Introduction

Thyroid carcinomas exhibit various genetic alterations [1], among which the fusion genes are important targets for molecular therapies [2,3,4]. Determining the presence or absence of these fusion genes is essential for developing novel therapeutic strategies. However, conducting comprehensive genetic analyses in every case is difficult because of technical and economic constraints [5, 6]. In Japan, since 2022, the Oncomine Dx Target Test (for thyroid carcinomas) has been approved for the genetic characterization of advanced or recurrent thyroid carcinomas. This has facilitated widespread genetic analysis of thyroid carcinomas using the Oncomine Dx Target Test, which is expected to be performed at many hospitals. Based on this advancement, we conducted detailed pathological examinations of thyroid carcinomas in which the RET and NTRK3 fusion genes were identified using the Oncomine Dx Target Test. This study aimed to elucidate the specific pathological characteristics associated with the RET and NTRK3 fusion genes and effectively identify cases likely to exhibit one or both of these fusion genes in routine diagnostic work. Additionally, we aimed to characterize the specimens that were most suitable for genetic analysis.

Materials and methods

In the present study, we enrolled patients with thyroid carcinoma from the Kanagawa Cancer Center (Yokohama, Japan) who were tested using the Oncomine Dx Target Test between May 2022 and October 2023. The rationale for using the Oncomine Dx Target Test, which is approved for use in Japan for patients with advanced or recurrent thyroid carcinoma, was driven by clinical reasons (the test was outsourced). In addition to detecting the BRAF, RAS, and other mutations, it also targets fusion genes for identification, including ABL1, ALK, AXL, BRAF, ERBB2, ERG, ETV1, ETV4, ETV5, FGFR1, FGFR2, FGFR3, MET, NTRK1, NTRK2, NTRK3, PDGFRA, PPARG, RAF1, RET, and ROS1. Among these cases, those in which the fusion genes were detected were selected. Basic clinicopathological data, including age; sex; type of fusion gene; histological findings; fusion gene partners; tumor, nodes, and metastases (TNM) classification (UICC 8th edition); and outcome, were obtained from electronic medical records and pathological diagnostic reports. The response to radioactive iodine (RAI) therapy follows the American Thyroid Association Guidelines [7]. Subsequently, the specimens were thoroughly reviewed to validate these data and conduct additional pathological analyses. These analyses included assessing the presence or absence of papillary thyroid carcinoma (PTC) nuclei, counting mitoses per 10 high-power fields, determining the presence or absence of squamous solid nests, and performing immunohistochemical analyses using BRAF V600E (Clone: VE1), Ki-67 (Clone: 30–9), and Pan-TRK (Clone: EPR17341). PTC nuclei were assessed based on a previous report using a 3-point scoring system, with a score of 2 or higher being defined as PTC nuclei [8]. Furthermore, based on a previous report, we also assessed whether “BRAF-like” atypia or “RAS-like” atypia was present [9]. The primary tumor was analyzed; however, in cases where the primary tumor could not be obtained, recurrent lymph node metastases or metastases in other organs submitted for the Oncomine Dx Target Test were used. Given the known high frequency of RET fusion genes in diffuse sclerosing variant of PTC (DSVPTC) [10,11,12], we adopted the diagnostic criteria based on a previous report for identifying relevant factors in this study [9]. At the structural level, the patterns of papillary, follicular, and solid/trabecular/insular (STI) structures were analyzed, and their percentages out of the total were evaluated. Only the percentages of papillary, follicular, and STI structures were included, excluding other structures from the denominator. The degree of fibrosis and calcification was also evaluated. The degree of fibrosis was expressed as a percentage of the total tumor lesions. For calcification, two distinct types were evaluated: psammoma bodies (a small, concentrically layered calcification formed around a necrotic tumor cell nucleus, typically found within the tumor stroma and lymphatic vessels [13]), and coarse calcification. Henceforth, the term “calcification” will refer exclusively to the latter, that is, coarse type. Calcification was evaluated on a scale of 0–3B. Each level was defined as follows: 0, no obvious calcification; 1, psammoma bodies or calcified nests that did not require a decalcification procedure; 2A, requiring only ethylenediaminetetraacetic acid (EDTA)-based decalcification for less than half of the specimens; 2B, requiring only EDTA-based decalcification for more than half of the specimens; 3A, requiring hydrochloric acid-based decalcification for less than half of the specimens; 3B, requiring hydrochloric acid-based decalcification for more than half of the specimens. When both EDTA and hydrochloric acid decalcification were present, hydrochloric acid decalcification was prioritized.

Results

In our hospital, 74 thyroid carcinoma cases were tested using the Oncomine Dx Target Test between July 2022 and October 2023. This excludes one case in which the test was canceled owing to a low tumor cell count and prior hydrochloric acid decalcification. The histological types are as follows: PTC in 66 cases, follicular thyroid carcinoma (FTC) in three cases, poorly differentiated thyroid carcinoma (PDTC) in four cases, and medullary thyroid carcinoma (MTC) in one case. No anaplastic thyroid carcinomas were present. Genetic analysis revealed that PTC primarily exhibited the BRAF V600E mutation (47 out of 66 PTC cases, 71.2%), whereas FTC exclusively showed RAS mutations (three out of three FTC cases, 100%). The RET fusion gene was detected in eight cases (8/74, 10.8%), whereas the NTRK3 fusion gene was detected in one case (1/74, 1.4%). No other fusion genes, including ALK, ROS1, and PPARG, were detected. Detailed data are summarized in Supporting Information 1. Among the nine cases exhibiting the fusion genes, the median patient age was 53 years, and the male-to-female ratio was 2:7. None of the nine patients in this study had a history of radiation exposure. For the response to RAI therapy, six cases exhibited a “structural incomplete response,” two cases exhibited an “indeterminate response,” and one case did not receive RAI therapy. In the cases with the fusion genes, primary tumors were confirmed in only seven cases (1, 2, 4, 6, 7, 8, and 9) in our hospital. Moreover, in two cases (3 and 5), we could not obtain the primary tumor because they involved surgeries performed at external hospitals. Most of the patients were diagnosed with PTC. Only one patient (case 9) exhibited lymph node metastases characteristics of a PDTC. All of the obtained cases, except for PDTC in case 9, had PTC nuclei. In addition, detailed examination revealed “BRAF-like” atypia in cases 2–4, 6, 8, and 9, and “RAS-like” atypia in cases 1, 5, and 7 (Fig. 1b–c). Note that case 9 was assessed only on the primary tumor, as the lymph node metastases had progressed to PDTC. The mitoses of the primary tumors were all < 1 per 10 high-power fields. Lymphatic invasion was observed in two cases (cases 1 and 4; 28.6%, 2/7), whereas vascular invasion was observed in all cases. Squamous solid nests were confirmed in three cases (cases 1, 2, and 4; 42.9%, 3/7). Immunohistochemical examination revealed a median Ki-67 labeling index of 1.3% for primary tumors. All cases were negative for BRAF V600E, both in immunoreactivity (Fig. 2c) and mutations. Immunoreactivity of Pan-TRK was negative in case 7 (Fig. 2d), which had the ETV6::NTRK3 fusion gene. These clinicopathological data are summarized in Table 1. Pathologically, in primary tumors, the median percentage of papillary structures was 50%, and follicular and STI structures were observed (Fig. 2a, b) while lymph node metastases exhibited a more pronounced papillary structure. The prevalence of STI structures in the lymph node groups was low. Moreover, dysmorphic clear cells, identified by highly compressed nuclei and pale-to-clear cytoplasm, were observed to varying degrees in all cases except case 4 (Fig. 1a, b). The median percentage of fibrosis in the primary tumors was 40%. In the lymph node groups, the median percentages were 20% in the N1a group and less than 10% in the N1b group. A decalcification procedure was necessary in the majority case of primary tumors but it was not needed for the N1b lymph node except for one case (Fig. 3a–d and Table 2). Furthermore, in our study, while none of the nine cases completely matched the essential criteria for DSVPTC [9], several distinct characteristics were observed. These included a predominantly young and female patient population and significant occurrence of RAI resistance. Additionally, features such as intraglandular dissemination, lymphatic invasion, squamous solid nests, and/or chronic thyroiditis were confirmed in some cases. Varying degrees of fibrosis and patterns of follicular or STI structures were also observed across many cases. These data are summarized in Supporting Information 2.

Fig. 1
figure 1

Representative microscopic images of a thyroid carcinoma with fusion genes in the high-power fields. a Carcinoma cells in the primary tumor exhibit the characteristic “dysmorphic clear cell.” It was defined as a cell with highly compressed nuclei and clear to pale cytoplasm (Case 2, H&E stain, original magnification × 400, scale bar, 100 µm). b Carcinoma cells in lymph node metastases exhibiting typical eosinophilic (pink) or pale-to-clear cytoplasm. The former contributed to the typical pink intranuclear cytoplasmic inclusions, whereas the latter contributed to the washed-out-like intranuclear cytoplasmic inclusions. So-called “BRAF-like” atypia was present. Some cells also showed highly compressed nuclei and dysmorphic clear cells. These morphologic characteristics were observed in both decalcified primary tumors and non-decalcified lymph node metastases, indicating morphologic consistency between the different specimens (Case 3, H&E stain, original magnification × 400, scale bar, 100 µm). c Carcinoma cells in primary tumors exhibit so-called “RAS-like” atypia characterized by round nuclei, powdery chromatin, and lack of intranuclear cytoplasmic inclusions (Case 7, H&E stain, original magnification × 400, scale bar, 100 µm)

Full size image
Fig. 2
figure 2

Representative microscopic images of thyroid carcinoma with fusion genes in the middle-power fields. a Despite the occurrence of papillary thyroid carcinoma, the cancer cells exhibited follicular structure in addition to papillary structure (Case 6, H&E stain, original magnification × 100, scale bar, 400 µm). b Carcinoma cells exhibit small solid or insular structures without the characteristics of poorly differentiated thyroid carcinoma, such as convoluted nuclei, significant mitotic activity, and tumor necrosis (Case 6, H&E stain, original magnification × 100, scale bar, 400 µm). c Immunohistochemical staining for BRAF V600E. Carcinoma cells showed negative immunoreactivity (Case 6, immunohistochemistry, clone: VE1, original magnification × 100, scale bar, 400 µm). d Immunohistochemical staining for Pan-TRK. Carcinoma cells showed negative immunoreactivity (Case 7, immunohistochemistry, clone: EPR17341, original magnification × 100, scale bar, 400 µm)

Full size image
Table 1 Clinicopathological findings of thyroid carcinomas with the RET or NTRK fusion genes
Full size table
Fig. 3
figure 3

Macroscopic and microscopic images of primary tumors and lymph node metastases in thyroid carcinomas with fusion genes. a Macroscopic image of the thyroid gland in Case 7 showing extensive calcification and sclerosis (gross view; scale bar, 30,000 µm). b Microscopic image of the thyroid gland in Case 7, showing severe calcification. Owing to the severe calcification, hydrochloric acid-based decalcification was required to prepare the slide (H&E stain, original magnification × 40, scale bar, 1000 µm). c Macroscopic image of the lymph node metastasis in Case 3. As the clinician had informed pathologists that there was a large lymph node metastasis, an intentional partial incision was made for proper formalin fixation (gross view; scale bar, 30,000 µm). d Microscopic image of the lymph node metastasis in Case 3. The slide has a rough appearance due to the presence of psammoma bodies; however, a decalcification procedure was not required for slide preparation (H&E stain, original magnification × 20, scale bar, 2000 µm)

Full size image
Table 2 Assessment of structural and morphological fibrosis, and calcification patterns between primary and nodal lesions in thyroid carcinomas
Full size table

Discussion

Recently, the importance of molecular-targeted drugs in the treatment of thyroid carcinoma has been increasingly recognized [3, 4, 14,15,16]. Although the significance of BRAF inhibitors is well established, the role of targeted therapies for the fusion genes (such as RET, NTRK, ALK fusion genes) is also important [17, 18]. Notably, BRAF mutations are mutually exclusive with these fusion genes [19, 20]. Therefore, the efficient detection of these fusion genes is important for the selection of appropriate molecular-targeted therapies. Of course, recent advancements in preoperative molecular testing for genes have significantly aided early thyroid carcinoma diagnosis [1, 3, 15, 16]. However, limitations such as sampling variability and the heterogeneous nature of cancer can affect the comprehensiveness of these tests [21, 22]. The cost of testing is also an issue [21, 23, 24]. Considering economic and insurance constraints, it is important to establish a process that can detect cases with these fusion genes from routine diagnostic work. Our study, which analyzed surgical specimens, complements this approach by providing a more comprehensive molecular and histological assessment.

In thyroid carcinoma, the frequency of RET fusion genes ranges from approximately 6 to 10% [25,26,27,28], which aligns with our results. Similarly, the frequency of NTRK fusion genes ranges from 2 to 7% [26, 29,30,31,32]. However, the frequency of NTRK fusion genes in our study was lower. This discrepancy could be attributed to several factors. First, in our analysis, we used the Oncomine Dx Target Test, approved in Japan only for advanced or recurrent cases. Second, our study had a comparatively small sample size. Final, the detection of the NTRK fusion gene using the Oncomine Dx Target Test could be limited owing to difficulties in covering relevant intron regions, in which the NTRK fusion breakpoints commonly occur [33]. The frequency of ALK fusion genes in thyroid carcinoma ranges from approximately 1 to 4% [34,35,36], with a slightly higher frequency of 6% in pediatric PTC cases [37]. However, in our study, we did not detect any ALK fusion genes, which may have the same issues as those found with NTRK fusion genes. Moreover, our patient cohort included advanced cases, most of which were RAI therapy-resistant. This might introduce a selection bias. In addition, among the 66 PTC cases (10/66, 15.2%), ten exhibited no identifiable genetic alterations. A recent report indicates that the categorization of such PTC cases with no initial driver event identification as “dark matter” PTCs might be premature [9]. The absence of detectable genetic alterations in these cases may be because of limitations of the testing methods and may require a more comprehensive diagnostic approach, if necessary.

In our histological analysis, we observed the presence of non-papillary (follicular and/or STI) structures, compressed nuclei, and cells with clear to pale cytoplasm (reminiscent of dysmorphic clear cells), along with notable fibrosis and calcification. These features were prevalent in several cases, suggesting distinct histological patterns. Given that BRAF V600E mutations in papillary thyroid carcinomas are associated with distinct morphological features, including a well-developed papillary architecture [38], it can be hypothesized that non-papillary structures may be more pronounced in thyroid carcinomas with RET and/or NTRK3 fusion genes. In addition, case 7, the sole case with a detected NTRK3 fusion gene, exhibited “RAS-like” atypia. Each case with a RET fusion gene presented distinct morphologic features, with some cases showing “BRAF-like” atypia (cases 2–4, 6, 8, and 9) and others exhibiting “RAS-like” atypia (cases 1 and 5). This variability in cases with RET fusion genes and “RAS-like” atypia in cases with NTRK fusion genes is consistent with that in a previous report [9] and underscores the morphological distinctions associated with these specific genetic changes. In addition, dysmorphic clear cells were observed in both decalcified primary tumors and non-decalcified lymph node metastases, indicating morphological consistency between the different specimen preparations. Despite the small number of cases, the overall morphology of the cases with the RET fusion gene and the NTRK3 fusion gene was similar. This finding is consistent with that of a previous report on sarcoma cases with these fusion genes [39]. In this study, all cases with detected fusion genes exhibited negative immunoreactivity for BRAF V600E. These findings suggested a high probability of the presence of the RET or NTRK fusion genes in cases with various structural patterns, dysmorphic clear cells, and background fibrosis and/or calcification. However, as confirmed in case 4, the absence of these morphological characteristics does not indicate the absence of these fusion genes, emphasizing the limitations of morphological assessment alone. Nevertheless, the negative immunoreactivity for BRAF V600E in all cases indicates that the combination of morphological analysis with immunohistochemistry may improve the prediction for thyroid carcinoma harboring the RET and/or NTRK fusion genes. It should be noted, however, that Pan-TRK immunohistochemistry may not always yield positive results in cases with the ETV6::NTRK3 fusion gene, as previously reported [9].

Meanwhile, our cases exhibited some similarities to those of DSVPTC. These included a relatively young patient population and the occurrence of RAI resistance. Intraglandular dissemination, lymphatic invasion, squamous solid nests, and/or chronic thyroiditis were noted in some cases. Additionally, varying degrees of fibrosis and patterns of follicular or STI architecture were observed across many cases. However, it is important to recognize that in Japan, the Oncomine Dx Target Test is used primarily in advanced or recurrent cases. This specific limitation for the use of the test may explain the lack of definite DSVPTC in our series. Despite this, the clinicopathological similarities we observed suggest a potential association between thyroid carcinomas with detected fusion genes and DSVPTC. Furthermore, some reports have suggested that thyroid carcinomas with the RET and/or NTRK fusion genes tend to be aggressive [28, 29, 40, 41]; our findings indicate a more complex scenario. Although all our cases showed low mitotic activity and a low Ki-67 labeling index, distant metastases were observed in cases 1–8, and a PDTC component was present in case 9. This implies that the level of aggressiveness was not fully captured by these markers.

We would like to shed some light on the challenges and considerations in decalcification processes, particularly in thyroid carcinoma harboring fusion genes and varying degrees of calcification. In our hospital, the choice between EDTA-based and hydrochloric acid-based decalcification depended on the degree of calcification. EDTA-based decalcification is preferred to preserve nucleic acid quality [42, 43], which is important for detecting fusion genes. However, in cases of more severe calcification, which prevents sectioning for pathological diagnosis, hydrochloric acid-based decalcification is required despite the risk of affecting nucleic acid quality [42, 43]. In this study, significant calcification was observed in many primary tumors that required hydrochloric acid-based decalcification. In contrast, calcification in the lymph node metastases, particularly in the N1b group, tended to be mild. These findings suggest that lymph node metastases, particularly those in the N1b group, may be more suitable for genetic analysis of thyroid carcinoma cases. The reason for the mild calcification in the lymph node metastases is unclear. In our study, younger patients, such as those in cases 1 and 2, exhibited mild calcification of the primary tumor. This finding implies that mild calcification in lymph node metastases may reflect their recent occurrence compared to primary tumors. As these metastatic cells are derived from a late-stage primary tumor, their stay in the lymph nodes is relatively short, which may result in mild calcification. However, as lymph nodes are located in adipose tissues and are covered by a capsule [44], it may be necessary to remove the surrounding adipose tissue and make special incisions to obtain the specimen to ensure proper formalin fixation for genetic analysis.

In conclusion, our study has identified unique pathological characteristics associated with the RET and NTRK3 fusion genes in thyroid carcinoma cases, such as non-papillary structures, compressed nuclei, dysmorphic clear cells, calcification, and similarities with DSVPTC. These characteristics could serve as indicators for considering fusion gene testing. Additionally, our analysis of the varying degrees of calcification between primary tumors and lymph node metastases, particularly in the N1b group, underscores the importance of precise sample processing for efficient genetic analysis. Put together, these insights contribute to a more targeted approach in the molecular diagnosis and treatment of thyroid carcinomas (details in Supporting Information 3).

Limitations

This study has several limitations. This study involved a limited number of cases and subjective pathology analysis. As the Oncomine Dx Target Test is approved only for advanced or recurrent thyroid carcinoma cases in Japan, there may have been bias in the analyzed cases. Moreover, decisions regarding the selection of decalcification procedures (EDTA- or hydrochloric acid-based) depended on the judgment of the clinical laboratory technicians.