Thyroid nodules have a prevalence of approximately 70% in adults. Fine-needle aspiration (FNA) is a minimally invasive, cost-effective, standard method to collect tissue from thyroid nodules for cytologic examination. However, approximately 15% of thyroid FNA specimens cannot be unambiguously diagnosed as benign or malignant.
To investigate whether clinically actionable data can be obtained using next-generation sequencing of residual needle rinse material.
A total of 24 residual needle rinse specimens with malignant (n = 6), indeterminate (n = 9), or benign (n = 9) thyroid FNA diagnoses were analyzed in our clinical molecular diagnostics laboratory using next-generation sequencing assays designed to detect gene mutations and translocations that commonly occur in thyroid cancer. Results were correlated with surgical diagnoses and clinical outcomes.
Interpretable data were generated from 23 of 24 residual needle rinse specimens. Consistent with its well-known role in thyroid malignancy, BRAF V600E mutations were detected in 4 malignant cases. An NRAS mutation was detected in 1 benign case. No mutations were detected from specimens with indeterminate diagnoses.
Our data demonstrate that residual thyroid FNA needle rinses are an adequate source of material for molecular diagnostic testing. Importantly, detection of a mutation implicated in thyroid malignancy was predictive of the final surgical diagnosis and clinical outcome. Our strategy to triage thyroid nodules with indeterminate cytology with molecular testing eliminates the need to perform additional FNA passes into dedicated media or to schedule additional invasive procedures. Further investigation with a larger sample size to confirm the clinical utility of our proposed strategy is underway.
Thyroid nodules are very common in the general adult population, with a prevalence of approximately 70% in adults.1 Most thyroid nodules are benign, but because of their high prevalence, thyroid cancer is fairly common. Importantly, for reasons that remain uncertain, the number of new cases consistently increases annually.2
Fine-needle aspiration (FNA) biopsy is the standard method for a pathologist to examine thyroid nodules. However, approximately 10% to 15% of thyroid FNAs cannot unequivocally be diagnosed as benign or malignant, and receive an indeterminate diagnosis. The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) was published in 2010.3 This internationally recognized standard includes 6 diagnostic categories, ranging from nondiagnostic/unsatisfactory to malignant. Most categories include an estimated risk of malignancy that guides further clinical management. Using TBSRTC, nearly 80% of thyroid FNAs can be categorized as either benign or malignant. The indeterminate categories include atypia of undetermined significance or follicular lesion of undetermined significance, and follicular neoplasm or suspicious for follicular neoplasm. The risk of malignancy in these categories is institution dependent, ranging from 5% to 15% for atypia of undetermined significance/follicular lesion of undetermined significance to 15% to 30% for follicular neoplasm/suspicious for follicular neoplasm.
The diagnostic challenge posed by indeterminate thyroid FNAs results in clinical uncertainty. Management of an indeterminate FNA may include clinical observation, repeat FNA, or surgical lobectomy. Important to patient safety, scheduling additional procedures may delay the diagnosis in cases that are malignant, and surgical lobectomy may risk overtreatment of benign nodules. An alternative strategy is to triage indeterminate thyroid FNAs for rapid molecular diagnostic testing. In support, the 2015 American Thyroid Association guidelines for thyroid nodules include recommendations for molecular testing.4 Currently there are 3 major commercially available molecular tests for indeterminate thyroid FNAs.5 They include the Afirma Gene Expression Classifier (Veracyte Inc, South San Francisco, California), ThyGenX/ThyraMIR (Interpace Diagnostics, PDI Inc, Parsippany, New Jersey), and ThyroSeq v2 (University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania). Of note, each of these tests requires dedicated thyroid FNA material into special collection media. This collection is done at the time of the FNA procedure, before the cytology diagnosis is rendered. In order to use these available tests, the patient must return for a second FNA. Alternatively, an extra FNA pass could be performed at the time of the first procedure for every patient, even though it will not be needed for the 80% of cases that will have a definitive cytology diagnosis rendered. Newer on the market is a molecular microRNA-based assay, RosettaGX Reveal (Rosetta Genomics Inc, Philadelphia, Pennsylvania), which can be used on routinely prepared FNA slides.6
Although many commercially available molecular tests require specimen collection in proprietary media, liquid-based cytology has been shown to be an adequate source for DNA and RNA for molecular testing.7–10 A previous pilot study showed that FNA needle rinses in ThinPrep CytoLyt Solution (Hologic Inc, Marlborough, Massachusetts) are an acceptable specimen type for the ThyGenX molecular testing.11 Our hypothesis is that clinically relevant molecular data can be obtained from the residual, routinely discarded needle rinse material from thyroid FNA specimens.
MATERIALS AND METHODS
At the time of thyroid FNA, air-dried and alcohol-fixed smears are prepared from each pass, followed by the needle being rinsed in CytoLyt before being discarded. This needle rinse can be used to make an additional ThinPrep slide if the cellularity is low, or a cellblock if immunohistochemical stains are needed. In most cases, this residual needle rinse material is discarded. Residual thyroid FNA rinse material was collected from 24 specimens during a period of 5 months. The specimens were collected in ThinPrep CytoLyt solution (n = 22) or in RPMI (n = 2; Lonza Group AG, Basel, Switzerland). Thyroid FNA diagnoses were rendered by board-certified cytopathologists at Houston Methodist Hospital (Houston, Texas). Most cytology diagnoses followed TBSRTC.3 Cases were selected for molecular testing based on cytology diagnosis, including 9 benign, 9 indeterminate, and 6 malignant nodules. Diagnoses that were considered “indeterminate” were atypia of undetermined significance/follicular lesion of undetermined significance, follicular neoplasm/suspicious for follicular neoplasm, and any non-TBSRTC diagnosis. Our study did not include any cases with the cytology diagnosis of suspicious for malignancy. One malignant case was a benchtop FNA of a surgical resection specimen with the diagnosis of papillary thyroid carcinoma that was used as a positive control. This study was approved by the Institutional Review Board at Houston Methodist Hospital.
Residual FNA needle rinse materials are routinely stored at 4°C for 1 to 32 days (average, 14 days). Specimens selected for inclusion in this study were transferred to the molecular diagnostics laboratory and stored at −80°C.
DNA/RNA extraction and molecular testing were performed in our clinical molecular diagnostics laboratory by a trained medical technologist blinded to the cytology diagnoses. DNA and RNA were isolated from the residual needle rinse specimens using an AllPrep DNA/RNA Mini preparation kit (Qiagen, Alameda, California) according to the manufacturer's protocol.
Gene mutations were detected by next-generation sequencing performed on the extracted DNA. Ion PGM Template OT2 200 and Ion PGM Sequencing 200 kits (Life Technologies, San Francisco, California) were used according to the manufacturer's protocol. The assay detects mutations in approximately 200 targeted regions in 50 genes implicated in human malignancy. Tested genes include ABL1, AKT1, ALK, APC, ATM, BRAF, CDH1, CDKN2A, CSF1R, CTNNB1, EGFR, ERBB2, ERBB4, EZF2, FBXW7, FGFR1, FGFR2, FGFR3, FLT3, GNA11, GNAQ, GNAS, HNF1A, HRAS, IDH1, IDH2, JAK2, JAK3, KDR, KIT, KRAS, MET, MLH1, MPL, NOTCH1, NPM1, NRAS, PDGFRA, PIK3CA, PTEN, PTPN11, RB1, RET, SMAD4, SMARCB1, SMO, SRC, STK11, TP53, and VHL. The limit of detection is 10% mutant allele burden at 100× coverage.
Gene translocations were detected by next-generation sequencing performed on the extracted RNA. The FusionPlex CTL Kit (ArcherDX, Boulder, Colorado) and a NextSeq Instrument (Illumina, San Diego, California) were used according to the manufacturers' protocols. The assay detects gene rearrangements, splice site variants, and copy number variants in 195 targets in 40 genes implicated in thyroid and lung cancer. Tested genes include AKT1, ALK, AXL, BRAF, CALCA, CCND1, CTNNB1, DDR2, EGFR, ERBB2, FGFR1, FGFR2, FGFR3, GNAS, HRAS, IDH1, IDH2, KRAS, KRT20, KRT7, MAP2K1, MET, NRAS, NRG1, NTRK1, NTRK2, NTRK3, PIK3CA, PPARG, PTH, RAF1, RET, ROS1, SLC5A5, THADA, TTF1, CHMP2A, GPI, RAB7A, and VCP.
Surgical pathology follow-up was obtained from the laboratory information system. Clinical follow-up was obtained from the electronic medical record.
A total of 24 residual needle rinse specimens from thyroid FNAs from 22 patients were collected. All specimens would otherwise have been discarded. The cytology diagnoses were benign (n = 9), indeterminate (n = 9), and malignant (n = 6; Table). Adequate nucleic acid for gene mutation testing and gene translocation testing was isolated from 23 of 24 and 22 of 24 needle rinse specimens, respectively. One case with insufficient nucleic acid had very low cellularity, had a ThinPrep slide prepared to aid in the cytology diagnosis, and did not receive a diagnosis using the Bethesda System (Figure 1).
Gene mutations were detected in 5 specimens, including 4 of 6 malignant (66.7%), 0 of 8 indeterminate, and 1 of 9 benign cases (Table). Consistent with its known role in thyroid tumors, a BRAF V600E mutation was detected in the 4 positive malignant cases.
Of note, a KRAS G60R variant was detected in the 1 indeterminate case. A repeat FNA also yielded a diagnosis of atypia of undetermined significance. Because KRAS codon 60 alterations have not been implicated in tumorigenesis, this variant was considered to be predictive of a benign process. Similarly, detection of KDR K286N and APC I1557Y variants in 1 benign case were also classified as not likely being tumorigenic in a thyroid nodule. Unexpectedly, a NRAS Q61R mutation was detected in 1 benign case. On clinical follow-up, the patient had no issues related to the thyroid nodule. No gene translocations were detected in the 23 specimens (0%).
Follow-up molecular testing was performed on available surgical resection specimens. Of cases with a preceding indeterminate cytology diagnosis, BRAF V600E was detected in the case with insufficient nucleic acid on cytology. In the case with an indeterminate cytology diagnosis of suspicious for follicular neoplasm and a resection diagnosis of papillary thyroid carcinoma, follicular variant, no molecular variants were detected (Figure 2). Follow-up molecular testing on the surgical resection specimens with a malignant cytology diagnosis but no molecular alterations detected on cytology found PTEN N323fs*21 in 1 case and no molecular variants in the other. Of the cases with a malignant cytology diagnosis and BRAF V600E detected, 1 case had follow-up molecular testing on the resection specimen that confirmed the BRAF V600E mutation.
We demonstrate that thyroid FNA needle rinses, which would otherwise be discarded, are an adequate source for molecular diagnostics testing using next-generation sequencing to detect gene mutations and gene translocations.8,10,11 Our next-generation sequencing data are consistent with those of Krane et al,8 who used polymerase chain reaction to detect BRAF and RAS point mutations and RET/PTC and PAX8/PPARγ rearrangements in 597 FNA rinse specimens.8 They also generated interpretable results in most cases, including 90 with an indeterminate cytology diagnosis. Similarly, others have successfully analyzed DNA from residual thyroid FNA material collected in an ethanol-based fixative solution10 or used rinses in CytoLyt with the ThyGenX reference laboratory.11
We detected molecular variants in approximately 20% (5 of 23) of our cohort, including 1 of 9 benign, 0 of 8 indeterminate, and 4 of 6 malignant cytology diagnoses. The mutations detected were of a similar pattern as has been previously reported,8,11,12 including NRAS in a case with benign cytology, and BRAF V600E in 4 cases with malignant cytology that were confirmed to be papillary thyroid carcinoma on surgical resection. We also report 1 indeterminate case with a likely benign KRAS variant and another with KDR and APC variants detected. To our knowledge, only 1 other KDR variant has been reported in the thyroid, in a pediatric patient with an encapsulated follicular variant of PTC detected after neuroblastoma treatment, so its biologic significance is uncertain.13 One unusual case with a surgical diagnosis of follicular neoplasm with neuroendocrine features was found to have a PTEN frameshift variant (N323fs*21), which was not detected on the preceding thyroid FNA (Figure 3). Although often associated with Cowden syndrome, somatic PTEN variants have also been described in differentiated thyroid tumors.14,15
The currently commercially available molecular tests for thyroid FNAs are thyroid-specific panels. Previously, other groups have used a general DNA hotspot panel on thyroid FNAs with variable success.16,17 Our study supports the utility of a hot spot molecular panel rather than a thyroid-specific panel; the former has the advantage of being a test that was already validated in our clinical laboratory, has multiple applications, and is routinely performed. Additionally, the hot spot panel includes genes, such as AKT1, APC, PTEN, RB1, and TP53, that are known to be mutated in thyroid cancer18 but may not be included in all currently commercially available thyroid panels.
The newly renamed noninvasive follicular thyroid neoplasm with papillary-like nuclear features and subsequent reclassification as a nonmalignant neoplasm may cause some changes in thyroid cytopathology. For example, if noninvasive follicular thyroid neoplasm with papillary-like nuclear features is considered a benign diagnosis, then the risks of malignancy in many of the Bethesda System categories may decrease. In the limited literature about noninvasive follicular thyroid neoplasm with papillary-like nuclear features thus far, these cases commonly receive an indeterminate cytology diagnosis.19,20 Noninvasive follicular thyroid neoplasm with papillary-like nuclear features cases also have unique molecular features that may be detectable on the FNA cytology specimen.21 Additionally, the TBSRTC is currently under revision, and the second edition may address some of these issues.22
Routinely discarded residual thyroid FNA needle rinses are an adequate source of DNA and RNA for molecular testing. In this modestly sized study, we demonstrate that clinically relevant gene mutations can be identified from this material. This strategy has the advantage of not needing additional FNA passes into dedicated media or a return of the patient for a repeat procedure. Although our initial results are promising, larger, well-controlled trials are needed to validate the assay for routine clinical use.
This project was funded by a Houston Methodist Department of Pathology microgrant (winter 2016).
From the Department of Pathology & Genomic Medicine, Houston Methodist Hospital, Houston, Texas (Drs Fuller, Mody, and Olsen, and Mss Hull, Pepper, and Hendrickson); and the Department of Pathology, Weill Cornell Medical College, New York, New York (Drs Mody and Olsen).
The authors have no relevant financial interest in the products or companies described in this article.
A poster based on our preliminary data was presented at the United States & Canadian Academy of Pathology 2017 Annual Meeting; March 6, 2017; San Antonio, Texas.