFV600E mutation analysis. A second cytologic review was performed by 2 cytopathologists in light of the mutation status.
Of the 313 patients who received surgery, 200 (63.9%) had been initially diagnosed as malignant by cytology. They were surgically confirmed as PCs, and the BRAFV600E mutation was detected in 82.5% of the cases. Ninety-five of 102 cases (93.1%) with indeterminate cytology turned out to be malignant, and the mutation was present in 63.3% of PCs. The sensitivity, accuracy, and negative predictive value (NPV) of the second review were better than those of initial cytologic diagnosis (P <.001). The addition of the mutation analysis significantly increased the sensitivity, accuracy, and NPV for detecting PCs compared with those of cytology alone.
Qualified cytologic diagnosis increases the effectiveness of FNA, forgoing the need for repeat biopsy or intraoperative frozen section evaluation. Preoperative BRAF mutation testing can supplement the routine cytology in the selection of thyroid nodules for surgery. Cancer (Cancer Cytopathol) 2012. © 2011 American Cancer Society.
Fine-needle aspiration (FNA) has been accepted as a primary diagnostic tool for the evaluation of thyroid nodules in euthyroid patients. The main role of FNA is to triage patients with malignant nodule for timely clinical intervention and to reduce unnecessary surgery for patients with benign nodule.
However, the terminology for thyroid FNA has not been standardized, and diagnostic categories and criteria vary from one institute to another and even between individual pathologists at the same institute. Interobserver variability is a well-established limitation, especially for indeterminate cytology (ie, not clearly benign or malignant).1 The lack of uniformity in reporting may create confusion to referring clinicians in some instances, can result in inconsistent patient management, and also hinder data sharing among institutes. In an effort to establish standard terminology, including the assessment of specimen adequacy and the diagnostic criteria of 6 categories, the Bethesda System for Reporting Thyroid Cytopathology (BSRTC) has been introduced.2 Each diagnostic category has a suggested cancer risk, which ranges from 0% to 3% for the benign category to virtually 100% for the malignant category and is linked to evidence-based clinical management guidelines.3-5
Although most aspirates provide diagnostic cytology, approximately 30% are classified as nondiagnostic/unsatisfactory (ND/UNS) or indeterminate.4-8 Repeated FNA, core biopsy, and even intraoperative frozen section are often unsuccessful at clarifying the nature of these ambiguous nodules. BRAF, RET/PTC, RAS, or PAX8/PPAR are involved in thyroid tumorigenesis, and there have been efforts to combine molecular tests with cytologic interpretation.9-11BRAF is 1 of 3 isoforms of RAF, a family of proteins that encode serine/threonine kinases, and it plays a role in cell proliferation, differentiation, and apoptosis.12 The BRAFV600E mutation results in constitutive activation of the MEK-MAPK pathway, and has been focused on as a diagnostic and prognostic marker in papillary carcinomas (PCs).7, 9, 13 PC is the most common thyroid cancer and accounts for more than 90% of thyroid malignancies in Korea. Moreover, the prevalence of the BRAFV600E mutation in PCs is much higher in Korea (73-86%) than in Western countries.14-16 Therefore, the BRAFV600E mutation can be a useful molecular marker for preoperative diagnosis of PC. However, performing BRAFV600E mutation analysis in all FNA samples is unlikely to be cost-effective, and indications of a molecular test for thyroid nodule detected on ultrasonography (US) are not well established.
The aim of this study is to determine whether the second review of FNA cytology improves the diagnostic values for detecting PCs and to investigate usefulness of preoperative BRAFV600E mutation testing as an adjunctive diagnostic tool along with traditional FNA. We also analyzed cytologic diagnosis according to the mutation status in cases with surgical follow-up, and compared the efficacy of 2 detection methods for BRAFV600E mutation in FNA specimens.
MATERIALS AND METHODS
This study involved thyroid aspirates from 1060 patients who had undergone US-guided FNA at our institute between April 2008 and July 2009. Informed consent for BRAFV600E mutation testing was obtained from all patients, and this study was approved by the institutional review board.
Thyroid nodules were classified by US as malignant when they met at least 1 of the following criteria: 1) presence of calcification, 2) spiculated margin, 3) marked hypoechogenicity, and 4) taller-than-wide shape.17 Indeterminate nodules were defined as having neither benign nor malignant features. When a patient had multiple lesions, the sample for molecular testing was obtained from the most suspicious one. FNAs were performed by skilled radiologists using 22- or 23-gauge needle attached to a 2-mL syringe. Aspirates were expressed onto glass slides, and immediately fixed in 95% alcohol for Papanicolaou and hematoxylin and eosin (H&E) staining. H&E stain is routinely used in our institute and provides the familiarity of the color scheme to pathologists. The needle and syringe was then washed with normal saline for BRAFV600E mutation analysis. The FNAs, initially diagnosed by pathologists at our department, were reclassified into 6 categories based on the BSRTC.2 Cases of ambiguous cytologic report, which could not be categorized, were excluded from the present study. Indeterminate categories, including atypia of indeterminate significance (AUS), follicular neoplasm (FN), or suspicious for FN (SFN) and suspicious for malignancy (SFM), as well as the mutation-positive cases were re-reviewed by 2 experienced cytopathologists (Y. Oh and E. Cho) who were blinded to histologic diagnosis.
Detection of BRAFV600E Mutation
Allele-specific polymerase chain reaction (AS-PCR) and direct DNA sequencing were used to identify the BRAFV600E mutation in FNA specimens, and the results were considered positive when the mutation was detected by either method.
Genomic DNAs were extracted from FNA samples using QIAamp DNA mini kit (Qiagen, Chartsworth, Calif) according to the manufacturer's instructions. PCR amplification of exon 15 of BRAF was performed using the oligonucleotide primers: 5′-TTCATGAAGACCTC ACAGTAAAAA-3′ (forward) and 5′-CCACAAAATGG ATCCAGA- CA-3′ (reverse). PCR was carried out on a GeneAmp 9700 thermocycler (Applied Biosystems, Foster City, Calif) with an initial denaturation step at 94°C for 15 minutes, then 35 cycles at 94°C for 30 seconds, 62°C for 30 seconds, and 72°C for 60 seconds, followed by a final extension cycle at 72°C for 10 minutes. A 6-μL sample of the template was amplified in a reaction mixture containing 4-μL 5× BRAF ACE PM and 10-μL 2× Multiplex master mix (Seegene Inc., Seoul, Korea). The products were visualized on a 2% agarose gel with ethidium bromide. Samples were also analyzed by direct DNA sequencing with an ABI PRISM 3100 sequencer using BigDye terminator cycle sequencing ready reaction kit (Applied Biosystems). Sequences were compared with those of normal BRAF exon 15 in the GenBank database with Sequencher 4.8 (Gene Codes Corporation, Ann Arbor, Mich).
The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of FNA cytology alone and combined cytology BRAFV600E mutation testing for detecting PC were calculated. Detection of BRAFV600E mutation in FNA specimens that were histologically confirmed as malignant was considered as true-positive, and detection of the mutation in FNA specimens that were confirmed as benign was categorized as false-positive. Not detecting BRAFV600E mutation in benign lesions was considered as true-negative, and failure to identify the mutation in malignant lesions was categorized to false-negative. The indeterminate group, including AUS, FN/SFN, and SFM, was categorized as benign (ie, negative preoperative diagnosis of PC), when calculating the sensitivity and specificity of cytologic diagnosis.18, 19 Pearson chi-squared or Fisher exact test was used to examine the association between categorical variables, and the McNemar test was performed to assess the difference between paired proportions. Statistical analyses were carried out using SPSS 15.0 (SPSS, Somers, NY), and a P value of <.05 was considered statistically significant.
The cytologic diagnosis and mutational status of thyroid nodules were analyzed in 1060 patients (Table 1). Of these patients, 247, 143, 634, and 36 were diagnosed as malignant, indeterminate, benign, and ND/UNS by FNA cytology, respectively. The indeterminate category included 7 AUSs, 14 FN/SFNs, and 122 SFMs, and all SFM cases were suspicious for PC (SPC). BRAFV600E mutation was detected in 82.2%, 53.1%, 3.8%, and 5.6% of patients with malignant, indeterminate, benign, and ND/UNS cytology, respectively, which was more frequent in the malignant category compared with the indeterminate category (P <.001).
Table 1. BRAFV600E Mutational Status According to Cytological Diagnosis in 1060 Patients
|SFM||122||76 (62.3%)||47 (38.5%)|
|Malignant||247||200 (81.0%)||171 (69.2%)|
|Total||1060||302 (28.5%)||218 (20.6%)|
Total thyroidectomy or lobectomy was performed in 313 patients. Initial and reviewed cytologic diagnosis with the mutation status and final pathology are summarized in Table 2. Of the patients who underwent surgery, 200 had been diagnosed as malignant by FNA cytology. All 200 patients were histologically confirmed to have PCs, and BRAFV600E mutation was detected in 82.5% of them. Among 102 cases of indeterminate cytology, 95 turned out to be malignancies, including 90 PCs and 5 follicular carcinomas (FCs). The mutation was present in 57 PCs (63.3%), but not in FCs or in other benign lesions. Most surgically proven PCs were conventional type of PC. Only 5 cases were classified as follicular variant, 1 of which revealed dispersed microfollicular clusters with subtle nuclear change and was diagnosed as AUS on cytology.
Table 2. Initial and Reviewed Cytologic Diagnosis With BRAFV600E Mutation Status and Final Pathology in 313 Patients Progressing to Surgery
|ND/UNS||7 (2)||0||0||0||0||0||7 PCs|
|Benign||0||2 (2)||1 (1)||0||1 (1)||0||2 PCs1 FA1 NH|
|AUS||0||0||3 (0)||0||0||0||1 PC1 FC1 HTA|
|FN/SFN||0||0||0||9 (0)||0||0||4 FCs2 FAs3 NHs|
|SPC||0||0||1 (0)||0||18 (11)||71 (46)||89 PCs1 NH|
|PC||0||0||0||0||0||200 (165)||200 PCs|
There was a statistically significant difference in the diagnostic values between the initial cytology and the second review by 2 cytopathologists, and the sensitivity increased from 66.9% to 90.6% with better accuracy and NPV (P <.001). Based on a review of the cytology, the rate of malignancy after surgery was 100% (271/271) for malignant (PC), 100% (19/19) for SPC, 20.0% (1/5) for AUS and 50.0% (1/2) for benign. Moreover, combining FNA cytology with mutation analysis significantly improved the sensitivities, accuracies, and NPVs for detecting PCs (Table 3). In cases of indeterminate cytology, the concordance rate between initial and reviewed diagnosis was 29.4% (30/102). Seventy-one of 90 patients (78.9%) who had been diagnosed with SPC were upgraded to the malignant category, and 1 was reclassified as AUS. There was no diagnostic change in AUS and FN/SFN cases. The sensitivity, accuracy, and NPV of the second review was 78.9%, 81.4%, and 38.7%, respectively. The diagnostic values of reviewed cytology were better than those of BRAFV600E mutation testing alone (P = .029). Higher sensitivity, accuracy and NPV were obtained by combining FNA cytology with the mutation analysis (P = .001) (Table 4).
Table 3. Diagnostic Values of FNA Cytology and BRAFV600E Mutation Analysis for Detecting Papillary Carcinoma
|Initial cytology||66.9% (99.7%)||100.0% (92.9%)||68.4% (96.5%)||100.0% (99.7%)||12.4% (56.5%)|
|Initial cytology with BRAFa||87.3%||85.7%||87.2%||99.2%||24.0%|
|Reviewed cytology||90.6% (97.0%)||100.0% (100.0%)||91.1% (97.1%)||100.0% (100.0%)||33.0% (60.9%)|
|Reviewed cytology with BRAFb||95.7%||85.7%||95.2%||99.3%||48.0%|
Table 4. Diagnostic Values of Reviewed Cytology and BRAFV600E Mutation Analysis for Detecting Papillary Carcinoma in 102 Cases of Indeterminate Cytology
|Reviewed cytology with BRAFb||91.1%||100.0%||92.2%||100.0%||60.0%|
The clinicopathologic features of histologically confirmed PCs according to mutational status are shown in Table 5. BRAFV600E mutation was detected in 75.6% of patients with PC. Patients were older in the mutation-positive group (P = .044). The mutation was more common in patients with multicentric tumors (P = .032) and in those without background lymphocytic thyroiditis (P = .001). However, it was not associated with patients' gender, tumor size, extrathyroidal extension, and lymph node metastasis. The mutation was present in 82.0% of ultrasonographically malignant nodules and in 69.1% of benign or indeterminate nodules, the difference being statistically significant (P = .010).
Table 5. Clinicopathologic Characteristics of 299 Papillary Carcinomas According to BRAFV600E Mutational Status
|Gender|| || || |
| Female||179||61|| |
|Age (years)||47.9 ± 9.9||45.1 ± 10.7||0.044|
|Tumor size (cm)||0.9 ± 0.7||0.8 ± 0.5||0.608|
|Multicentricity|| || || |
| Absent||146||57|| |
|Extrathyroidal extension|| || || |
| Absent||107||38|| |
|Lymph node metastasis|| || || |
| Absent||132||52|| |
|Lymphocytic thyroiditis|| || || |
| Absent||179||44|| |
|Ultrasonographic feature|| || || |
| Benign or Indeterminate||103||46|| |
The BRAFV600E mutation was detected in 72.8% (228/313) of the patients who underwent surgery. The diagnostic values of 2 molecular methods for detecting PC are shown in Table 6. The sensitivity, accuracy, and NPV of the AS-PCR method were higher than those of direct DNA sequencing (P <.001). As there are no false-positives for the mutation confirmed by direct DNA sequencing, the specificity and PPV were 100%. AS-PCR failed to identify the mutation in 3 patients with malignant cytology, in whom the BRAFV600E mutation was detected by direct DNA sequencing. Conversely, the mutation was detected by the AS-PCR method, but not by direct DNA sequencing, in 50 patients (Table 7). The cytologic diagnoses were variable, with the majority of cases interpreted as malignant or SPC (88.0%), and 48 were surgically proven to be PCs. Two of 4 cases that appeared as malignant lesions on US, but had benign cytology, were followed-up with repeat FNA and turned out to be PCs. The remaining 2 patients with benign cytology presented with multiple nodules and were finally confirmed to have benign lesions.
Table 6. Comparison of Two Molecular Methods Detecting Papillary Carcinoma of the Thyroida
Table 7. Cytological and Histologic Diagnosis of 50 Allele-Specific PCR-Positive and Direct DNA Sequencing-Negative Cases
Thyroid FNA is a highly sensitive and specific method for diagnosis of PC. The pathologist's knowledge and experience are important in initially recognizing the cases concerning PCs, and then classifying them into AUS, SPC, or PC.20 Pathologists have varied widely in reporting FNA cytology, and some have tended to use indeterminate categories with which they are comfortable. It is important for pathologists to communicate with referring physicians in terms that are precise, unambiguous, and clinically useful.
Because SPC has a higher risk of malignancy, it should be conceptually separated from AUS. The diagnosis SPC is an indication for surgery with or without intraoperative frozen section, whereas the clinical approach to a nodule with an initial AUS interpretation is a repeat FNA at an appropriate interval.20 Follow-up studies have shown different clinical outcomes for distinct subcategories within the generic indeterminate category.5, 20, 21 As the target PPV of SPC is known to be 55% to 85%, this category was not regarded as malignant for calculating diagnostic values and was integrated into the indeterminate category with AUS and SFN/FN in the present study.22
It is important for centers with large patients populations to determine their own malignancy rates as related to reporting categories, so they can share them with their clinicians.20 Our institute is one of the largest referral centers in Korea, and more than 7000 thyroid FNAs are performed annually. In this series, the percentage of total FNA cases that fell into AUS, SPC, and malignant category accounted for 0.7%, 11.5%, and 23.3%, respectively. The AUS result is far less than the previous reports of 3% to 18% of thyroid FNAs.4, 5, 20 This may be because the present study only included the patients in whom the mutation analysis was performed with informed consent. For the category SPC, the proportion of cases and surgical yield of PC were higher than those of previous large-scale studies, which can be explained by that a cytologic diagnosis is dependent on pathologist's experience, comfort level, and threshold for malignancy, and some pathologists prefer ambiguous description due to legal problems.5, 21, 22
In this study, indeterminate categories and the mutation-positive cases were re-reviewed by experienced cytopathologists. It could be argued that the second review was not performed in all FNA cases. However, the main purpose of this study was to investigate potential usefulness of BRAFV600E testing & second review of FNA cytology in identifying papillary carcinomas in the indeterminate category. In addition, mutation positive cases with ND/UNS or benign cytology were re-reviewed to evaluate the false positivity of BRAFV600E mutation analysis. Although a distinction between malignant and SFM (or between SFM and AUS) is admittedly subjective, segregating 2 categories is to manage patients as appropriate as possible, allowing for alternative options before definite surgery.1, 2 Nevertheless, there is confusion among clinicians on how to manage a patient who does not have a clear-cut positive or negative result. Thus, we evaluated whether the second review of FNA cytology or BRAFV600E mutation analysis is beneficial in patients with indeterminate cytology, selecting for surgery those nodules with a greater probability of malignancy.
After the second review, the diagnosis of 71 patients was changed from SPC to PC. The cytopathologic features that were not initially interpreted as PC are (1) monolayered sheets with mildly altered nuclear polarity; (2) follicular cells exhibiting characteristic nuclear alterations, but rare intranuclear pseudoinclusions; and (3) PC with single cell/follicular pattern or cystic change. These cases reflect a combination of inexperience in recognizing certain criteria as possibly indicative of PC and the reviewer's resulting level of suspicion for PC. We also noticed that the assessment of specimen adequacy was variable among individual pathologists. The definition of an adequate specimen may be subjective and any specific cellularity is not applicable to all cases. However, some pathologists insisted on the SPC diagnosis when a specimen did not show at least moderate cellularity. If a FNA sample meets adequacy criteria and shows typical nuclear features, it is mandatory to diagnose malignancy.
The sensitivity, accuracy, and NPV of the second review by experienced cytopathologists were significantly higher than those of initial cytologic diagnosis, either with or without combination with the BRAFV600E mutation analysis. Moreover, the sensitivity, accuracy, and NPV of reviewed cytology were superior to those of the mutation testing in 102 cases of indeterminate cytology. This finding was unexpected, because BRAF mutation testing has been shown to have better or comparable sensitivity for PC in the diagnosis of indeterminate nodules when compared with FNA.16, 19, 23, 24 Based on these results, we presume that pathologist's role is crucial in the evaluation of thyroid nodules, regardless of implementation of the mutation testing.
Thyroid surgeons usually request intraoperative frozen section examination for definitive diagnosis and for determining the extent of thyroid resection in patients with SPC. In this study, the combination of FNA cytology and mutation analysis provided improvement of sensitivity, accuracy, and NPV in those indeterminate cases (Table 4). The significant increase in sensitivity came from the detection of BRAFV600E mutation in 11 of 18 patients with SPC. As illustrated in a recent study, patients with nodules with that have an equivocal cytologic diagnosis and are positive for BRAFV600E mutation would be strong candidates for total thyroidectomy. Lobectomy with or without intraoperative consultations would be indicated for patients with nodules that have a suspicious cytology but are negative for the mutation.25 Thus, preoperative BRAFV600E mutation testing can help with plan for surgery in the mutation-prevalent populations, and the patients may forgo intraoperative frozen section evaluation, which can reduce operation time.
The present study also showed evidence that BRAFV600E mutation testing in FNA samples with benign or ND/UNS cytology may be diagnostically meaningful and could decrease the false-negative rate (Table 2). However, routine mutation analysis in all aspirates is unlikely to be cost-effective. Further studies are needed to identify a set of clinical and US criteria that would determine which nodules with negative or atypical cytology should be rescreened using molecular tests.24 The prevalence of BRAFV600E mutation in tissue samples is not dramatically affected by a detection method applied, as occurs for RET/PTC rearrangements.26 However, a sensitive technique is required because aspirates can be contaminated with the surrounding nontumor cells and the level of mutation-carrying cells is low. Thus, sensitivity, sensibility, and reproducibility must be taken into account when choosing the most appropriated method and a source of analyzing materials (eg, scratched from glass slides, left behind in needle or processed for liquid-based cytology). In this study, the sensitivity and accuracy of AS-PCR were significantly better than those of direct DNA sequencing, and 48 of 50 AS-PCR-positive and DNA sequencing-negative cases were surgically proven to be PCs. Although equivocal bands in the gel and categorization by visual analysis can cause false-positive results, AS-PCR seems to be a useful complementary tool with the additional benefit of detecting more PCs.
The preoperative detection of the BRAFV600E mutation provides prognostic information and may be helpful in the future to guide targeted therapies.27 The mutation serves as an independent marker of aggressiveness in PCs, and is correlated with the frequency of lymph node metastasis.28, 29 Recently, Xing et al reported a greater than 4-fold higher risk for tumor persistence/recurrence among BRAFV600E-positive PC patients than among patients with mutation-negative tumors.30 In our study, the mutation was significantly associated with tumor multicentricity. Prospective studies of whether BRAF status would be a useful decision-making tool regarding the need for total thyroidectomy and the extent of lymphadenectomy at initial surgery are, therefore, warranted.24, 31
In conclusion, FNA is an accurate and reliable tool for the diagnosis of thyroid malignancy, which cannot be replaced by BRAFV600E mutation analysis. A uniform approach, especially in the interpretation of indeterminate categories, is important to triage patients with thyroid nodules for proper clinical management. Preoperative BRAFV600E mutation testing in conjunction with improved cytologic analysis can refine the diagnosis in mutation-prevalent areas while recognizing variable diagnostic thresholds.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.