RASGEF1C methylation for the distinguishment and classification of benign and malignant thyroid tumors

Abstract

Background

The incidence of thyroid cancer (TC) has significantly increased, highlighting the need for effective and objective approaches for the early diagnosis of TC. This study aimed to explore RASGEF1C methylation as a biomarker for papillary thyroid cancer (PTC).

Methods

Formalin-fixed paraffin-embedded samples from a total of 363 PTC and 409 benign thyroid nodules from multiple centers were analyzed. RASGEF1C methylation profiles were examined via MALDI-TOFF mass spectrometry. Statistical analysis was performed via logistic regression adjusted for covariates, nonparametric tests, and receiver operating characteristic (ROC) analysis. Additionally, 40 follicular thyroid cancer samples, 45 medullary thyroid cancer samples, and 7 anaplastic thyroid samples from three hospitals were afterward collected to compare methylation patterns across subtypes.

Results

Hypomethylation of RASGEF1C in PTC was observed vs. BTN (all odds ratios (ORs) ≥ 1.57, p values < 0.001). Stratification analysis revealed a more pronounced association in younger patients, especially for BRAF V600E-positive PTC patients, than in individuals with benign tumors (all ORs ≥ 1.89, p values < 0.001). ROC analysis further demonstrated the outstanding diagnostic power of RASGEF1C hypomethylation for BRAF V600E-positive PTC cases (area under the curve (AUC) = 0.93), for cases < 55 years old (AUC = 0.88), and even for patients with a tumor length ≤ 1 cm (AUC = 0.83). Moreover, we observed the lowest RASGEF1C methylation level in anaplastic thyroid carcinoma, the most aggressive subtype of TC. Our results revealed similar RASGEF1C hypomethylation between chronic lymphocytic thyroiditis and papillary thyroid cancer, whereas RASGEF1C methylation in subacute thyroiditis patients was similar to that in the other benign subtypes.

Conclusion

Our study revealed RASGEF1C methylation as a promising biomarker for distinguishing and classifying benign and malignant thyroid tumors and even provided epigenetic evidence for the inflammatory-cancer transformation. Nevertheless, the limitation of tissue-based biomarkers should be well noted, and the development of more accessible biomarkers warrants further exploration in the future.

Introduction

Thyroid cancer (TC) is the most prevalent endocrine malignancy [1]. It has shown a notable increase in recent years, ranking among the most common cancers globally [2, 3]. China has the heaviest burden of TC worldwide [4], and TC is even the second most prevalent cancer among Chinese females [5]. Papillary thyroid carcinoma (PTC) accounts for more than 80% of TC cases, and follicular thyroid carcinoma (FTC), medullary thyroid carcinoma (MTC), and anaplastic thyroid carcinoma (ATC) are less prevalent subtypes of TC [1].

Several clinical diagnostic methods are currently employed. Neck ultrasonography supersonic is the primary imaging method in clinical practice. However, inconsistent reporting and diverse grading standards can lead to diagnostic discrepancies [6]. Fine-needle aspiration biopsy (FNAB) is another common practice and serves as an important means of further determining the benign or malignant nature of a nodule following ultrasonography [7]. Clinically, 15–30% of preoperative samples are classified as indeterminate (2023 Bethesda System for Reporting Thyroid Cytopathology [Third Edition], Categories III–V), with risk of malignancy spanning 22–74% [8, 9], which are often resorted to surgical removal. Excessive surgery in the indeterminate cases leads to various sequelae, whereas only 20% are finally determined as malignant [10,11,12]. The shortage of expert cytopathologists in China [13] (China [2.12/100,000], compared to America [3.94/100,000] [14, 15]) leads to overreliance on diagnostic lobectomy for indeterminate nodules—a procedure that overtreats 26–45% of benign/low-risk cases, despite postoperative histopathology providing definitive malignancy confirmation for resected tissues [16]. Furthermore, the diagnostic accuracy in cytopathological assessment of tissue malignancy is substantially influenced by the interobserver variability inherent in subjective morphological interpretation. Implementation of validated biomarkers could potentially mitigate this variability by providing objective diagnostic parameters, thereby not only alleviating diagnostic workloads but also enhancing the reliability and efficiency of cytopathological evaluations in clinical practice.

Certain molecular biology-assisted diagnostic approaches, such as detecting the BRAF V600E mutation, have demonstrated significant diagnostic efficacy, particularly in terms of specificity [17]. However, the sensitivity of the BRAF V600E mutation is only approximately 50% in distinguishing benign and malignant tumors in patients diagnosed with suspicious malignancies (Bethesda V) and only 30% in indeterminate nodules (Bethesda III–V) [18]. Somatic mutations have also been reported in less prevalent TC subtypes (RAS: mutated in 40–50% of FTCs; RET: mutated in approximately 50% of scattered MTCs; and TERT: mutated in approximately 40% of ATCs) [19,20,21]. Thus, the combination of multiple genetic mutations only slightly increases sensitivity while dramatically increasing costs. Moreover, certain benign thyroid tumors harbor cancer-associated genetic alterations as well. For examples, RAS mutations exist in around 26% follicular adenomas [22, 23]. Additionally, RET/PTC rearrangements have been reported in chronic lymphocytic thyroiditis [24]. Established commercial test kits such as Afirma and ThyroSeq v3 are limited by prohibitively high costs and the stability of RNA samples [25, 26]. Hence, there is an urgent need for a more feasible, objective, and accurate diagnostic approach for TC.

Numerous studies have confirmed that alterations in DNA methylation levels as early and concurrent events in cancer development [27]. The methylation analysis enables sensitive detection with trace DNA samples obtained from FNAB specimens (with as low as 1–10 ng DNA, approximately 150–1500 cells), thereby fulfilling analytical requirements for clinical utility [28], and is also feasible to detect DNA methylation in FFPE-fixed tissue samples [29]. Numerous studies have investigated DNA methylation in TC. However, most of these studies focused on the difference between malignant carcinoma and paracarcinoma normal tissues, and few studies focused on the difference between malignant and benign tumors. Although benign tumors share several characters as cancers, such as proliferation, angiogenesis, impaired growth signals, etc., benign tumors do not develop invasive and metastatic characteristics as malignant tumors. Therefore, understanding the epigenetic difference between malignant and benign tumors, which was reported as one of the driving forces of carcinogenesis in very recent studies, would be meaningful [30, 31]. Furthermore, differential diagnosis between benign and malignant tissues is crucial in clinical practice, with studies comparing carcinoma and paraneoplastic tissues offering limited significance. To date, many studies have addressed these issues, but most lack a sufficient number of subjects [32, 33].

In this study, we initially conducted a preliminary investigation using the Infinium MethylationEPIC array (850 K) platform on 32 specimens (unpublished data). Candidate genes for experimental validation were selected followed a dual rationale: primary evidence from the 850 K methylation profiling coupled with functional genomic considerations. Priority was placed on identifying novel loci with functional relevance to thyroid carcinogenesis and absence of prior documentation regarding methylation associations with PTC in existing literatures. The RASGEF1C (RasGEF Domain Family, Member 1C) gene, located on chromosome 5q35.3, is expressed primarily in the cerebral and cerebellar hemispheres [34]. Members of the RASGEF family catalyze the conversion of RAS-GDP, the inactivated state, to RAS-GTP, the activated state, thereby activating the RAS signaling pathway [35, 36]. RASGEF1A and RASGEF1B have been well reported in multiple cancers [37,38,39]. Given the shared conserved domain in RASGEF family, RASGEF1C may influence biological processes such as cell growth, differentiation, and migration [40]. So far, studies focusing on the RASGEF1C gene are mainly in noncancer areas [41, 42], with only one study examining DNA methylation and aging [43]. In this study, we investigate RASGEF1C methylation for differentiation benign and malignant thyroid tumors with a large study cohort.

Methods

Study design and patients

From 2017 to 2023, a total of 409 benign thyroid nodules (BTN) FFPE samples (205 adenomas, 187 goiters, 3 subacute thyroiditis, and 14 chronic lymphocytic thyroiditis) and 363 PTC FFPE samples were collected from two hospitals (Huai’an Hospital of Xuzhou Medical and University and the Affiliated Hospital of Nantong University). The PTC patients did not receive any cancer-related treatments before surgery. The BTN patients were included on the basis of age, sex, and years of diagnosis and were matched with the PTC patients. To evaluate RASGEF1C methylation specificity across thyroid malignancies, we additionally involved Jiangsu Province Hospital of Chinese Medicine and expand the collection from 2015 to 2023, finally obtained 40 FTC, 45 MTC, and 7 ATC surgical FFPE samples. None of the patients received any cancer-related treatments before surgery. Histopathological diagnosis was conducted by two qualified pathologists for all the subjects. The stage of each malignant case was determined according to the 8th edition of the American Joint Committee on Cancer Staging System (AJCC8) [44].

This study received approval from the ethics committee of each center. All participants provided informed consent.

DNA extraction and bisulfite conversion

After surgery, both benign and malignant tissues were fixed in formalin, and paraffin-embedding (FFPE) technology was used for preservation. Genomic DNA extraction from FFPE samples was carried out via the Tissue DNA Isolation Kit (DC105, Vazyme, Nanjing, China), followed by bisulfite conversion via the Sulfurization Kit (BT201, Tantica, Nanjing, China). All the TC and BTN samples were processed in parallel for DNA extraction and bisulfite conversion.

MALDI-TOF mass spectrometry

The CpG site cg26704844 (GRCh38/hg38: chr5:180,135,325–180,135,326) was selected based on its annotation in the UCSC Genome Browser (University of California, Santa Cruz, https://genome.ucsc.edu), location at the S_shore of a CpG island, and presence within a proximal enhancer-like region (ENCODE cCREs database). This site spans intronic regions and is hypothesized to regulate RASGEF1C expression via chromatin looping or transcription factor recruitment (Fig. 1A).

Fig. 1

The schematic diagram, sequence, and the mass peaks of RASGEF1C amplicon. A The location of the investigated 228 bp amplicon in RASGEF1C gene (from the UCSC Genome Browser, GRCh38/hg38). B The sequence of the RASGEF1C primers and amplicon (chr5: 180,135,285–180135512, GRCh38/hg38, defined by the UCSC Genome Browser). Uppercase letters indicate the sequence-specific primer regions, and the lowercase indicated the tags for PCR. The measurable CpG sites in the amplicon were underlined. C The MassARRAY assay yielded four distinguishable peaks. CpG_3, CpG_4, and CpG_5 generated a single peak each, CpG_1 and CpG_2 were in the same digested fragment and thus presented as a combined peak. The downward red arrow pointing at the signal peak indicates the non-methylated CpG site, and the blue line above the signal peak indicates the methylated CpG site. The mass is represented at the bottom

DNA methylation was quantified using MALDI-TOF MS (Agena MassARRAY). Bisulfite-converted DNA was amplified via primers lacking SNPs/CpGs (designed by Epidesigner, Agena), generating a 228-bp amplicon (chr5:180,135,285–180,135,512) that encompassed 5 CpG sites (Fig. 1B). The PCR products were then treated with shrimp alkaline phosphatase (SAP) and subjected to a T-cleavage assay according to standard instructions (Agena Bioscience, San Diego, California, United States). After resin cleaning, the final products were dispensed onto a 384 SpectroCHIP using a nanodispenser. The levels of DNA methylation were determined semiquantitatively via Agena matrix-assisted laser desorption/ionization (MALDI)-time-of-flight (TOF) mass spectrometry (Agena Bioscience, San Diego, California, United States). The chips were read by a MASSARRAY system. The quantitative methylation levels for each CpG site were obtained via SpectroACQUIRE v3.3.1.3 and visualized via EpiTyper v1.3 software. All TC cases and controls were processed in parallel, with equal sample numbers per MassARRAY chip to minimize batch effects. In this method, the conversion of unmethylated cytosine to uracil during bisulfite treatment generated base-specific cleavage products that reflected underlying methylation patterns. Methylation status of CpG sites was determined by MALDI-TOF MS through mass signal ratios of cleavage fragments. The amplicon yielded measurable data for 5 CpG sites, including cg26704844 as mentioned above (CpG_5), with 4 distinguishable mass peaks. CpG_1.2 (spanning cg16931200 and an adjacent CpG) was analyzed as a merged peak, with methylation levels averaged (Fig. 1C). The relative methylation level was quantified as the β value, derived from the ratio of methylated (M) to total (methylated + unmethylated + 100) allele intensities: β = M / (M + U + 100), where M and U represent the peak areas of methylated and unmethylated alleles measured by MALDI-TOF MS. The constant 100 was introduced to suppress background noise [45]. Hypomethylation or hypermethylation was defined by β values significantly lower or higher than those in benign controls, as determined by nonparametric tests with p < 0.05.

Immunohistochemistry

Formalin-fixed, paraffin-embedded (FFPE) thyroid tissues were deparaffinized in xylene, rehydrated through graded ethanol, and subjected to heat-induced antigen retrieval in sodium citrate buffer (pH 6.0, 95 °C, 20 min). Endogenous peroxidases were blocked with 3% H, followed by 5% BSA blocking. Sections were incubated with primary antibody against RASGEF1C (Abmart TD9809; 1:50, 4 °C overnight), then with HRP-conjugated secondary antibody (Agilent K5007, RT, 1 h). DAB chromogen (Dako K3468) was used for signal development, with hematoxylin counterstaining. Slides were dehydrated and mounted for analysis. Quantification: DAB-positive area (%) was calculated using ImageJ (v1.53) via color deconvolution (DAB/hematoxylin separation) and threshold-based segmentation:

Positive area (%) = (DAB-positive pixels / Total tissue pixels) × 100.

Statistical analyses

Continuous variables are represented as medians with interquartile ranges (IQRs) [M (Q1, Q2)]. All data processing was conducted via SPSS 29.0 software and GraphPad prism 9.5 software. The Chi-square test was used to assess categorical variables. Nonparametric tests, including the Mann‒Whitney test and Kruskal‒Wallis test, were used to compare the differences between different groups. Logistic regression models were applied to assess the association between DNA methylation and PTC, and the results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Receiver operating characteristic (ROC) curves were used to evaluate the efficacy of DNA methylation in each subgroup for PTC diagnosis. We conducted ROC curve analysis via multivariable logistic regression models adjusted for age and sex. Three models were constructed to predict malignancy: Methylation model: Logit (P) = β0 + β1 (CpG_1.2) + β2 (CpG_3) + β3 (CpG_4) + β4 (CpG_5) + β5 (Age) + β6 (Sex); BRAF model: Logit (P) = β0 + β1 (BRAF) + β2 (Age) + β3 (Sex); Combined model: Logit (P) = β0 + β1 (CpG_1.2) + β2 (CpG_3) + β3 (CpG_4) + β4 (CpG_5) + β5 (BRAF) + β6 (Age) + β7 (Sex); Predicted value was obtained by logistic regression model as the probability of malignancy. Probability thresholds were determined separately for each subgroup by maximizing the Youden index (J = sensitivity + specificity − 1) from their respective ROC curves. All analyses were two-sided, and statistical significance was defined as a p value < 0.05 or a p value < 0.0125 (0.05/4 adjusted by Bonferroni).

Results

Significant hypomethylation of RASGEF1C in PTC

The association between RASGEF1C hypomethylation and PTC was investigated with a total of 363 PTC patients (median age: 50 years) and 409 BTN subjects (median age: 51 years). The BTN group comprised 22.00% males and 78.00% females, with a sex composition similar to that of the PTC group (23.14% males, 76.86% females, p value = 0.771, Chi-square test). The detailed clinical characteristics are presented in Table 1. Methylation levels of the five CpG sites were lower in PTC than in BTN (median of methylation in PTC vs. BTN, CpG_1.2: 0.43 vs. 0.64; CpG_3: 0.36 vs. 0.68; CpG_4: 0.37 vs. 0.65; and CpG_5: 0.35 vs. 0.60; Fig. 2A and Additional file 1). Logistic regression adjusted for age and sex revealed that each 10% decrease in methylation conferred an increased risk of PTC (ORs per -10% methylation: CpG_1.2: 1.93; CpG_3: 1.99; CpG_4: 1.96; and CpG_5: 1.57; all p values < 0.001 by Bonferroni adjustment; Additional file 1).

Table 1 Clinical characteristics of subjects

Fig. 2

RASGEF1C hypomethylation in PTC cases compared with BTN subjects. Box plots show the distribution of methylation levels in case and control groups. Results expressed as a combination of all subjects (A), stratified by age 55 years (B–C), and stratified by sex (D–E), respectively. p values were calculated by logistic regression with adjustment for age and sex (in sex study with adjustment for age only), and significant p values were indicated with *: < 0.05, **: < 0.01, and ***: < 0.001. BTN, benign thyroid nodule and PTC, Papillary Thyroid Cancer

Associations between RASGEF1C hypomethylation and PTC stratified by age and sex

Age and sex were used as stratification criteria to comprehensively understand the relationship between RASGEF1C methylation and PTC. Stratifying by the AJCC8 age cutoff (55 years), younger patients (< 55 years) showed stronger associations between RASGEF1C hypomethylation and PTC (ORs per -10% methylation: CpG_1.2: 2.16; CpG_3: 2.21; CpG_4: 2.01; and CpG_5: 1.58; p < 0.001 by Bonferroni adjustment; Fig. 2B) compared to older patients (≥ 55 years; ORs = 1.55–1.86, p < 0.001 by Bonferroni adjustment; Fig. 2C). Males exhibited higher ORs (ORs per − 10% methylation: CpG_1.2: 2.08; CpG_3: 2.32; CpG_4: 1.86; and CpG_5: 1.60; Fig. 2E) than females (ORs = 1.55–1.98; Fig. 2D). Details see Additional file 2.

Association between RASGEF1C hypomethylation and PTC stratified by the BRAF V600E mutation status

We further investigated the clinical relevance of RASGEF1C hypomethylation in PTC progression. Tumors exceeding 1 cm in length exhibited significantly reduced methylation at CpG_3 (P = 0.003), CpG_4 (P = 0.001), and CpG_5 (P = 0.003) compared to smaller lesions (≤ 1 cm). Similarly, advanced T-stage (T2–4) tumors showed lower methylation levels at CpG_1.2 (P = 0.026), CpG_3 (p < 0.001), CpG_4 (P < 0.001), and CpG_5 (p < 0.001) relative to T1 tumors. Lymph node metastasis (pN1) was associated with pronounced hypomethylation at CpG_3 (P = 0.001), CpG_4 (P = 0.003), and CpG_5 (P = 0.017), suggesting a potential role of RASGEF1C epigenetic dysregulation in metastatic progression (Additional file 3).

Strikingly, BRAF V600E mutation status profoundly influenced methylation patterns. BRAF-positive PTCs demonstrated markedly lower methylation across all CpG sites compared to BRAF-negative cases (p < 0.001 for all sites; Fig. 3 and Additional file 4). Compared to the BTN, the logistic regression results suggested that the interaction with RASGEF1C hypomethylation had approximately double the risk of developing cancer in those with BRAF V600E mutation than the ones without (CpG_1.2: OR = 2.60 vs. 1.55; CpG_3: OR = 2.75 vs. 1.62; CpG_4: OR = 2.40 vs. 1.62; CpG_5: OR = 1.89 vs. 1.32; all p values < 0.001; Fig. 3 & Additional file 4). Despite this divergence, both BRAF-positive and -negative PTC cohorts remained significantly hypomethylated versus BTN controls (p < 0.001, Kruskal–Wallis with Dunn’s post hoc test; Additional file 5).

Fig. 3

RASGEF1C hypomethylation levels in papillary thyroid cancer (PTC) with/without BRAF V600E mutation vs. benign thyroid nodule (BTN) controls. P values were calculated by logistic regression adjusted for age and sex. BTN, benign thyroid nodule and PTC, Papillary Thyroid Cancer

The clinical value of RASGEF1C hypomethylation in distinguishing PTC from BTN

RASGEF1C hypomethylation robustly differentiated PTC from BTNs, achieving an area under the curve (AUC) of 0.86 (95% CI: 0.83–0.88; Fig. 4A). This performance significantly surpassed BRAF V600E mutation testing alone (AUC = 0.78; 95% CI: 0.74–0.81; Fig. 4A). For micropapillary carcinomas (tumor length ≤ 1 cm), methylation retained strong discrimination (AUC = 0.83) Fig. 4B). Age stratification revealed enhanced performance in patients < 55 years (AUC = 0.88; Fig. 4C). Integrating RASGEF1C methylation with BRAF V600E mutation further enhanced diagnostic efficacy across all groups (Total cohort: AUC = 0.89; Micropapillary carcinomas: AUC = 0.86; Age < 55 years: AUC = 0.91; Age ≥ 55 years: AUC = 0.86; Fig. 4A–4D). Notably, BRAF V600E-positive PTCs achieved near-perfect separation from BTNs (AUC = 0.93; Fig. 4F), whereas BRAF-negative cases still demonstrated moderate discrimination (AUC = 0.78; Fig. 4E).

Fig. 4

Diagnostic value of RASGEF1C methylation in distinguishing PTC from BTN patients. The methylation levels of five CpG sites within the RASGEF1C gene were utilized to generate a prediction probability via logistic regression with covariates adjusted. Receiver operating characteristics (ROC) curves were conducted to assess the discriminatory power of RASGEF1C methylation, the BRAF V600E mutation, or combined the above two biomarkers together for distinguishing PTC cases from BTN subjects containing all participates (A), tumor length ≤ 1 cm (B), and stratified by the age of 55 years (C–D), or the BRAF V600E status (E–F), respectively. BTN, benign thyroid nodule; PTC, Papillary Thyroid Cancer; and AUC, area under the curve

Expression characteristics and methylation correlation analysis of RASGEF1C

To investigate the expression and regulation of RASGEF1C in thyroid cancer, we first analyzed TCGA data, which showed higher (though not statistically significant, P = 0.227) mRNA expression of RASGEF1C in thyroid tumor tissues (N = 510) compared to normal tissues (N = 58) (Fig. 5A). Immunohistochemical (IHC) imaging (Fig. 5B) and quantitative analysis (Fig. 5C) of our in-house samples revealed significantly elevated RASGEF1C expression in PTC versus BTN (p < 0.001). Methylation correlation analysis across distinct CpG sites (Fig. 5D) indicated varying degrees of negative associations with RASGEF1C expression, suggesting that DNA methylation may modulate its expression. Additionally, leveraging TCGA data, we found a statistically significant positive correlation between the mRNA expression of RASGEF1C and the downstream oncogene HRAS (r = 0.47, p < 0.001), BRAF (r = 0.31, P = 0.007), MEK (r = 0.57, p < 0.001), and ERK1 (r = 0.50, p < 0.001) (Additional file 7). These correlations imply that the expression of RASGEF1C may activate MAPK-related signaling pathways.

Fig. 5

Expression, immunohistochemical analysis, and correlation between expression and methylation of RASGEF1C. A Boxplot displaying the normalized read count of RASGEF1C in normal tissues (N = 58) and tumor tissues (N = 510) derived from TCGA data, with a reported P of 0.227. B Immunohistochemical images of BTN (upper) and PTC (lower) tissues. C Bar graph presenting the quantitative analysis of the relative expression level of RASGEF1C in BTN and PTC tissues. *** p < 0.001 by T-test. D Scatter plots illustrating the correlation between the relative expression levels of RASGEF1C and methylation levels at distinct CpG sites. BTN, benign thyroid nodule and PTC, Papillary Thyroid Cancer

Methylation differences among subtypes

We compared RASGEF1C methylation patterns among eight histologically confirmed thyroid subtypes: adenoma, goiter, subacute thyroiditis (SAT), chronic lymphocytic thyroiditis (CLT), PTC, FTC, MTC, and ATC. The clinical features see Additional file 6. While FTC and MTC showed divergent methylation patterns, the most notable finding was CLT exhibiting hypomethylation levels nearly identical to PTC (adjusted p > 0.05 vs. PTC), distinct from benign adenomas and goiters (both hypermethylated vs. CLT, adjusted p < 0.05, Fig. 6), distinct from benign adenomas and goiters (both hypermethylated vs. CLT, adjusted p < 0.05, Fig. 6). Notably, among malignant subtypes, ATC—the most aggressive TC variant—showed the lowest methylation at CpG_3, CpG_4, and CpG_5 (Fig. 6), though limited sample size (n = 7) precluded definitive conclusions. FTC and MTC displayed CpG-specific differences: While both were hypermethylated at CpG_1.2, CpG_3, and CpG_4 compared to PTC (adjusted p < 0.001), FTC exhibited moderate hypomethylation at CpG_5 versus goiter (adjusted P = 0.020), and MTC showed lower methylation than benign nodules (adjusted p < 0.001), highlighting subtype-specific epigenetic landscapes.

Fig. 6

Differences in RASGEF1C methylation levels among benign and malignant thyroid nodules subtypes. A Boxplot shows the methylation levels of the CpG sites in benign and malignant thyroid nodules subtypes. B–E RASGEF1C methylation differences among different benign and malignant thyroid nodule subtypes. The numbers in the figure indicate p values from Kruskal–Wallis with Dunn’s post hoc test. Significant p values were shaded in gray. B CpG_1.2; C CpG_3; D CpG_4; and E CpG_5. SAT, Subacute Thyroiditis; CLT, Chronic Lymphocytic Thyroiditis; PTC, Papillary Thyroid Carcinoma; FTC, Follicular Thyroid Carcinoma; MTC, Medullary Thyroid Carcinoma; and ATC, Anaplastic Thyroid Carcinoma

Discussion

The accurate identification of thyroid nodules is crucial for preventing overtreatment and managing tumor metastasis. Compared with clinicopathological markers, molecular markers offer objectivity and reliability for auxiliary diagnosis [29, 46]. The utilization of DNA methylation, with its microsampling ability, stability, cost-effectiveness, and prioritization over cancer symptoms, can be an effective complement to FNAB for the diagnosis of TC. This study demonstrated that RASGEF1C exhibited differential methylation patterns between BTN and PTC subjects. These findings provide new evidence for the potential clinical application of methylated genes.

RASGEF1C has been identified as a specific activator of RAP2, a member of the RAS family of GTPases [47, 48]. Regulated by a set of guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs), RAP2 functions as a slow-response molecular switch in the RAP1 signaling cascade [49] and is involved in the migration, invasion, and poor prognosis of multiple cancers [39, 50,51,52]. RAP2 is also involved in the Wnt signaling pathway [53, 54] and maintains the homeostasis of the Wnt signaling pathway by regulating the LRP6 receptor [55]. Activation of this signaling pathway triggers β-catenin translocation into the nucleus, thereby promoting TC cell proliferation [56]. Our results indicated that the hypomethylation of the enhancer element of RASGEF1C, the regions have been linked to transcriptional repression in TC [57], may modulate its expression and further regulate the signal transduction of the RAP2 pathway, which is subsequently involved in the initiation and progression of PTC. Our study is limited by functional exploration, and the detailed mechanism requires further investigation in the future.

TC incidence in China significantly increases with age, peaking at 55 years, and is three times more common in women than in men [5, 58]. Differences in physiological structure, hormone levels, and lifestyle habits influenced by age and sex can lead to distinct disease spectra. Estrogen may stimulate the generation of mutagenic molecules in thyroid cells and promote the proliferation and invasion of tumor cells [59]. Recent studies have highlighted the mechanistic role of DNA methylation in aging, revealing distinct methylation patterns between younger and older individuals [60, 61] as well as between sexes [62]. Our study revealed a more pronounced association between RASGEF1C methylation and PTC in males or younger individuals, suggesting the potential roles of age and sex in modulating the methylation status of the RASGEF1C gene. These findings underscore the importance of considering age and sex when interpreting the molecular mechanisms underlying TC.

RASGEF1C is an activator of RAP2, and activated RAP2 binds to BRAF to form the RapGTP-B-Raf complex, which phosphorylates MEK, sustains ERK activation, and is considered a major ERK activator [63, 64]. BRAF plays a pivotal role in the MAPK pathway, and its mutation is relatively common in PTC [65]. The most prevalent BRAF mutation is V600E, which leads to the activation of the BRAF protein and results in a relatively poorer prognosis, higher mortality rates, and increased recurrence rates in carriers [66,67,68]. As shown in Fig. 2, RASGEF1C exhibited much lower methylation levels in BRAF V600E-mutated PTC patients than in those without the BRAF V600E mutation. Considering the poorer prognosis of BRAF V600E carriers, our results suggest that ultrahypomethylation of RASGEF1C may even imply a worse prognosis for PTC patients, via the activation of MAPK pathways as presented in Addition file 7. Very recent studies have reported that epigenetic changes could be one of the driving forces of cancer initiation and progression [30, 31], and our findings suggest interactions between genetic and epigenetic alterations. Advanced TC, characterized by larger tumor length, lymph node involvement, and BRAF V600E mutation, often implies a poorer prognosis [66]. Since approximately 99% of the PTC patients in our study were at the early stage (Stages I and II), we were unable to evaluate the correlation between RASGEF1C methylation and stage in our study. Nevertheless, RASGEF1C methylation was lower in patients with a tumor length > 1 cm, larger tumor size and involved lymph nodes than in those with small tumors and no involved lymph nodes, indicating the involvement of hypomethylation of RASGEF1C in the progression of cancer (Additional file 3). It would be very meaningful to determine the causative or consequent effects of aberrant DNA methylation on genetic changes in the future through functional studies, and the mechanism of RASGEF1C hypomethylation at the enhancer region and TC initiation and progression is also worth exploring.

Moreover, although the BRAF V600E mutation has been widely used as a biomarker to distinguish benign and malignant thyroid tumors, its sensitivity is limited owing to the approximately 50% mutation rate in PTC [18]. In this study, as shown in Fig. 4A, ROC curves for RASGEF1C methylation demonstrated robust diagnostic efficiency (AUC = 0.86), highlighting the potential of DNA methylation as a promising diagnostic biomarker for PTC, whereas the AUC of BRAF V600E was only 0.78. The combination of RASGEF1C methylation and BRAF V600E could even reach an outstanding AUC of 0.89, not only exhibiting outstanding performance for the diagnosis of BRAF V600E-positive PTC cases (AUC = 0.93), but RASGEF1C methylation also performed sufficiently well in diagnosing BRAF V600E-negative PTC cases (AUC = 0.78), suggesting the superiority of RASGEF1C as a diagnostic marker for PTC. Research efforts incorporating larger sample sizes are essential for comprehensively assessing the clinical significance of RASGEF1C methylation in distinguishing various subtypes and assessing the aggressiveness of PTC. Furthermore, there is a pressing need for in-depth investigations into the functional aspects and underlying molecular mechanisms associated with RASGEF1C hypomethylation in the malignant progression of PTC. Given the limitations of FNAB in indeterminate cytology (Bethesda III/IV, malignancy risk 22–74%), RASGEF1C methylation offers a critical molecular adjunct to refine risk stratification, particularly in cases where BRAF testing alone is insufficient due to its limited sensitivity in non-BRAF-mutant PTCs. Our data show that even in the BRAF-negative tumors, RASGEF1C hypomethylation retains diagnostic utility (AUC = 0.78), and combining both markers achieves an AUC of 0.89—potentially reducing unnecessary surgeries in up to 45% of benign cases currently overtreated under standard guidelines. This is particularly impactful in resource-limited settings or regions with cytopathologist shortages, where molecular profiling can supplement subjective cytology interpretations.

Chronic lymphocytic thyroiditis (CLT, also known as Hashimoto’s thyroiditis) is one of the most prevalent autoimmune disorders and is characterized by the infiltration of T and B cells and subsequent damage to the thyroid gland, ultimately leading to primary hypothyroidism [69]. Recent studies and meta-analyses have validated that CLT patients are more likely to develop PTC with a risk rate (RR) of 1.4–2.1 compared with those who did not have the disease [70, 71]. Furthermore, studies have indicated that incidentally discovered papillary thyroid microcarcinomas (PTMCs) are more frequently found in patients with CLT than in those with goiter [71]. Additionally, CLT is common in patients with PTMC [72]. If a high-risk CLT (the CLT with the potential of malignant transformation) could be identified, intervention for TC at the preclinical stage or even prevention of TC may be possible. As previously mentioned, methylation often precedes the development of tumor symptoms [27]. Malignant transformation in the thyroid gland may be caused by cellular mediators produced by immune cells in states of chronic inflammation [73, 74]. This transformation has also been linked to epigenetic influences [69, 75]. Notably, our study identified a similar epigenetic pattern between CLT and PTC. It is critical to note that our cohort predominantly comprised surgically resected CLT cases due to suspicious ultrasound features (e.g., microcalcifications, irregular margins, or taller-than-wide shape) or indeterminate fine-needle aspiration cytology (Bethesda III–V categories), potentially containing the PTC microfoci, and thus, overestimating the methylation overlap in the general CLT population (409 (3.4% CLT) in our BTN samples vs. 5%–15% CLT in benign thyroid tumors in general [76, 77]). Given that surgical intervention is generally not recommended clinically in low-risk CLT cases without compression symptoms, we postulate that RASGEF1C hypomethylation might specifically characterize high-risk CLT subgroups requiring surveillance, rather than representing the general CLT population where methylation levels could approximate those of benign thyroid nodules. This epigenetic signature may, therefore, serve dual clinical purposes: (1) as a dynamic monitoring biomarker for malignant transformation risk stratification in high-risk CLT cases, enabling early intervention before histopathological PTC manifestation and (2) as a molecular adjunct to current diagnostic modalities. Integration with single-cell sequencing could further refine detection sensitivity for pre-malignant clones. However, validation of this hypothesis remains methodologically challenging due to the ethical and practical constraints of obtaining serial surgical specimens from non-operated CLT patients. Future longitudinal studies employing pre-surgical samples such as FNAB or liquid biopsy-based methylation analysis may circumvent this specimen acquisition barrier while enabling prospective validation.

A critical limitation of this study is the reliance on postoperative FFPE tissues, which provide definitive histopathological diagnosis but lack the preoperative context of FNAB specimens—characterized by smaller cell yields, potential sampling bias, and variable DNA integrity, that are critical for biomarker discovery and development. To circumvent challenges in prospectively obtaining preoperative FNAB samples with surgical correlation, we first analyzed postoperative FFPE tissues, which could provide unambiguous malignant/benign classification and also sufficient DNA materials for the biomarker development and methylation analyses. Future studies will focus on translating these findings to preoperative FNAB specimens to address the unmet need for accurate risk stratification in indeterminate cytology cases.

Conclusion

Our study establishes RASGEF1C hypomethylation as a robust epigenetic marker for distinguishing PTC from benign nodules, with significant associations in younger patients and males. RASGEF1C methylation demonstrated superior diagnostic performance to BRAF V600E testing alone and enhanced accuracy when combined with BRAF analysis, a synergy that suggests epigenetic–genetic interactions in PTC pathogenesis. The correlation with adverse clinical features highlights its potential to refine risk stratification for aggressive phenotypes. Our observation of shared hypomethylation between high-risk CLT and PTC also provided further evidence to support the inflammation-mediated carcinogenesis [69, 73,74,75]. These discoveries highlight RASGEF1C as a promising biomarker, though validation in postoperative samples and clarification of its causal role in carcinogenesis. Longitudinal studies and functional explorations are needed to translate these findings into clinical practice.