What's new?

Despite developed diagnostic methods for thyroid nodules, fine‑needle aspiration biopsy (FNAB) may still provide inconclusive results. This is the first study evaluating a wide panel of potential malignancy markers in material obtained from glass slides with FNAB specimens. We showed that the heterogeneity in the genetic background of indeterminate thyroid nodules corresponds to their histopathologic diversity. Of the gene mutations studied, the KDR gene tended to be more common in malignant samples and may be a possible malignancy marker. Glass slides with FNAB samples were shown to be a reliable source of genetic material for NGS studies.

Introduction

Thyroid nodules may be detected in up to 67% of the adult population.¹ Ultrasound examination remains the gold standard for a rapid diagnosis of potential malignancy.^2,3 It may be complemented by fine‑needle aspiration biopsy (FNAB). Both methods enable a diagnosis and choice of appropriate treatment in 75% to 90% of patients.⁴ However, according to the Bethesda reporting system,⁵ cytologic results of FNAB are inconclusive in 10% to 25% of patients with thyroid nodules.⁶ These results include atypia or follicular lesion of undetermined significance (AUS/FLUS; Bethesda category III) and follicular neoplasm or suspicious for follicular neoplasm (FN/SFN; Bethesda category IV). In those cases, a differentiation between follicular thyroid adenoma and follicular thyroid cancer (FTC) is possible only on the basis of a histopathologic assessment of vascular or capsular invasion (or both),⁷ thus making surgery unavoidable.⁸

To avoid the need for a diagnostic surgery, new markers of presurgical differentiation between follicular thyroid adenoma and FTC are being extensively studied. Currently, one of the main areas for research is the identification of genetic alterations that may become markers of malignancy.⁹ The BRAF, 167 Gene Expression Classifier (GEC), and ThyroSeq tests are dedicated to the assessment of indeterminate thyroid nodules.¹⁰ However, they have not yet provided the results that would be sufficiently accurate for routine clinical use,¹⁰ while the comprehensive assessment of the genetic background of AUS/FLUS, FN/SFN, as well as FTC (not covered in the Cancer Genome Atlas Program) is still lacking.

The aim of the present study was to assess the genetic background of indeterminate thyroid nodules based on the material from FNAB samples in order to identify a mutational status that might facilitate the understanding of these nodules. We also aimed to discover novel genetic pathways potentially involved in FTC.

Patients and methods

Results

The most common final benign diagnosis on histopathologic examination was a hyperplastic nodule (n = 15; 60%), followed by a follicular adenoma (n = 3; 12%). The most common malignant diagnosis was papillary thyroid cancer, diagnosed in 6 patients (24%): TNM stage pT1aN0M0 in 3 patients (50%) and pT2N0M0 in the other 3 patients (50%). The malignancy rate was 28% (18.8% and 44.4% for FLUS and SFN, respectively). The final malignant diagnosis more often coexisted with CLT than benign diagnosis (OR, 12; 95% CI, 1.16–123.69; P = 0.03; Table 2).

Any mutation from the 50‑gene Cancer Hotspot Panel v2 was identified in 14 patients, which constituted 56% of all FLUS and SFN samples. The RET mutation was detected in 11 patients (79% of the samples, regardless of the final outcome). The results showed that 10 samples (56% of all samples with benign diagnosis) harbored mutated genes. The most common was RET (n = 8; 80%), followed by TP53 (n = 7; 70%), FLT3 (n = 6; 60%), APC (n = 6; 60%), and PDGFRA (n = 5; 50%). In contrast, among the samples with malignant diagnosis, mutations were detected in 4 (24%). The most frequent were KDR and RET (both n = 3; 75%), followed by TP53 (n = 2; 50%). The KDR mutation tended to be more common in cancer samples than in benign ones (n = 3; 75% vs n = 2; 20%; P = 0.1). The ERBB4, IDH1, JAK3, PTEN, RB1, and SMAD4 mutations were only found in benign samples. There were no mutations found exclusively in cancer samples. The genetic landscape of both groups is presented in Figure 1.

The total number of mutated genes was significantly higher in patients with benign tumors (17 vs 11, P = 0.02). The mean total number of mutations per every specimen did not differ between groups (4.9 mutations per sample for the entire group; range, 1–9). The distribution of mutations, type of genetic changes (point mutations [single nucleotide variants: intron, missense, synonymous variants] as well as duplications, insertions, and deletions), and the results of in silico analyses are presented in Table 3.

Discussion

In our study, we assessed the genetic alterations of indeterminate thyroid nodules in preoperative FNAB specimens, using a large NGS panel. The results showed that the genetic heterogeneity of the FLUS and SFN samples in our patients corresponded with the heterogeneity of the final histopathologic diagnoses. A total of 17 mutated genes were found. Also, the number of mutations in individual samples was high, with a mean of 4.9 mutations per sample. This heterogeneity may be due to different routes of proliferation and differentiation,¹⁶ driven by completely different signals for benign and malignant diagnoses,^17,18 as well as tumor heterogeneity.¹⁹ The latter can be divided into intertumor heterogeneity, which shows diverse genetic alterations based on the tumor sites, and intratumor heterogeneity, which contains different genetic alterations within the same tumor.²⁰ Intratumor heterogeneity may derive from specific morphohistologic features, whereas at the molecular level, clonal intratumor heterogeneity originates from genomic instability, and nonclonal, from a microenvironment interaction.²¹ An extremely diverse genetic background, aggressive clone, and tumor heterogeneity in thyroid cancer have been described before.^22-24

The overall malignancy rate in our patients was 28%. The previously described rates showed a considerable variation. For AUS/FLUS, the rates of 12%,²⁵ 26.6%,²⁶ and 29.9%²⁷ were reported, and for FN/SFN, 10.6%,²⁸ 20.8%,²⁹ 32.6%,³⁰ 37%,³¹ 49%,³² and up to 53%.³³ The discrepancies may relate to random variation, prior patient selection such as through referral bias, or institutional differences in pathologic interpretation.²⁶ Nonetheless, the CIs reported in those series largely overlap with those in the current study, further supporting cancer incidence in FLUS and SFN shown by our group.

In recent years, significant advances have been made in thyroid cancer research. Molecular diagnostics of thyroid nodules extend the initial diagnostics of thyroid cancer, and molecular tests help distinguish between benign and malignant nodules, particularly in indeterminate FNAB results (Bethesda III–V).² The fast‑evolving NGS technology offers a cost‑effective approach in cancer genomics and thyroid cancer.²⁰ The NGS has already been used to study indeterminate thyroid nodules also based on DNA acquired from preoperative FNAB specimens.^27,34,35 However, instead of commercially available thyroid‑specific panels such as the 7‑gene panel test, which covers a limited number of gene alterations (BRAF and NRAS/HRAS/KRAS point mutations and RET/PTC and PAX8/PPARg translocations),³⁶ or its successors (v2, v2.1, and v3),³⁷ we used a large NGS panel of not only thyroid‑specific genes. The panel, in addition to BRAF, RAS, and RET hotspot mutations, covers numerous other genes involved in thyroid carcinogenesis and malignant progression (eg, AKT1, APC, ATM, CTNNB1, PI3CKA, PTEN, RB1, and TP53).^38-40 The quality of DNA obtained from cytology specimens is considered to be better than that of formalin‑fixed paraffin‑embedded tissue.²⁰ Besides, as previously reported, NGS can also provide information on routine smears that have suboptimal DNA input quantity and quality as well as post‑sequencing metrics.⁴¹

Ultrasound features considered to be predictors of malignancy do not allow a reliable differentiation between malignant and benign nodules,³ neither do the GEC or the ThyroSeq v2 tests.⁴² Therefore, our goal was to identify novel genetic pathways potentially involved in FTC. KDR and RET (both present in 75% of samples), followed by TP53, were among the most common mutated genes found in our study in patients with a cancer diagnosis. KDR tended to be more common in malignant than benign samples, and as such, it appears to be a possible candidate as a genetic marker for malignancy. Its role was previously reported in breast cancer.⁴³ KDR encodes kinase insert domain receptor of the vascular endothelial growth factor, which acts as a mediator in the regulation of angiogenesis, vascular development, vascular permeability, and embryonic hematopoiesis, as well as promotes proliferation, survival, migration, and differentiation of endothelial cells. KDR gene mutations may play a major role in tumor angiogenesis.⁴⁴

The number of mutated genes was higher in the group with benign diagnosis. This may indicate the complex genetic background not only of thyroid cancers but also of follicular adenomas or other benign nodules. For instance, a determination of the tumors and malignant transformation is still related to common KDR, RET, and TP53 alterations, which may result from subclonal evolution. Both KDR and RET encode kinase domain receptors. The important presence of the RET gene mutation and its possible diagnostic value for identifying benign tumors have been previously described by our team.⁴⁵ However, the high level of genomic dysregulation found in benign tissue, as reflected by a high number of identified mutations and exclusive contribution of the mutated ERBB4, IDH1, JAK3, PTEN, RB1, or SMAD4 genes (not found in malignant cases) as the markers of indolent thyroid tumors, requires further comprehensive studies. The lack of the BRAF mutation, reported also in other FTC studies,⁴⁶ and the occurrence of KDR and RET may encourage further research of new genetic pathways of follicular malignancy. The in silico analyses showed that numerous mutations commonly occur in other malignancies, showing a possible intersection of cancer pathways, beyond the previously known ones for thyroid cancer. The detailed description of genetic findings shows that the mutations interpreted by various algorithms as benign or possibly benign have been reported to occur in various other malignancies, which may prove that the role of particular mutations may be more complex and our understanding of them still incomplete. The specific role of each variant is to be determined in functional studies.

Another interesting finding is that the malignant final diagnosis showed a high degree of correlation with CLT. This is in line with previous reports showing that in patients with differentiated thyroid cancer, the occurrence of CLT was higher than in those with benign thyroid diseases.⁴⁷ This link was also studied in detail regarding CLT‑associated clinicopathologic features of thyroid cancer.⁴⁸

Although the Polish population is in part considered to have iodine depletion,⁴⁹ previous research showed no differences in genetic alterations of thyroid cancers between iodine‑rich and iodine‑deficient countries, which may suggest that iodine intake might not affect the genetic alterations of differentiated thyroid cancer.⁵⁰ This makes our study findings even more generalizable.

The major limitation of our study is the small number of the analyzed cases. Therefore, the conclusions should be formulated with caution, and a much larger group of patients is needed to clarify the clinical importance of the findings. However, this preliminary study emphasizes the potential utility of applying a broader NGS panel to describe the genetic background of indeterminate thyroid nodules. Despite the lower statistical power, the present results indicate the genetic heterogeneity of indeterminate thyroid nodules. Aiming to increase the homogeneity of the group and the reliability of results, we opted only for the FLUS and SFN samples.

In conclusion, our study shows that the heterogeneity of the genetic alterations of indeterminate thyroid nodules corresponds to its histopathologic heterogeneity, which may be even higher for benign tumors. The role of KDR as a potential malignancy marker needs to be confirmed, as does the possible role of ERBB4, IDH1, JAK3, PTEN, RB1, or SMAD4 as the markers of indolent thyroid tumors. Glass slides with FNAB samples may constitute a reliable source of genetic material for NGS analysis. The new genetic pathways for FTC should be searched for using new sequencing technologies, such as NGS with large gene panels, to improve our understanding of the genetics of thyroid nodules and its overlap with other malignancies. Consequently, it may facilitate a more effective preoperative cancer diagnosis of thyroid nodules.