MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
- FUNDING SOURCES
- CONFLICT OF INTEREST DISCLOSURE
This study was approved by the Cleveland Clinic Institutional Review Board. Forty-three thyroid FNA samples were retrospectively collected after rendering clinical cytologic diagnoses between 2011 and 2012. These cases were purposefully selected to have varying degrees of morphologic suspicion for PTC. After each pass, a direct smear was made and the FNA needle was rinsed in CytoLyt (Hologic, Marlborough, Mass). One ThinPrep (Hologic) slide was prepared from each case for assessment of both morphology and cellularity. Quantitative evaluation of cellularity on the ThinPrep slide was performed based on the number of groups of 10 or more follicular cells as follows: 1+ (1-5 groups), 2+ (6-10 groups), 3+ (11-20 groups), 4+ (>20 groups). The cellularity was evaluated by 2 cytopathologists. Examples of samples from each category are illustrated in Figure 1.
Figure 1. Thyroid fine-needle aspirate samples exhibited a wide range of cellularity. Quantitative evaluation of cellularity on the ThinPrep slide was performed based on the number of groups of 10 or more follicular cells as follows: (A) 1+, 1-5 groups; (B) 2+, 6-10 groups; (C) 3+, 11-20 groups; (D) 4+, > 20 groups.
Download figure to PowerPoint
The residual liquid-based materials from each case were stored at 4°C and submitted for molecular testing within 2 weeks of rendering a cytologic diagnosis. Genomic DNA was extracted using the Gentra Puregene kit (Qiagen, Chatsworth, Calif) and the QIAsymphony DNA Mini Kit with the QIAsymphony SI instrument (Qiagen) following the manufacturer's stated protocol. DNA concentration was measured using the NanoDrop ND1000 spectrophotometer (ThermoFisher Scientific, Kalamazoo, Mich), and the purity of genomic DNA was assessed by calculating both the protein/nucleic acid ratio and the DNA/RNA ratio. Cellularity and DNA yield were then correlated.
To compare different methods for BRAF mutational assay, 37 samples were analyzed using an allele-specific polymerase chain reaction (AS-PCR) assay (Rotor-Gene; Qiagen). In addition, Sanger sequencing was performed on all 43 samples (ABI 3730 Genetic Analyzer; Applied Biosystems, Foster City, Calif) following the manufacturer's recommendations. Sanger sequencing targeted a 224–base pair region of the BRAF gene that includes the hotspot for mutations around codon 600 in exon 15.
- Top of page
- MATERIALS AND METHODS
- FUNDING SOURCES
- CONFLICT OF INTEREST DISCLOSURE
A total of 43 thyroid FNA samples from 40 patients, including 17 males and 26 females (mean age of 51.5 years), were collected for this study. Patients with multiple samples had samples collected from separate locations (ie, left versus right or isthmus thyroid); samples were interpreted separately and an independent diagnosis rendered for each. Cytologic diagnoses included 6 suspicious for follicular neoplasm (SFN), 13 suspicious for PTC (SPC), and 24 positive for PTC. At the time of the study, 38 resections (on 35 patients) had been performed. Among those, PTC was confirmed histopathologically in 34 of 35 patients. One patient had chronic thyroiditis with focal Hürthle cell hyperplasia. Four thyroidectomy resections had non-PTC neoplasms, most commonly Hürthle cell adenomas. The tumor size of PTC ranged from 0.2 to 4.5 cm. The demographic data and pathologic diagnoses are summarized in Table 1.
Table 1. Demographics and Histopathologic Diagnoses of Thyroid Lesions
|Age range||21-88 y|
|Histopathologic diagnosis|| |
|No surgical follow-up||5|
|Tumor size (PTC)|| |
On surgical follow-up, 34 patients had confirmed PTC in resection specimens. Histopathologically, 19 patients had predominantly classic type PTC, 8 had follicular variant, and 7 had tall cell variant. Cytologically all the classic and tall cell variants of PTC were diagnosed either definitively as positive or suspicious for PTC. Six of 8 follicular variants of PTC were diagnosed as positive or suspicious for PTC and the remaining 2 cases were diagnosed as SFN. Table 2 highlights and compares the cytologic and histologic diagnoses of our study population.
Table 2. Correlation Between Cytologic and Histopathologic Diagnosis
|Cytologic Diagnosis||No. of Cases||Histopathologic Diagnosis|
|PTC, Classic||PTC, FV||PTC, TC||Non-PTC Neoplasm||No Surgery|
In our study, 30.2% of cases had 1+ cellularity and approximately 49% of cases had 1-2+ cellularity. It appears that the variation of cellularity correlates to some extent with cytologic diagnosis (Table 3 and Fig. 1). For example, low cellularity (1-2+) was seen in 63.2% of cases with cytologic diagnosis of SFN or SPC, compared with only 37.5% in those cases with a diagnosis of positive for PTC. In contrast, high cellularity (3-4+) was seen in 36.8% of cases with a diagnosis of SFN or SPC, compared to 72.5% in those diagnosed as positive for PTC. The number of atypical diagnoses gradually decreased along with the increase in cellularity. Table 3 shows that atypical cytologic diagnoses decreased from 62% at 1+ cellularity to 25% in 4+ cellularity.
Table 3. Correlation of Cellularity and Cytologic Diagnosis
|Cellularity||SFN||SPC||PTC||No. of Cases||Atypical Diagnoses|
We evaluated the minimum amount of genomic DNA required to successfully run BRAF mutational analysis. A minimum of 200 ng of DNA was required to successfully perform BRAF mutational analysis using Sanger sequencing methodology and a minimum of 150 ng DNA was needed to conduct BRAF assay using AS-PCR. The ThinPrep cellularity correlates proportionally with genomic DNA yield, which is summarized in Table 4. The range of genomic DNA yields in our thyroid FNA samples varied greatly from 146 ng to more than 50,000 ng, with a minimum median of 699 ng and minimum mean of 1299 ng. A trend of gradually increasing DNA yield is observed from low to high cellularity. The average DNA yield increased nearly 4-fold from 1+ to 4+ cellularity. There are 3 cases with DNA yields less than 200 ng, which is below the threshold for successful testing for BRAF mutation in our laboratory. The cellularity of these 3 cases was in the low cellularity range, including 2 cases with 1+ cellularity (168 and 170 ng, respectively) and one case with 2+ cellularity (146 ng of DNA). Overall, 93% (40 of 43) of thyroid FNA samples yielded >200 ng of genomic DNA. When using 2+ cellularity (>6 groups of more than 10 cells) in ThinPrep slides as a cutoff for the adequacy of DNA yield, approximately 97% (29 of 30) of cases yield more than 200 ng of genomic DNA for BRAF mutational analysis.
Table 4. Correlation of Cellularity and Genomic DNA Yield
|Cellularity||No. of Cases||DNA Range (ng)||Median (ng)||Mean (ng)||Standard Deviation|
DNA yield did not appear to be related to type of PTC. We found that both follicular variant (FV) and tall cell variant (TC) generated sufficient DNA for molecular testing. Eight of our samples showed FV PTC on surgical resection. These cases generated a great deal of DNA (560-2992 ng; mean, 1476.5 ng). All the cases were successfully tested for BRAF mutation. The majority of cases tested were wild-type BRAF (75%). Three cases of PTC had oncocytic features and generated a mean DNA amount of 4086 ng with range of 560-10982 ng. BRAF mutational assay showed that all 3 samples contained wild-type BRAF. Seven of our samples showed TC PTC on surgical resection. These cases generated a mean DNA amount of 1298.5 ng with a range of 146-3232 ng. Mutational analysis showed that all 7 cases had the BRAF V600E mutation. Classic type PTC showed the widest variation in DNA yield (471-53,907.4 ng; mean, 6635.1 ng). Seven cases with cystic changes had low cellularity, including 6 cases with 1+ cellularity and 1 case with 2+ cellularity. Although DNA yield in these samples ranged from 543.6-10,982 ng, with a mean of 3783 ng, the presence of cystic changes still adversely affected molecular analysis. Of the 4 cases that yielded false negative results for BRAF mutation by Sanger sequencing (as compared to PCR), 3 samples (75%) had cystic changes.
Using Sanger sequencing (ABI 3730 Genetic Analyzer; Applied Biosystems, Foster City, Calif), we conducted at least 1 successful BRAF mutational analysis on all 43 cases. Twenty-four cases (55.8%) harbored BRAF V600E point mutations, whereas 19 cases had wild-type BRAF (Table 5). Successful molecular testing relies not only on DNA amount but also on the ratio or percentage between target cells and background noise such as inflammatory cells and benign follicular cells. To compare the sensitivity of detection of BRAF mutation in thyroid FNA samples, we compared 2 technical platforms, traditional Sanger sequencing using the ABI Genetic Analyzer and a real-time AS-PCR–based assay using Rotor-Gene platform. However, 6 (14%) cases could not undergo BRAF mutational testing for both Sanger sequencing and AS-PCR due to limited amount of DNA yields. All 6 cases were in the low cellularity range including 5 cases in 1+ and one case in 2+ cellularity. Therefore, only those 37 cases with sufficient DNA for duplicate analyses were subjected to BRAF mutational testing using both Sanger sequencing and AS-PCR (Table 6). Our study showed that Sanger sequencing detected 24 cases with BRAF mutations, whereas 13 cases were found to be the wild type. In contrast, AS-PCR assay detected BRAF V600E mutations in 26 cases, with 11 cases having no mutation detected. There were 4 cases with discordance between Sanger sequencing and allele-specific assay (Table 6). Among those, 3 cases were wild type by Sanger sequencing, but were found to have BRAF V600E mutation detected by AS-PCR. The fourth case is discussed below. These 3 discordant cases had low cellularity of 1+ or 2+ with approximately 10% to 20% target malignant cells present. There is a 100% concordance between Sanger sequencing and AS-PCR in higher cellularity groups (3-4+) with greater than 25% of malignant cells usually present. Our study further demonstrates that although Sanger sequencing technology can be successfully conducted with a minimum DNA amount of 200 ng, it demands a higher percentage of target cancer cells (>25%) to detect the BRAF V600E mutation due to its lower analytical sensitivity. In contrast, AS-PCR assays can detect > 10% target cells at even lower cellularity in thyroid FNA samples.
Table 5. Cellularity, DNA Yield, and BRAF Assay by Sanger Sequencing
|Cellularity||DNA Range (ng)||BRAF Status (ABI Sequencing)|
Table 6. BRAF Mutational Testing by Sanger Sequencing and Allele-Specific PCR Assay
|Cellularity||ABI Sequencing||AS-PCR||Sanger-PCR Concordance|
|Mutated||Wild Type||Mutated||Wild Type|
A rare BRAF mutation, L597Q, was detected by Sanger sequencing in a case with 4+ cellularity. This case is the fourth discordant case between Sanger sequencing and allele-specific assay. The cytologic diagnosis in this sample was “atypical cannot exclude PTC” in a background of lymphocytes. Follow-up surgical resection showed a hyperplastic nodule with Hürthle cell change and background of chronic lymphocytic thyroiditis. The patient's TSH was not elevated. However, this rare BRAF point mutation was not detected with repeated AS-PCR assay. The AS-PCR assay used in this study was designed to be specific for the c.1799T > A mutation, which results in the BRAF V600E substitution, and is not capable of detecting the L597Q (c.1790T > A) mutation. Electropherograms of the BRAF V600E mutation (c.1799T > A) and this rare BRAF L597Q mutation (c.1790T > A) are included as Fig. 2A and 2B, respectively.
Figure 2. (A) Electropherogram showing BRAF mutation at V600E (c.1799T>A). (B) Electropherogram showing BRAF mutation at L597Q (c.1790T>A)
Download figure to PowerPoint
- Top of page
- MATERIALS AND METHODS
- FUNDING SOURCES
- CONFLICT OF INTEREST DISCLOSURE
Papillary thyroid carcinoma is the most prevalent of all thyroid carcinomas, comprising approximately 80% of thyroid malignancies. Thyroid FNA is currently accepted as the best initial diagnostic tool to establish diagnosis and guide further management.[9, 10] The standard treatment protocol includes thyroidectomy followed by radioiodine (131I) to remove residual tumor and treat metastases. Most patients have a good prognosis, with indolent behavior and high long-term survival rates.2-4 However, a small subset of patients experiences a more rapid clinical course.
BRAF mutations are the most common known genetic events in PTC, being present in approximately 20% to 50% of tumors in the literature.[2, 5, 12] More than 95% of reported mutations are c.1799T>A point mutations, which result in the substitution of valine with glutamic acid at residue 600 (p.V600E), leading to constitutive activation of BRAF kinase and chronic stimulation of the mitogen-activated protein kinase pathway, and eventually results in carcinogenesis in thyroid epithelial cells. BRAF mutations are fairly specific for PTC, not being found in benign thyroid lesions or other types of thyroid cancers. In addition, BRAF mutations have consistently been associated with more aggressive clinicopathologic behavior, including extrathyroidal extension, advanced stage of disease at presentation, and lymph node or distant metastasis.[2, 5, 8, 12] Therefore, identification of BRAF mutations has not only diagnostic utility, but potential therapeutic and prognostic implications as well, serving as a prognostic biomarker for clinical management. The emergence of thyroid FNA as the preferred diagnostic modality for PTC highlights the importance of optimizing detection of BRAF mutations in cytologic specimens.
BRAF mutational analysis has traditionally been performed on formalin-fixed paraffin-embedded surgical tissue. Recently, several studies have been published on the feasibility of successfully testing for BRAF mutations using cytologic materials.[7, 9],14-20 It has been shown that detection of BRAF mutations can be helpful in facilitating accurate diagnosis in thyroid FNA cytology.[5, 13] A successful cytologic diagnosis relies on adequate sample with good cellularity. National consensus guidelines on specimen adequacy for cytomorphologic evaluation have been published recently to standardize the practice of thyroid FNA and cytologic diagnosis. However, discussion of sample adequacy for molecular testing is noticeably incomplete in many published studies. There is a general lack of systematic approach in the literature to validate specimen adequacy for successfully conducting molecular testing for BRAF mutations in thyroid FNA samples. It is imperative to establish cellularity adequacy criteria and minimal DNA amount needed in any molecular cytology laboratory in order to successfully perform molecular testing and to avoid false-negative report. In this study, we examined the minimum cellularity necessary to yield enough DNA for molecular testing, assessed the minimum amount of DNA required to successfully perform BRAF mutational analysis, and compared the sensitivities of different molecular methods in detection of BRAF mutations in cytologic samples. We have studied 43 cases of thyroid FNA samples with cytologic diagnoses of either positive for malignancy, atypical suspicious for malignancy, or atypical suspicious for follicular neoplasm. All the cases had a ThinPrep slide made before submission of the remainder of the vial for genomic DNA extraction for molecular testing. We used the ThinPrep slides to evaluate cellularity of the specimen and divided cellularity from 1+ to 4+ based on number groups of at least 10 cells on the slides. The cellularity was then correlated with cytologic diagnosis, genomic DNA yields, and conduction of BRAF mutational assay.
Low cellularity is the nature of thyroid FNA. We found that nearly half of the cases in our study population (21 of 43, or 49%) had either 1+ or 2+ cellularity. In fact, the lowest cellularity (1+) was seen in 30% of cases. It is well known that limited cellularity hinders diagnostic certainty in cytology; our finding shows a great correlation between cellularity and diagnostic certainty. For example, a definitive positive diagnosis of PTC was seen in 37.5% of cases with lower cellularity (1+ and 2+) compared to 72.5% of cases with higher cellularity (3+ and 4+). In contrast, atypical cytologic diagnoses (SFN and SPC) were seen in nearly two-thirds of cases with 1+ cellularity compared to only 25% of cases with 4+ cellularity (Table 3).
Similarly, cellularity also correlates well with genomic DNA yields. We found the DNA yields varied from sample to sample with a trend that was proportionate to cellularity. DNA yield tended to be lower when cellularity was low (1+ or 2+), with approximate median DNA yield of 700 ng. Three cases with DNA yields lower than 200 ng fell into 1+ and 2+ cellularity categories. In contrast, every sample with cellularity greater than 3+ yielded sufficient DNA for molecular testing. The median DNA yield in high-cellularity samples (3-4+) was almost 2.5- to 4.1-fold higher than in lower cellularity samples. On average, a typical thyroid FNA sample generates more than 1200 ng genomic DNA, enough to perform clinical testing in triplicate. When using 6 groups of 10 or more follicular cells as a cutoff, the vast majority (97%) of routinely collected clinical thyroid FNA samples contain sufficient genomic DNA for molecular testing. Interestingly, the same adequacy criteria are also used by the Bethesda System to evaluate sample adequacy for cytomorphologic interpretation. Therefore, based on this study, we recommend applying the same adequacy criteria both for cytomorphologic interpretation and for molecular testing for BRAF mutation. The only notable difference is that we emphasize using the ThinPrep slide, as opposed to conventional smears, in order to more accurately reflect the cellularity remaining in leftover vials for molecular adequacy.
DNA yield is not related to the subtype of PTC. Some types, especially the follicular variant, can be difficult to diagnose cytologically due to only focal features of PTC. These cases become even more challenging when limited by low cellularity. However, molecular studies are not hampered by morphologic uncertainty. Fifteen cases had a final histopathologic diagnosis of either follicular variant or tall cell variant of PTC on surgical resection. These cases generated a great deal of DNA, and all the cases were successfully tested for BRAF mutations. The majority of FV PTC cases was wild-type BRAF (75%) which is consistent with the literature. All of the TC PTC cases had the BRAF V600E mutation (100%), which is also consistent with what has been recently reported. The classic type of PTC can show a wide range of DNA yield based on cellularity. Seven cases in our series had cystic changes, most of which also had low cellularity (1+). Although DNA yield in these samples was adequate (543.6-10982 ng; mean, 3783 ng), 3 of 4 false negatives for BRAF mutation by Sanger sequencing (V600E by PCR, wild type by Sanger) were found in these cases with cystic changes, mainly due to diluted target cell population by background histiocytes and inflammatory cells.
It is rather challenging to conduct molecular testing in liquid cytologic specimens than in tissue-based samples, because there is no practical way to enrich or separate target tumor cells from non-neoplastic cells. Successful molecular testing on cytologic materials relies not only on the absolute DNA amount but also on the ratio of target malignant cells to background noise cells such as benign follicular cells and inflammatory cells. The sensitivity of detection of minimal target cells with genetic alterations is very much methodology-dependent. A sensitive testing platform should be able to detect a small number of malignant cells in a mixture of a large number of non-neoplastic cells. Therefore, choosing a proper technical platform is an important factor for successful conducting of molecular testing and providing accurate molecular information for personalized medicine. BRAF mutations can be detected via many different methods; direct sequencing is currently the gold standard with relatively high specificity (96%), but it suffers from mediocre sensitivity (83%) and is technically very time-consuming. Other available modalities include polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP), mutant allele-specific PCR, probe-specific real-time PCR, and pyrosequencing, which provide better sensitivity than direct sequencing.
To determine the sensitivity in detecting BRAF mutations in cytologic specimens, we compared 2 technical platforms for BRAF mutational analysis: Sanger sequencing using ABI 3730 and allele-specific PCR. Platforms were compared using 37 cases, with enough genomic DNA present for both assays. It should be noted that both methods have been successfully used in our molecular laboratory for detection of BRAF mutations in paraffin-embedded formalin-fixed tissue sections with ample recuts. Although Sanger sequencing was run successfully with a minimum of 200 ng DNA in all 37 cytologic samples, there were 4 cases with discordance between Sanger sequencing and AS-PCR. Among them, 3 cases had BRAF V600E mutation detected by AS-PCR but missed by Sanger sequencing. All 3 cases had lower cellularity and also had a lower percentage (5%-15%) of target malignant cells. One group has reported detection of as low as 1% of target DNA using optimized AS-PCR. In cases with higher cellularity and higher percentage of target malignant cells (>20%), there is a perfect concordance between the 2 methods. Our study further demonstrates that when applying a molecular detection system in cytologic specimens with limited cellularity, the method with higher sensitivity prevails with both more accurate detection rates and fewer false negative results. In our laboratory, although Sanger sequencing can be used confidently in samples with higher cellularity, PCR has become the method of choice to identify BRAF mutations in low-cellularity specimens, which comprise nearly half of all routine thyroid FNA samples.
In the fourth discordant case, a 1790T > A mutation resulting in L597Q was detected by Sanger sequencing but was not detected by AS-PCR. The latter is not specifically designed in detection of BRAF mutations other than V600E. This is a rare BRAF point mutation that has been previously reported and described in malignant melanoma and childhood acute lymphoblastic leukemia.[25, 26] However, to our knowledge, it has never been reported in thyroid tissue. Follow-up on this case showed Hashimoto's thyroiditis with no evidence of malignancy. The clinical significance of this unusual BRAF mutation is unknown, and this case will be closely followed to see if this peculiar mutation is associated with any potential for thyroid malignancy.
In summary, using the ThinPrep slide as a surrogate to estimate cellularity is a practical approach to evaluate adequacy for molecular testing. It is straightforward to count groups of follicular cells, and it would be simple for both cytotechnologists and cytopathologists to use this method in routine clinical practice. Coincidently, the minimum cellularity we assessed based on this study mirrors the Bethesda System adequacy criteria for cytomorphologic interpretation. Besides minimum DNA yield, the percentage of malignant cells (especially in low-cellularity samples) is another important factor affecting the success of molecular testing. To overcome both low cellularity and a low percentage of malignant cells in liquid cytology samples, careful selection of a sensitive detection system or technical platform is another key element to ensure the successful application of molecular tools in liquid-based cytologic specimens.