Mario Vitale, Dipartimento di Endocrinologia ed Oncologia Molecolare e Clinica, Via S. Pansini, 5 Napoli, 80131 Italy. Tel.: 39 0817464983; Fax: 39 0817463668; E-mail: email@example.com
Background Fine-needle aspiration biopsy (FNAB) is the primary means to distinguish benign from malignant nodules and select patients for surgery. However, adjunctive diagnostic tests are needed because in 20–40% of cases the FNAB result is uncertain.
Objective We investigated whether a search for the oncogenes RET/PTC, TRK and BRAFV600E in thyroid aspirates could refine an uncertain diagnosis.
Patients and methods A total of 132 thyroid aspirates, including colloid nodules, inadequate samplings, indeterminate and suspicious for malignancy were analysed by reverse transcription polymerase chain reaction (RT-PCR) and mutant allele-specific amplification techniques for the presence of oncogenes.
Results No oncogenes were detected in 48 colloid nodules, 46 inadequate and 19 indeterminate FNABs, then confirmed to be benign at histology. No oncogenes were detected in one follicular thyroid cancer (FTC) with indeterminate cytology. Five out of six papillary thyroid cancers (83%) with FNAB suspicious for malignancy were correctly diagnosed by the presence of oncogenes. Among these, four (67%) contained the BRAF mutation and one (17%) contained RET/PTC-3. On final analysis, no false-positive results were reported in 131 samples and five out of seven carcinomas (71%) were correctly diagnosed. The finding of oncogenes in FNAB specimens suspicious for malignancy guided the extent of surgical resection, changing the surgery from diagnostic to therapeutic in five cases.
Conclusions Detection of RET/PTC, TRK and BRAFV600E in FNAB specimens is proposed as a diagnostic adjunctive tool in the evaluation of thyroid nodules with suspicious cytological findings.
A palpable thyroid nodule is very common in the general population and is usually a benign lesion. However, 4–5% of clinically apparent thyroid nodules harbour malignancy.1 An accurate preoperative diagnosis is needed to avoid delay, incomplete surgery, or unnecessary surgical intervention. Fine-needle aspiration biopsy (FNAB) is the gold standard for the diagnosis of thyroid nodules. The sensitivity and specificity of FNAB are variable and depend on the skill of the pathologist. In most studies, sensitivity and specificity are reported to be in ranges 70–98% and 55–100%, respectively.2 However, these percentages do not take into account the fact that 20–40% of FNABs yield uncertain results because of inadequate sampling or because cytology findings do not clearly indicate the nature of the lesion.1,3 In these cases, in the absence of clinical signs of malignancy, careful follow-up is recommended. In the presence of signs suggestive of malignancy, hemithyroidectomy is necessary. Intraoperative analysis of frozen sections is used as a means of guiding the extent of surgical resection. Frozen section analysis, however, not always impact on the intraoperative management of suspicious nodules.4 Therefore, adjunctive assays to improve preoperative assessment of thyroid nodules with inconclusive FNAB results are much needed.
Genetic alterations have been shown to play a pathogenetic role in thyroid tumorigenesis. The tyrosine kinase (TK) receptor/Ras/Raf/MEK/ERK signalling pathway is involved in cell growth and proliferation of most normal cells and, when altered, in tumorigenesis. Several genetic alterations have been identified among the components of this enzymatic cascade in thyroid carcinoma. Some of these genetic alterations are exclusive to thyroid cancer or are restricted to malignant tumours. The TK receptor RET is a component of a multiprotein complex, activated by the glial cell line-derived neurotrophic factor (GDNF) family molecules, that plays a crucial role in the development of the enteric nervous system and the kidney. Fusion of the sequence coding for the TK domain of RET to the 5′ sequence of genes that are constitutively expressed in thyroid follicular cells leads to the generation of a number of chimeric oncogenes known as RET/PTC.5,6 The gene NTRK-1 encodes a cell-surface TK receptor that binds the nerve growth factor, whose expression is restricted to the peripheral nerve ganglia. Similar to RET/PTC rearrangements, oncogenic activation of NTRK-1 (TRK) can occur through creation of chimeric fusion proteins.7,8 Both these oncogenes are exclusively present in papillary thyroid carcinoma (PTC). The thymine-to-adenine transversion at nucleotide position 1799 of BRAF, which results in a valine to glutamate substitution at residue 600 (BRAFV600E), occurs in a broad range of human cancers including PTC.9–12 To date, BRAF is the most frequent oncogene identified in PTC.
In view of their selective expression and high prevalence, the oncogenes RET/PTC, TRK and BRAF are ideal hallmarks of thyroid cancer with a papillary component. The prevalence of RET/PTC, TRK and BRAFV600E in PTC varies according to different studies from 8% to 85%, 5% to 12·6% and 28·8% to 45%, respectively,6,13 The sensitivity of a method based on the detection of these three oncogenes depends strictly upon their cumulative prevalences. Because overlap of these oncogenes has been excluded, one of them is present in the majority of PTC. Thus, because they offer excellent specificity and significant sensitivity as molecular markers of PTC, we investigated whether detection of RET/PTC, TRK and BRAF transversion mutation on FNAB specimens can be proposed as a diagnostic adjunctive tool in evaluation of thyroid nodules with indeterminate cytological findings.
Patients, materials and methods
Patients and FNAB
A total of 131 patients from the Azienda Ospedaliera Universitaria ‘Federico II’, Naples and the Ospedale Mauriziano, Turin were entered in the study after giving their consent and with approval from the institutional review boards. Patients harboured solitary or multiple thyroid nodules. In patients with multinodular goitre, the dominant nodule was examined. Cytology samples were obtained using a syringe with a 22-gauge needle passed three to four times. Patients whose FNAB results were uncertain, either because of insufficient material or indeterminate cytology diagnosis, were subjected to a second FNA procedure, performed twice consecutively in the same session. One sample was used for cytological examination, the other was dispersed into TRI Reagent buffer (Sigma) and stored at −20 °C until DNA and RNA extraction. Smears were classified according to Baloch and LiVolsi14 as suspicious for PTC, aspirates of moderate to low cellularity and equivocal malignant atypia such as abundant nuclear folds but no pseudo-inclusions; indeterminate, aspirates with high to moderate cellularity and the presence of microfollicular patterns with or without Hürthle cell change and scant colloid; inadequate, limited cellularity or poor preservation and fixation (or a combination of these factors). A specimen was qualified as satisfactory if there were six groups of epithelial cells with at least 10 cells per group. Only patients with persisting inadequate, indeterminate or suspicious cytology were selected for this study. Forty-eight benign nodules were also included as controls.
DNA extraction and detection of BRAF T1799A
DNA extraction was performed according to the manufacturer's recommendations. The final pellet was resuspended in 10 µl diethylpiyrocarbonate (DEPC) water. Searching for the BRAF mutation was performed by mutant allele-specific polymerase chain reaction (PCR) amplification (MASA), as described previously.15 Two different forward primers with substitution of a single base at the end of the primer were used to amplify the wild-type allele or the BRAF T1799A transversion mutation with an unique reverse primer (Table 1). PCR reactions were performed with 50–100 ng genomic DNA, 0·5 µM of each primer and 2·5 U Euro-Taq DNA polymerase (EuroClone, Celbio, Italy). All primers were obtained from Primm (Milan, Italy). All PCR reactions were performed separately in a PTC 100 Peltier Thermal Cycler (MJ Research Bio-Rad, Milan, Italy), including an initial denaturation of 2 min at 94 °C and subsequent denaturation for 30 s at 94 °C, annealing for 30 s at 58 °C and extension for 30 s at 72 °C. All samples were re-examined for the BRAF mutation at least three times. cDNA from NPA cells (harbouring BRAF T1799A transversion mutation) and from WRO cells were used as positive and negative controls, respectively.
Table 1. Primers used for the PCR amplifications
RET-TK, RET tyrosine kinase domain; RET-EC, RET extracellular domain.
RNA extraction and reverse transcription polymerase chain reaction (RT-PCR) amplification control
Total RNA was extracted using TRI Reagent following the suggested protocol. The pellet of RNA was resuspended in 10 µl DEPC water and reverse transcribed with SuperScript III (Invitrogen, Milan, Italy) in a 20 µl reaction volume with random primers. The presence of thyroid follicular epithelial cells, the integrity of the RNA and the efficiency of the RT reaction in each sample was confirmed by PCR for thyroglobulin mRNA (Table 1). The primers were chosen in such a way that the amplified fragments would include one splicing junction to avoid confusion with PCR products amplified from contaminating genomic DNA. The PCR conditions were the same as those used for BRAF, with standard conditions; the annealing temperature was 60 °C.
Detection of TRK rearrangements
To establish the presence or absence of the gene rearrangements we have used a standard method of RT-PCR. This method consists in primary and nested amplifications followed by agarose gel electrophoresis. For the first PCR amplification 2% of cDNA was used with the set of primers showed in Table 1, with forward primers in the 5′ region of TPM-3 and reverse primers in the TK domain of NTRK-1. For the second round of amplification, 1 : 10 of the first-round PCR product was used. The PCRs were performed as described previously; the thermal cycler (MJ Research PTC-100) conditions used for both amplifications were 94 °C for 2 min followed by 35 cycles at 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 30 s, and a final step at 72 °C for 5 min. Ten microlitres of the PCR products were separated on 1·2% agarose gel and visualized by ethidium bromide staining. cDNA from a PTC containing TRK was used as a positive control.
Detection of RET/PTC rearrangements
Detection of RET/PTC rearrangements was performed with RT-PCR using the following approach: all the samples were studied for the expression of the TK and extracellular (EC) domains of RET. Samples positive for the RET EC domain were screened for calcitonin expression by RT-PCR to exclude the abundant presence of C cells, which normally express RET. Only samples showing TK expression not associated with EC were considered positive for a rearrangement and analysed with specific primers for primary and nested amplifications for the most common chromosomal rearrangements found in sporadic and radiation-induced PTC. In brief, we used forward primers in 5′ of H4, Riα and Ele1 for RET/PTC-1, RET/PTC-2 and RET/PTC-3, respectively, and the same reverse primers in the TK domain of RET (Table 1). PCR was conducted in a One Advanced thermocycler (EuroClone, Celbio, Milan, Italy) using the same conditions for each amplified fragment: 2 µl of cDNA were added to a mixture containing 2·5 mM MgCl2, 0·5 µM of each primer, 200 µM dNTP, and 2·5 U Taq (EuroClone) in a final volume of 25 µl. Thirty-five cycles of denaturation (94 °C for 30 s), annealing and extension (72 °C for 30 s) were conducted. The annealing temperatures used for analysis of RET/PTC were the following: 58 °C for RET TK; 68 °C for RET/PTC-1 and 60 °C for the others (RET-EC; RET/PTC-1 nested; RET/PTC-2; RET/PTC-2 nested; RET/PTC-3; RET/PTC-3 nested). Ten microlitres of PCR product were analysed by electrophoresis in a 1·2% agarose gel. cDNA from TPC-1 cells (harbouring RET/PTC-1) and from PTC samples containing RET rearrangements were used as positive controls.
We analysed for the presence of RET/PTC and TRK rearrangements and for BRAF transversion mutation in a total of 132 thyroid aspirates, including 48 colloid nodules, which served as controls, 46 inadequate samplings, 21 indeterminate samplings and 16 suspicious for PTC. The results of this study are summarized in Figs 1 and 2. In one sample, thyroglobulin mRNA could not be detected and the search for the oncogenes was not performed. In the 48 colloid nodules and the 46 inadequate FNABs, no oncogenes were detected. Hemithyroidectomy was performed in 18 of the patients with colloid nodules and in 26 of the 46 patients with inadequate FNABs. None was suspect for cancer on clinical grounds and all proved to be benign nodules on histopathology after surgery. Of 21 indeterminate FNAB specimens, eight displayed increased cellularity, microfollicular patterns and Hürthle cell change. No oncogenes were found in this group and none of the nodules proved to be malignant at histopathology. Thirteen indeterminate FNABs yielded similar cytological findings but did not display Hürthle cell change. In this group no oncogenes were found in 11 samples; however, 10 nodules proved to be benign and one was found to be a follicular thyroid carcinoma (FTC) at histopathology after surgical resection. The cDNA for both the extracellular and the TK domains of RET was detected in two FNABs, suggesting the transcription of the entire receptor. In one patient, a medullary thyroid carcinoma (MTC) was suspected because of the presence of a high serum calcitonin concentration. Surgery was performed and the diagnosis of MTC was confirmed at histopathology. The other patient lacked any clinical suspicion of MTC and surgical resection was delayed until mutational analysis of RET was completed. Of the 16 FNABs suspicious for malignancy, six proved to be thyroid cancer (37%). Among these, four (67%) contained the thymine to adenine BRAF mutation, one (17%) contained RET/PTC-3 and TRK was not detected. On final analysis, no false-positive results were reported in 130 samples, five out of seven carcinomas (71%) were correctly diagnosed (five out of six PTC, 83%), while two cancers were not detected by this method. The analysis of surgical specimens confirmed the FNAB findings and no oncogenes were detected in surgical samples of negative FNAB.
We analysed 132 thyroid aspirates for RET/PTC and TRK rearrangements and for the BRAF mutation at nucleotide 1799. In only one sample the thyroglobulin transcript could not be detected, demonstrating that the quality of mRNA extracted from the samples was sufficient to allow RT-PCR analysis. In 46 inadequate FNABs no oncogenes were detected and 26 proved to be benign nodules after surgery had been performed for cosmetic or other reasons. This result is consistent with the expected prevalence of cancer in thyroid nodules in the absence of clinical suspicion for malignant neoplasm.16,17 Considering cost-effectiveness, the search for these three oncogenes is not advisable in inadequate FNAB specimens, in the absence of clinical suspicion. While no false-positive results were reported in 130 samples, none of the three oncogenes was found in one PTC and one FTC proved at histopathology. While RET/PTC, TRK and the BRAF mutation are not expressed in FTC and thus this tumour invariably escapes detection, the search for oncogenes was also negative in the tissue obtained from the resected PTC. The lack of genetic markers in a consistent percentage of thyroid carcinomas represents to date the main limitation of the technique. Analysis of PTC is necessary to identify other unknown genetic abnormalities that are responsible for or are associated with this tumour. In this small series, five of the six PTC (83%) and five of seven carcinomas (71%) were correctly diagnosed in indeterminate cytology. In two aspirates the presence of mRNA for both the EC and the TK domains of RET indicated the presence of a high number of thyroid medullary cells. This does not provide any evidence of whether the increased copy number of RET mRNA in the samples was due to C cell hyperplasia or MTC. In one of two cases that occurred in our study, the increase in serum calcitonin concentration indicated an MTC that was then confirmed at histopathology. In the other patient, in the absence of clinical suspicion of MTC, surgery has been delayed until the analysis of RET mutations is completed.
FNAB provides an accurate and cost-effective method of identifying patients with thyroid cancer. However, when the FNAB result is uncertain, the selection of patients requiring surgery is based solely on clinical grounds. In this situation thyroid surgery will be recommended to the patients who are assigned to the high category of suspicion for malignant neoplasm, while the other patients will be subjected to long-term follow-up. The type of surgery, when recommended, is also a controversial issue. Total thyroidectomy is indicated in the presence of cancer but is potentially associated with a higher risk for operative complications, such as parathyroidectomy and recurrent laryngeal nerve injury. Hemithyroidectomy plus resection of the isthmus is an alternative choice to total thyroidectomy. However, if cancer is demonstrated at histology, thyroidectomy must be completed with a second operation. In addition, intraoperative analysis of frozen sections is not always useful because of its low sensitivity to detect FTC and the follicular variant of PTC. In our study, thyroid cancer was correctly diagnosed in five indeterminate FNABs. This impacted on the decision regarding the extent of the surgery performed. The selection of thyroid nodules for surgical removal would certainly be improved if diagnostic strategies incorporated the detection of genetic markers of malignancy. Searching for oncogenes in biopsies suspect for malignancy may enhance the accuracy of FNAB but has some important limitations that need to be considered by the clinician. Conventional cytological diagnoses are divided into four categories: benign, insufficient, indeterminate, or malignant. The results of the search for oncogenes in FNAB specimens only fall into two categories: malignant (papillary or anaplastic) and unknown. Thus, while the presence of one oncogene is indicative of cancer, its absence has no impact on the diagnosis and on the final clinical decision. This will be the situation in the majority of cases as most of the nodules with indeterminate cytology are benign. The sensitivity of the method is limited by the global prevalence of RET/PTC, TRK and BRAFV600E in thyroid cancer with a papillary component and by their absence in follicular carcinomas, which in most series represent about 5% of thyroid carcinomas, and in metastatic cancer.
The identification of a genetic alteration specific for FTC would greatly improve the sensitivity of the method, mainly because this cancer very frequently yields indeterminate cytology that cannot be distinguished from benign adenoma.
Another important limitation is the heterogeneity of the tumour and the size of the sample analysed. In some studies, RET/PTC rearrangements were demonstrated only in a small proportion of cells within the tumour mass of PTC, supporting the idea of tumour heterogeneity.18–20 As for RET/PTC, theoretically, TRK and BRAF mutation could also be present in only a proportion of tumour cells. In a previous study, RET/PTC rearrangements present in the surgically resected tumours could not be detected in FNAB samples, suggesting a failure to biopsy the tumour, the inclusion of nontumour cells or the presence of tumour heterogeneity.21 A too small proportion of cells harbouring the oncogene in the sample would reduce the sensitivity of the method. The passage of the needle three to four times into the nodule mass ensures the collection of cells from different sites within the nodule. The analysis of a small sample, as in the remaining material after the smear preparation, risks being representative of only a limited nodule area and would reduce the sensitivity of the method. For this reason we decided to use the entire material obtained by one biopsy and no discordance resulted from the genetic analysis of FNAB and of larger postsurgical tissue samples.
The search for oncogenes in indeterminate FNAB samples can be considered as a source of additional information helpful in the selection for surgery; thus we propose this method as a useful tool to improve the accuracy of preoperative diagnosis in identifying PTC from biopsies with indeterminate cytological findings.
We thank G.N. Burrow for critical reading of the manuscript. This work was supported in part by Ministero dell’Istruzione, dell’Università e della Ricerca (to M.V., G.F. and G.R.) and Compagnia di San Paolo di Torino, Project Oncology (to P.P.L.).