Neoplasms of the thyroid gland are the largest group of malignancies in the endocrine system [3]. Thyroid nodules, likewise, represent a significant reason for endocrinology outpatient visits. The management of thyroid nodules hinges on assessing the risk of malignancy [4]. Therefore, accurately determining the nature of the nodule is crucial. Recently, molecular profiling of thyroid nodules has gained traction. While these molecular tests have become integrated into clinical practice in the United States, their usage still needs to be improved in many other countries [5]. In this study, for the first time in Turkey, molecular genetic analysis of thyroid nodules was performed by an NGS panel on 20 thyroid FNAB samples. Somatic variants were detected in ten samples located in six genes: NRAS (2), BRAF (3), PIK3CA (2), TP53 (1), TERT (1), PTEN (1)).
According to the ATA 2015 guideline, the malignancy risks of Bethesda Group III and IV nodules are 5–15% (median 14%) and 15–30% (median 25%), respectively [2, 4]. However, risk data can vary between publications. For instance, in a study from China, the malignancy rate of nodules with indeterminate cytology was 71%, significantly higher than the 10–40% rate observed in the United States [5]. Especially in tertiary care centers, the malignancy rate in this group of nodules can reach 52% [6]. In our study, postoperative pathology results of 5 out of 20 samples were benign (25%), and 15 of them (75%) were malignant. Several factors contribute to our study's high percentage of malignancy rates. One of these is that surgical decisions were not solely based on the Bethesda classification but also considered ultrasonography findings and patient characteristics. As a result, the nodules that underwent surgery were more likely to have a high suspicion of malignancy. Another significant factor is that this period coincided with the COVID-19 pandemic, reducing the number of surgeries. Patients at lower risk may have chosen to postpone surgery. While sample selection was random, the high malignancy rate affects sensitivity, specificity, NPV, and PPV. For future studies, a larger sample size may be preferable.
According to recently published guidelines, repeating the FNAB or reevaluating the risk using molecular tests are two different options within the same approach for managing indeterminate thyroid nodules [2]. FNAB repetition leads to a diagnosis of AUS/FLUS cytology in 10–30% of specimens in this group the second time and fails for the rest [7]. It is recommended to choose based on patient preference, cost-effectiveness, and accessibility and to avoid an approach that will not alter patient management [4]. Significant somatic variants were found in 10 out of 20 samples in our study. If this analysis were conducted in routine practice as the second step in pathological assessment, nine out of 10 samples would have been correctly referred for surgery with just a single FNAB. The sole specimen with benign postoperative pathology among the variant-positive samples carried the PIK3CA-E545A variant. Despite having an allelic fraction of under 5%, it was reported to be a well-defined oncogenic alteration in many different cancers. This variant may represent a relatively small clone. The PIK3CA-E545A mutation was initially identified in the cribriform-morular variant PTC and is not frequently reported in thyroid carcinoma [8].
The TCGA study found at least one molecular alteration in 97% of papillary thyroid carcinomas, with 90% of these being well-known point mutations or gene fusions [9]. Our study found no significant variant in six malignant samples (Table 3). An important reason for this was that the panel in the study was not designed for gene fusion and other rearrangement detection. The study design was based solely on DNA to reduce costs and obtain optimal nucleic acid quality from archival material. Extracting RNA from tissues not stored under appropriate conditions is much more challenging than securing DNA, and nucleic acid degradation presents a significant issue [10]. The data for our study were obtained from the archived material without requiring additional intervention.
The mean age at diagnosis of thyroid cancer is 50 [11]. Genetic driver mutations in thyroid neoplasms display age-related patterns [12, 13]. Translocations involving tyrosine kinase receptors dominate the genetic etiology in the pediatric age group, while point mutations of BRAF and RAS take the lead in the etiology in the adult group [13]. The most common molecular change is RET/PTC translocation in the pediatric group and BRAF-V600E mutation in adults [14]. In our study, the mean age of patients with a variant was 56, while patients without a variant were 46. There may be several reasons for this difference. Firstly, there could be a genetic etiology difference between the young population and the elderly population. Genomic rearrangements are more frequent in the younger population and may have been overlooked in our study. Secondly, malignancy is less common in the younger population, so a driver mutation would not be anticipated. However, the second explanation is less likely, as in the study sample, young patients also had malignant pathology, yet no significant genetic variation was detected through molecular analysis. In more detail, there were six patients under the age of 45 in the sample, and their ages were 18, 21, 37, 41, 42, and 45 (Fig. 1). The genetic results of five out of these six patients did not provide information about their pathology results (inconsistent/uncorrelated results). The remaining one patient (41-year-old) had a PIK3CA inversion (AF: 4%), which was classified as Tier 2, and the variant had the lowest allelic fraction together with the other reported PIK3CA variant (AF: 3%). Considering all these data, five patients were under the age of 45 among seven patients with uncorrelated data between genetic analysis and pathology results (Fig. 2). The remaining two uncorrelated results belonged to 62-year-old and 59-year-old patients. In a 59-year-old patient, the PIK3CA variant was detected with a 3% allelic fraction and classified as Tier 2. Given the low allelic fraction status, the first explanation for this is that it may represent a minimal clone that could have been overlooked in the pathology examination, or the second explanation is that this alteration might not have caused the cancerous change at all. If PIK3CA variants with AF below 5% had not been reported in our study, six of the seven discordant patients would have been under 45. It highlights that the test's efficacy is limited in the younger age group.
PIK3CA variants are more common in anaplastic thyroid carcinoma than well-differentiated thyroid carcinomas (0.5% vs 18%) [15]. They are often associated with a poor prognosis and may co-occur with BRAF-V600E. The most prevalent PIK3CA mutation, particularly in anaplastic thyroid carcinoma, is PIK3CA-E545K [15]. In a previous study involving an adult group, PIK3CA mutations were most associated with the follicular variant of papillary thyroid carcinoma among well-differentiated thyroid cancers [16]. It has been previously suggested that certain PIK3CA single nucleotide polymorphisms (SNPs) may impede PIK3CA amplification, potentially inhibiting the formation of FTC [17]. Primary PIK3CA mutations are not expected in PTC, as most are secondary mutations. In our study, PIK3CA mutation was detected in two patients. While one sample showed malignant pathology, the other was benign. The PIK3CA variants had the lowest allelic fraction among all variants in the study (less than 5%). Variants in PIK3CA were reported because they are classified as Tier 2 according to the literature. Both were in exon 2, where oncogenic mutations are commonly found. Among all the reported variants in our study, variants in PIK3CA have the lowest level of confidence and the most uncertain clinical significance.
The molecular mechanisms identified in thyroid cancer have also influenced the previous classification, primarily based on tissue architecture. In 2022, the WHO updated the classification of thyroid cancers, placing particular emphasis on molecular characteristics in the 5th version of the classification, even separating some groups due to their molecular origins [3]. For instance, the follicular variant of papillary thyroid carcinoma (FVPTC) has been subdivided into groups based on its molecular foundation in the new WHO classification [3]. Infiltrative (not encapsulated) FVPTC remained a subtype of papillary thyroid carcinoma, while invasive encapsulated FVPTC is distinguished from PTC due to its molecular basis closer to FTC. The fact that encapsulated FVPTCs exhibit RAS-like molecular characteristics, rather than BRAF-V600E and are differentiated from follicular carcinomas only by nuclear features has led to their inclusion in another major malignancy category (invasive encapsulated follicular variant papillary carcinoma, IEFVTP). Additionally, Non-Invasive Follicular Thyroid Neoplasm with Papillary-Like Nuclear Features (NIFTP) was separated from malignant follicular neoplasms and included in the low-risk neoplasm subgroup [3]. In our study, two patients with NRAS-Q61R positive and follicular variant PTC pathology detected according to the old classification both had capsules in their tumors. In our study, two patients with NRAS-Q61R positivity and follicular variant PTC pathology, as per the old classification, had capsules in their tumors. While one lacked capsule invasion, the other tumor did show capsule invasion. These cases are now categorized as IEFVTP and NIFTP according to the 2022 classification. These tumors are now considered to be close to FTC rather than PTC. Similarly, BRAF-K601E positivity was detected in one sample. The pathology result of the patient was FV-PTC. Upon re-evaluation of the microscopic features of the tumor, it was classified as invasive encapsulated FVPTC according to the new WHO 2022 classification. This re-evaluation result aligns with the molecular basis. Invasive encapsulated FVPTC belongs to the RAS-like group, as discussed above, and BRAF-K601E positivity can be detected in up to 10% [3]. BRAF-K601E has been associated with less aggressiveness [18].
Notably, an oncocytic variant PTC was detected in both cases with NRAS-Q61R positivity. Among our 20 samples, three had oncocytic variant PTC, and NRAS-Q61R positivity was found in two (66%). Although RAS mutations in FTC and PTC with an oncocytic variant have been reported in the literature, the oncocytic variant PTC is more likely to be in the BRAF-V600E-like group [19]. The pathological type of our patients changes from FV-PTC to IEFVTP and NIFTP according to the 2022 WHO classification, as discussed in detail above. Therefore, they are not expected to be in the BRAF-V600E-like group. However, one study reported that they did not detect NRAS-Q61R or any RAS positivity in the oncocytic variant FTC/FTA [20]. In this sense, our data show differences from the literature.
In 2015, Beg et al. reported that loss of PTEN was associated with FV-PTC [21]. PTEN promoter region methylation is frequent in follicular tumor formation [22]. Although PTEN regulates central mechanisms in thyroid cancer, intergenic mutations have rarely been reported. In our study, PTEN-N48K mutation was detected in one case. PTEN-N48K is a mutation that has been previously identified as pathogenic in other cancers and is known to be oncogenic. We classified it as Tier 2 because it has never been previously reported in thyroid cancer.
There are several prominent platforms worldwide for evaluating thyroid nodules. Among them, Afirma (latest version GSC) and Thyroseq (latest version v3) stand out due to their extensive research [23]. According to the ATA guideline recommendations, it has been reported that the test for molecular analysis of the nodule should have high PPV and specificity (rule-in test, inclusion test) or high NPV and sensitivity (rule-out test, exclusion test) [2]. None of the molecular tests available in the guidelines have been tagged as the gold standard [5]. In a meta-analysis, the sensitivity of Afirma GSC in 472 nodules was 96.6% (95% CI = 89.7–98.9%), specificity was 52.9% (CI = 23.4–80.5), PPV 63%, and NPV 96%. The sensitivity of Thyroseq v3 over 550 nodules was 95.1% (91.1–97.4%), specificity was 49.6% (29.3–70.1%), PPV 70%, and NPV 92% [23]. The same study found no statistically significant difference between the two tests [23]. Due to the unavailability of Afirma or Thyroseq in China, the results of RNA and RNA/DNA-based tests, which were designed for the study for the first time, were shared in a study from China. In these RNA and RNA/DNA-based tests, PPV was 81.6% and 84.4%, respectively; NPV was 66.7% and 61.5% [5]. In our study, NPV was 40%, PPV was 90%, sensitivity was 60%, and specificity was 80%. The fact that our study's NPV and sensitivity rates are lower than other studies suggests that the panel should be evaluated as a rule-in test rather than a rule-out test.
The most significant limitation of this study is the small sample size. Additionally, the benign sample group is relatively small and disproportional. These factors may have affected the PPV, NPV, specificity, and sensitivity results of the study. In the study, considering cost-effectiveness, a non-specific and solely DNA-based next-generation sequencing technique was used. Genome rearrangements and expression profiling, especially in the younger patient population, would increase the power of the test. RNA-only-based tests do not cover changes in non-coding parts, such as the promoter region of the TERT gene. In our study, 67 genes were targeted, but changes were detected in six genes. Designing a panel of specific genes for the thyroid could increase cost-effectiveness and facilitate analysis. Nevertheless, it could be concluded that the panel used in the study would be much more efficient in the group over the age of 45 than in the younger group.
In summary, molecular tests are increasingly vital in the follow-up and treatment within daily medical practice. Similarly, they offer a promising option for managing thyroid nodules. As advances continue in genetics, molecular analysis will play an even more prominent role in future applications. For the first time in Turkey, our study investigated molecular genetic changes in Bethesda III and IV thyroid nodules using an NGS panel. It also demonstrated that FNAB archival materials are suitable for DNA-based genetic analysis. NPV, PPV, sensitivity, and specificity of the test were 40%, 90%, 60%, and 80%, respectively. According to these results, the test could be a rule-in test. However, it has a low NPV and sensitivity rate for the rule-out test. The main disadvantage of the test is its inability to detect genomic rearrangements. It is thought that the sensitivity and NPV values of the test could be increased by the RNA-based parameters. Since the genetic etiology is affected by age, it is considered a more appropriate test for the population over 45 years old. Lastly, a thyroid-specific panel would increase the power of the test and contribute more information to patient management. Our study aimed to emphasize the importance of this area by conducting the first molecular analysis of thyroid nodule archival material in Turkey. More comprehensive studies are still needed.