Thyroid nodules are common and clinically significant because about 5% of thyroid nodules are malignant (1). The incidence of thyroid cancer is on the rise and is expected to become the fourth most common cancer globally (2). The distinction between benign and malignant thyroid nodules has important therapeutic implications. Fine needle aspiration (FNA) cytology is the gold standard for the diagnosis of thyroid nodules. However, its results indicate cytologically indeterminate nodules (Bethesda classification III-IV) in approximately 20% of cases, posing a diagnostic challenge (3).

Approaches for obtaining information on cancer risk are necessary for establishing management strategies. With the recent development of molecular biological research and genetic analysis technology, molecular tests for the accurate diagnosis of thyroid cancer are being developed rapidly (4, 5). Improved molecular testing for common somatic mutations and identification of mRNA and microRNA expression classifiers have emerged as the most promising approaches (6). Somatic mutation testing is currently the most studied molecular diagnostic test for FNA biopsy (7, 8). However, mutation tests, except for the BRAF mutation, are not yet widely available in the real clinical world, and mutations alone cannot explain all aspects of the tumor. The analysis of differentially expressed genes has emerged as an alternative to mutation testing (9). However, although up to 70% of the human genome is transcribed to RNA, only 2% of it represents protein-coding genes, and transcript levels are not sufficient for the prediction of protein levels (10, 11). MicroRNAs act by base-pairing with target mRNA to negatively regulate the latter’s expression; recently, microRNA analysis has gained an important place in the study of molecular markers (12–14).

The Encyclopedia of DNA Elements project revealed that approximately 80% of the human genome is transcribed into 14,880 long non-coding RNAs (lncRNAs) from 9,277 loci (15). LncRNAs are defined as non-protein coding transcripts longer than 200 nucleotides, although recent literature suggesting some lncRNAs can encode small peptides or small proteins (16). Currently, lncRNA research has moved to the forefront of human cancer research, as recent findings have revealed a crucial role for lncRNAs in gene modulation. LncRNAs are primarily involved in the epigenetic regulation of the expression of various genes at different levels, including chromatin, splicing, transcription, and the post-transcription stages (17–19). LncRNAs are potent biomarkers, although their expression levels are considerably lower than those of mRNAs. Because lncRNAs do not encode proteins and perform their biological functions directly, most of their functions correlate with their expression levels (15, 17). The expression patterns of lncRNAs are more tissue-specific than those of protein-coding genes, and this property makes them useful as highly specific diagnostic biomarkers (20).

The aims of this study were to evaluate the diagnostic value of selected lncRNAs from our previous study and The Atlas of Noncoding RNAs in Cancer (TANRIC) thyroid cancer dataset to determine whether lncRNA expression analysis can be a promising approach for advancing the molecular diagnosis of papillary thyroid cancer (PTC) (21, 22).

In this study, we evaluated the diagnostic value of selected lncRNAs for the molecular diagnosis of PTC ( Figure 7 ). LncRNA expression analysis exhibited high diagnostic efficiency, and certain combinations of lncRNAs exhibited better results than single lncRNAs in the differential diagnosis of benign thyroid tumors and PTC.

First, we selected candidate lncRNAs with PTC diagnostic values by analyzing TANRIC thyroid cancer dataset, an open-access resource for interactive exploration of lncRNAs in cancer (22). In our previous study, we established that one of the candidates, LINC02082, had higher levels in almost all (60/64, 93.8%) PTC tissues than in the matched normal tissues (21). For the diagnosis of PTC, the differences in expression between malignant nodules and benign nodules are more important than those between malignant nodules and normal thyroid tissues. Therefore, to check and confirm the separation capability of candidate lncRNAs, we compared the expression levels of candidate lncRNAs in the malignant nodular tissues, benign nodular tissues, and paired normal thyroid tissues of a few patients who underwent thyroidectomy at our institution. The expression levels of LRRC52-AS1, LINC02082, UNC5B-AS1, MPPED2-AS1, LNCNEF, and LOC100129129 were significantly increased or decreased in malignant nodules compared to those in paired normal thyroid tissues and also in benign nodules.

Next, ROC curves were used to analyze the potential diagnostic efficacy of these lncRNAs in TANRIC thyroid cancer dataset, and these lncRNAs showed potential for diagnostic use. It was expected that the potential diagnostic efficacy could be further improved when using the modified expression level as calculated by dividing the increased lncRNA level by the decreased lncRNA level. When measuring the RNA expression levels for cancer diagnosis, a housekeeping gene control such as GAPDH, whose expression levels remain the same, is required (26). However, most lncRNAs have a low expression level, and there is no known lncRNA housekeeping gene; most housekeeping genes, such as GAPDH, have a relatively high expression level, which may be inappropriate for accurate lncRNA quantitative analysis (27). When dividing the increased lncRNA level by the decreased lncRNA level, the use of the housekeeping gene becomes unnecessary, and by maximizing the difference in the levels, it is possible to significantly increase the accuracy of cancer diagnosis.

We applied the data of these lncRNAs to FNA samples for verification. However, in the preliminary FNA test, the lncRNAs whose levels decreased in PTC did not exhibit pronounced diagnostic performance as expected. Moreover, due to the limited amount of FNA samples, we focused on the lncRNAs whose levels increased in PTC. The lncRNAs whose levels increased in PTC exhibited a favorable diagnostic performance, and two or more lncRNAs above the cut-off value showed an even better diagnostic performance. Additionally, in a cumbersome category, Cat III, these lncRNAs showed a remarkable diagnostic value. These data suggest that LRRC52-AS1, LINC02082, and UNC5B-AS1 can be used to differentiate between benign and malignant tumors in thyroid nodules by measuring their expression levels, which has not been reported in PTC thus far.

Interestingly, the expression levels of LRRC52-AS1, LINC02082, and UNC5B-AS1 were not elevated in a surgical specimen of non-invasive encapsulated follicular thyroid neoplasm with papillary-like nuclear feature (NIFTP), although the data were not included in the study. These lncRNAs are believed to aid in the diagnosis of borderline thyroid tumors (hyalinizing trabecular tumor, follicular tumor of uncertain malignant potential, well-differentiated tumor of uncertain malignant potential, and NIFTP) that are difficult to diagnose; however, additional research such as follicular neoplasm is needed (28).

Although the data from this study demonstrated the diagnostic value of lncRNAs, interestingly, limited information is available regarding the function of these lncRNAs. Zhou et al. showed that LRRC52-AS1 is associated with clinical progression and regulates cell migration and invasion in PTC (29). Our previous study demonstrated that LINC02082 expression is elevated in human thyroid cancer, and it may play a critical role in thyroid carcinogenesis (21). Wang et al. showed that the lncRNA UNC5B-AS1 promotes proliferation, migration, and invasion in PTC cell lines (30). There are also several studies showed that other many lncRNAs, such as MALAT1, H19, BANCR, HOTAIR, GAS5, and PCA3, play an important role in regulation of different processes involved in the development and progression of various thyroid cancers, and suggested these lncRNAs could be used as novel biomarkers for early diagnosis or even treatment (17, 31, 32). However, these previous studies only analyzed in thyroid cancer cell lines or only compared thyroid cancer tissues and adjacent normal thyroid tissues from surgical specimen. In the present study, we additionally compared the lncRNA data of these tissues with those of benign nodules and applied to actual FNA samples to show that these lncRNAs have a real diagnostic value. Although further research is needed, these lncRNAs are expected to be helpful in the prognosis as well as diagnosis of PTC, based on currently available reports (19, 33).

To date, studies on lncRNAs in thyroid cancer have been limited compared with studies on protein-coding RNAs (34). LncRNAs, such as BANCR, are relatively well-known; however, most have been only discovered recently (35). The functions of lncRNAs are dictated by their secondary structures rather than their primary sequences, and their subcellular localization is critical for their function (36, 37). Researchers may need to perform gene regulation via CRISPR activation, CRISPR inhibition, or antisense LNA GapmeRs rather than via plasmid cDNA, siRNA, or shRNA to study lncRNA functions (38–41). Animal experiments can be challenging because lncRNAs exhibit evolutionarily poor sequence conservation across species (17). The fact that lncRNAs are the functional units themselves may be more meaningful for diagnosis than measuring mRNA levels (37). However, the measurement of lncRNA levels is difficult, as the expression of lncRNAs is generally considerably lower than that of mRNAs (17). In this study, several representative lncRNAs were measured using the laboratory-level real-time PCR method; however, if more lncRNAs are measured using a more sophisticated method, lncRNAs could be used as effective markers for advancing the molecular diagnosis of PTC.

This study has several limitations. First, the number of nodules included in the study was small, and there were not many indeterminate nodules that were difficult to diagnose. Second, only PTC was included in the study since it is the most common type of thyroid cancer, while the other types of thyroid cancer were not included. Third, in the preliminary FNA test, the diagnostic performance with the decreased lncRNA levels in PTC was lower than expected, and we did not discover additional candidate lncRNAs to improve the diagnostic accuracy. Fourth, false positive or false negative results might exist for the cytologically proven nodules in the FNA samples, as not all patients were surgically confirmed. Nevertheless, we showed that lncRNA expression analysis could be a promising approach for advancing the molecular diagnosis of PTC.

In summary, we selected candidate lncRNAs from TANRIC thyroid cancer dataset, confirmed the diagnostic potential of lncRNAs through our surgical specimen analysis, and verified the diagnostic value of lncRNA by using FNA samples at our institution. We showed that more accurate results could be obtained using a combination of specific lncRNAs. With further research on the identification of lncRNAs with additional diagnostic value and verification in large-scale studies of indeterminate nodules that require actual molecular diagnosis, the diagnostic method using lncRNAs could potentially supplement or replace FNA cytology and other existing molecular tests.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Severance Hospital (No. 4-2019-0335). The patients/participants provided their written informed consent to participate in this study.

DK: conceptualization, methodology, formal analysis, investigation, resources, data curation, writing - original draft, visualization, funding acquisition. JY: validation, formal analysis, investigation, data curation, writing - review and editing. JK: resources, writing - review and editing. Y-aH: resources, writing - review and editing. JKK: methodology, investigation, resources, writing - review and editing. CRK: methodology, resources, writing - review and editing. JHY: methodology, investigation, resources, writing - review and editing. JYK: methodology, investigation, resources, writing - review and editing. K-HN: conceptualization, methodology, investigation, resources, writing - review and editing, supervision, project administration, funding acquisition. EJL: conceptualization, writing - review and editing, supervision. All authors contributed to the article and approved the submitted version.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2021R1I1A1A01044993) and the Severance Hospital Research fund for clinical excellence (SHRC) (C-2019-0016).

The authors would like to thank Editage (www.editage.co.kr) for English language editing.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.