Cyto-Histological Profile of microRNAs as Diagnostic Biomarkers in Differenciated Thyroid Carcinomas

1. Introduction

The incidence of DTC, with over 95% of them diagnosed as papillary thyroid carcinoma (PTC) [1,2], has increased in the last years, mainly attributed to enhanced sensitivity of complementary diagnostic techniques and the influence of environmental factors, but not accompanied by a corresponding change in mortality rate [3,4]. Despite the favorable prognosis of PTCs, it was advanced that some clinicopathological features and the presence of some genetic alterations can influence their progression [5]. An accurate diagnosis in a pre-surgical phase is crucial to optimize medical or surgical treatment, and to avoid unnecessary surgeries and complications [6].

Ultrasound-guided fine needle aspiration cytology (US-FNAC) is the gold standard method in the diagnosis of nodular thyroid disease, but up to 30% of nodules remain without a definitive diagnosis [7]. Bethesda II and VI cytological results have excellent performance to assess the benign or malignant nature of a thyroid nodule, respectively. Diagnostic limitations are particularly significant in indeterminate Bethesda categories: III – atypia of undeterminate significance (AUS), and IV – follicular neoplasm. Each category harbors a different risk of malignancy (ROM) and a different option for treatment and follow-up [8,9].

Although the American and European Thyroid Association [10,11] consider the inclusion of molecular tests in US-FNAC, no consensus is yet achieved on the best molecular biomarkers panel to use in pre-surgical diagnosis and prognosis of thyroid cancer [12,13].

A promising area of study in thyroid cancer (TC) is the assessment of microRNAs (miRNAs) expression and its contribution for diagnosis and prognosis of thyroid nodules [14,15,16]. They are stable molecules that can be detected in cytology and histology samples [17,18], in small amounts of fresh or stored material, and can also be studied in the blood [19,20].

MiRNAs are an important class and numerous, of small (with a length of approximately 22 nucleotides) non-coding RNAs. MiRNAs are supposed to modulate 30% of the human genome and play an important role as global regulators of gene expression [21,22]. They have a key role in posttranscriptional gene expression, by reducing the stability of the target messenger RNA (mRNA) and/or repressing its translation. These events may have influence in intracellular regulatory processes implicated in carcinogenesis, including differentiation, proliferation, and apoptosis [23,24].

Studies of the miRNA expression profile in human tumors allowed the identification of specific "signatures" associated with diagnosis, staging, prognosis, and response to treatment [25,26]. Unraveling their potential poses a substantial scientific challenge, underscoring the depth and importance of this tool.

When TC grows due to the action of oncogenes, a series of alterations on miRNAs biogenesis is revealed. Altered expression of miRNAs occurs, contributing to a constitutive activation of MAPK pathway [27]. MiRNAs can act as tumor promoters (oncomiRs) which downregulates tumor suppressor genes or act as tumor suppressors leading to upregulation of oncogenes with consequences in tumor progression and growth [28].

In 2008 Nikiforova et al. [29] described the biological importance and diagnostic value of miRNAs analysis in thyroid tumors, estimating that in thyroid tumors 32% of miRNAs are upregulated while 38% are downregulated. When compared with thyroid normal tissue, specific up and downregulated miRNAs arose as involved in the progression of PTC such as miR-146b, miR-221, miR-222 (upregulated) and miR-15a (downregulated) (29). The miRNA expression is tissue specific, varying among the different types of TC. Upregulated miR-146b, miR-221 and miR-222, are shown to downregulate proto-oncogene receptor tyrosine kinase (KIT), which is involved in cell differentiation and growth, while miR-15a downregulates B-cell lymphoma 2 (BCL-2) and controls the proliferation and apoptosis of PTC through AKT pathway [29].

Castagna, et al. [30] have shown, in a series of fine-needle aspiration samples, that miR-146b, miR-221 and miR-222 are overexpressed in malignant or suspicious for malignancy nodules confirming the utility of these miRNAs in the classification of thyroid nodules. The overexpression of these miRNA was correlated with tumor aggressiveness, extra thyroidal extension, recurrence and metastases, representing a signature of PTC [31]. Therefore, the study of miRNA expression can be useful in the early and accurate diagnosis, as well as in the evaluation of biological behavior of PTC [32,33,34].

Although the identification of miRNAs expression profiles in thyroid cancer is being revealed, more studies are needed to establish its diagnostic, prognostic and predictive role in clinical practice. The identification of miRNAs and their target genes can allows us to envisage new and more personalized therapies [35,36].

This study aims to evaluate the expression of selected miRNAs (miR146b, miR221, miR222 and miR15a) in thyroid nodules, their association with the presence of clinicopathologic features and genetic alterations (TERTp, BRAF and RAS mutations). As an ultimate goal we want to evaluate the reliability of molecular tests on US-FNAC as a pre-surgical diagnostic tool in the improvement of DTC diagnosis.

2. Materials and Methods

2.1. Study Design

This retrospective study was conducted in a series of patients with thyroid nodular disease suspicious of malignancy, which underwent surgery, in a single non-oncologic hospital, between 2013 and 2020. Criteria for suspicion of malignancy in thyroid nodules were based in clinical evaluation, image features and Bethesda classification for cytology [7], in accordance with 2015 ATA [10] and 2017 ETA guidelines [37]. Surgical protocols and recommendations for the extent of surgical treatment followed the same guidelines [10,37]. A series of 13 histological benign lesions were also included for comparative purposes.

Study inclusion criteria: patients diagnosed, after surgery, with differentiated thyroid carcinoma (cases) or benign thyroid nodules (controls), of both genders, over 18 years old, with available representative histological and cytological samples.

Study exclusion criteria: patients with toxic goiter, papillary carcinoma smaller than 1 cm, and patients with other than thyroid neoplasm at the time of thyroid surgery.

This series includes 176 samples (98 cytology samples and 78 formalin-fixed paraffin embedded (FFPE) tissues) obtained from 106 patients-submitted to surgery. In 70 patients, paired cytology and histology samples were analyzed. Epidemiological and clinicopathological features of patients and tumors were gathered based on the information available in the clinical and histopathological reports from the reference hospital.

All the samples were reviewed by two independent pathologists based on the 4th edition of the World Health Organization (WHO) classification of tumors of endocrine organs [38].

2.2. MicroRNAs Expression Profile Analysis

2.2.1. MicroRNA Extraction from FFPE Tissues and FNAC Samples

The initial step is different for FFPE tissues and FNAC samples. For FFPE tissues, a slide with a 10 μm cut was deparaffinized and the tumor area was manually micro dissected. For FNAC samples, the entire smears were manually scraped. MicroRNA extraction was performed with the miRNeasy Mini kit (1038703 Qiagen, Germany) following the manufacturer’s instructions.

RNA concentration was determined using the NanoDrop (ND-1000, Thermo Fisher Scientific, Lithuania) spectrophotometer. RNA handling was always performed on ice.

2.2.2. Reverse Transcription and Quantitative Real Time PCR (RT-qPCR)

Complementary deoxyribonucleic acid (cDNA) synthesis was achieved with TaqMan® Advanced miRNA cDNA Synthesis Kit (refª A28007 Applied biosystems, USA) following the manufacturer’s instructions.

RT-qPCR was performed to evaluate the relative expression levels of miR-146b, miR-221, miR-222 and miR-15a (refª A25576, ThermoScientific, USA) [39]. MiR-16 (refªA25576, ThermoScientific, USA) was used as an endogenous control for normalization of the expression values [40]. The relative quantification of gene expression was attained using the 2-ΔΔCT method [41].

RT-qPCR was performed using TaqMan® Fast Advanced (refª 4444557, ThermoScientific, USA) system according to the manufacturer´s instruction on a QuantStudio 5 Real-Time PCR System (ThermoScientific, USA). Duplicates were done for all samples.

2.3. Genetic Analysis

2.3.1. DNA Extraction

Genomic DNA was extracted using the GRS Genomic DNA BroadRange Kit (refª GK06.0100 GRiSP Research Solutions, Portugal) or the QIAmp® DNA Investigator Kit (refª 56504 Qiagen, Germany), for FFPE and US-FNACs, respectively, according to the manufacturer’s instructions. NanoDropTM One UV185 Vis Spectrophotometer (Thermo Fisher Scientific Inc) was used for quantification of isolated DNA.

2.3.2. Mutational Screening

For genetic characterization of the lesions PCR (SimpliAmp TM Thermalcycler, Applied Biosystems, USA) and Sanger sequencing (Applied Biosystems 3130/3130XL Genetic Analyzer, USA) were used. Mutations in TERTp, BRAF and RAS (NRAS, HRAS and KRAS), were evaluate by PCR/sequencing with primers for the most frequent regions mutated on thyroid carcinoma, as previously described [42,43]. All detected mutations were validated by performing a new independent analysis.

2.4. Clinicopathological Features

Only PTCs were considered in the statistical analysis of the miRNAs expression and clinicopathological features of the nodules. The tumor’s clinicopathological features analyzed were: tumor size, extra thyroidal invasion, capsule invasion, vascular and lymphatic invasion, presence of fibrosis, inflammatory infiltrate, tall cell, oncocytic component, psammoma bodies, calcification, necrosis, and presence of lymph node metastases, focality, and laterality.

2.5. Statistical Analysis

An exploratory analysis was performed with categorical variables presented with frequencies (percentages), and quantitative variables with mean, and standard deviation (SD) or median, and percentiles (25-75) and/or range (minimum-maximum), as appropriate. Regarding quantitative variables, Student’s t and non-parametric Mann Whitney tests were used to compare groups, as appropriate. In the case of categorical variables, the Chi-square or Fisher’s exact tests were used, as required.

The discriminative ability of biomarkers regarding malignancy was performed by estimating sensitivity, specificity, positive and negative predictive values, with corresponding 95% confidence intervals. A level of significance α=0.05 was considered. Data analysis was performed using the SPSS software version 27.0 (IBM Corp. Released 2020. IBM SPSS Statistics for Windows, Version 27.0. Armonk, NY: IBM Corp).

3. Results

3.1. Series Description

3.1.1. Epidemiologic Data

From the 106 patients, 89 (84%) were female, with mean age at diagnosis of 52.2 (16.4) years (range 18-84), and 17 (16%) were male, with mean age of 55.2 (13.5) years (range 35-79). The distribution of age was the same across categories of gender (p =0.564), and across categories of histological diagnosis (p=0.503).

Mean tumor size was 33.1 mm (11.94) for benign lesions and 28.3 mm (14.76) for malignant lesions; the distribution of tumor size was the same across categories of histological diagnosis (p=0.152).

The histology samples were composed by 13 (12.3%) benign lesions, and 93 (87.7%) malignant lesions (DTC). PTC represented the larger group in malignant histology, with 81 (87.1%) cases.

Cytology samples were distributed according to the Bethesda classification, as follows: I. non diagnostic (ND), 3 samples (2.8%); II. benign (B), 23 samples (21.7%); III. atypia of undetermined significance (AUS), 23 samples (21.7%); IV. Follicular Neoplasm (FN), 29 samples (27.4%); V. suspicious for malignancy (SM), 14 samples (13.2%); VI. malignant 14 samples (13.2%).

The 49.1% (n=52) of nodules whose cytological result was indeterminate (Bethesda III and IV) corresponded in histology to one (1.9%) benign and to 51 malignant lesions (98.1%).

The distribution of cytology samples within histological subtypes in all series is presented in Table 1.

3.1.2. MiRNAS Profile in Cytology Samples

The miRNAs expression levels of miR-146b, miR-221, miR-222 and miR-15a in cytology samples by Bethesda categories is summarized in Supplementary Table S1. The levels of miRNAs expression observed across Bethesda categories exhibit elevated values, characterized by high median values, progressing from category I to category VI. Expressive variations in maximum values were noted among all miRNAs for each category, with the exception of miR-15a, which precludes statistically significant differences.

Indeterminate nodules (Bethesda III and IV) and nodules suspicious or malignancy (Bethesda V and VI) with higher miRNAs expression levels, proved to be malignant in histology, except in one benign case (Bethesda IV) as seen also in Table 1.

The miRNA expression levels of miR-146b, miR-221, miR-222 and miR-15a in cytology samples by histology diagnosis, in all series, is summarized in Figure 1. and Supplementary Table S2. The miRNAs profile observed in cytology samples showed a tendency to be overexpressed in malignant tumors when compared with benign lesions, with higher median for miRNAs 146b and 222 and wider amplitude of expression for all miRNAs in malignant tumors, although without statistical significance.

When we analyzed miRNA values obtained in cytology samples, according to the final diagnosis of malignancy, we noted that the malignant cases presented miRNA expression values above those obtained in benign cases. The number of malignant cases that presented (in the cytology sample) value of miRNAs expression above the maximum levels obtained for benign tumors, were 34.1% (n= 29) cases for miRNA-146b, 17.4% (n= 16) for miRNA-221, 35.4% (n=29) for miRNA-222, and 10.3% (n=9) for miRNA-15a, as showed in Table 2.

Malignant tumors presented, in cytology, higher values of miRNAs expression when compared with benign tumors. The five highest extreme values for miRNAs expression in cytology by histological diagnosis were shown in Supplementary Table S3.

3.1.3. MiRNAs Profile in Histology Samples

The miRNAs expression levels of miR-146b, miR-221, miR-222 and miR-15a in histology samples in all series is showed in Figure 2. and Table 3.

The miRNAs profile observed in histologies presented a statistical significant difference between benign and malignant tumors for all miRNAs (Table 3), showing higher median values and wider amplitude of expression in malignant tumors.

The miRNAs expression levels of miR-146b, miR-221, miR-222 and miR-15a in histology samples within histological subtypes is summarized in Supplementary Table S4.

Most histological subtypes of DTCs were PTCs (81/93), and presented higher median values and wider amplitude of expression for all miRNAs. Twelve cases were non-PTCs, from which two cases of HCC presented higher median values and wider amplitude of expression for miRNAs 221 and 222.

3.2. MiRNA Expression and Mutations in Cytology and Histology Samples in PTCs

For the analysis of the association between miRNAs expression levels and the presence of genetic mutations in cytology and histology samples, only PTCs were considered due to the limited number of the other subtypes of DTC.

In cytology samples, miRNAs profile revealed a higher median expression value in the presence of genetic mutations, as showed by miRNA-146b in the presence of TERTp and BRAF mutations, miRNA-221 and miRNA-222 associated with BRAF mutations, and miRNA-15a with TERTp and RAS mutations, although no statistically significant association was observed Supplementary Table S5.

In histology samples, miRNAs expression revealed a higher median value in the presence of almost all genetic mutations, and statistically significant associations between miRNAs expression levels and the presence of genetic mutations were obtained. MiRNA-146 and miRNA-221 expression levels were significantly associated with the presence of BRAF and RAS mutations, miRNA-222 with the presence of BRAF mutations, and miRNA-15a with the presence of TERTp and RAS mutations (Table 4).

3.3. MiRNA Expression and Clinicopatological Features in PTCs

The association between microRNAs expression and clinicopathological features of the tumors was evaluated in PTCs, being the results presented in Supplementary Table S6. A,B,C,D.

No statistically significant differences were achieved for gender or age. The distribution of tumor size was the same across microRNAs expression at diagnosis.

MiR-146b expression showed an association with capsular invasion, psammoma bodies (p=0.019) and calcification (p=0.010). Overexpressed miR-221 was associated with capsule invasion and vascular invasion (p< 0.001), oncocytic component (p=0.002) and calcification (p=0.031). miR-222 expression presented associations with the presence of extra thyroidal invasion, capsule invasion, vascular invasion, lymphatic invasion, oncocytic component, psammoma bodies, inflammatory infiltrate, tall cell component, calcification, and focality (p<0.05 for all parameters). miR-15a expression was significantly associated with capsular invasion and focality (p=0.41).

3.4. The Discriminative Ability of miRNAs in Histology for the Diagnosis of Malignancy

The discriminative ability of miRNAs profile for the diagnosis of malignancy in our series was analyzed only for PTCs, due to the limited number of the other malignant subtypes of DTC, as shown in Figure 3. and Table 5.

Area under the curve (AUC) for all miRNAs showed acceptable to exceptional discriminative ability regarding histological diagnosis, with estimates from 75.8% (95%CI: 62.7-88.9) for miR-222 to 93.5% (95%CI: 86.5-100) for miR-146b.

Regarding PTCs diagnosis and considering the cutoff values obtained by maximizing sensitivity and specificity, all miRNAs exhibited sensitivity (se) and specificity (sp) over 70%, with higher estimates for miR-146b (89.1% se and 87.5% sp). The positive predictive value (PPV) was over 93.5 % for all miRNAs, and attained a maximum of 97.6% for miR-146b.

The statistical analysis of combined miRNAs did not improve the discriminative ability of miRNAs for PTC diagnosis in our study. The analysis of miRNAs expression in 70 paired cyto-histology samples did not showed statistical significances.

4. Discussion

The expression of miRNAs is deregulated in many types of human cancers, including thyroid cancer, leading to alterations of different cellular processes, namely in cell differentiation, proliferation, migration and invasion [26]. Allied to genetic alterations, microRNAs have been described as new biomarkers in TC and are pointed as possible therapeutic targets [43,44,45].

In 2022, the 5th WHO classification of thyroid neoplasms [46] and in 2023, the 3rd Bethesda Classification [8], included genetic and molecular biomarkers for pre-surgical diagnosis to refine the diagnostic categories and their risk of malignancy. However, in addition to the lack of consensus on which panels of molecular biomarkers to use, their implementation in clinical practice is not yet possible in the vast majority of countries worldwide for technique and economic reasons.

This study aimed to determinate the expression levels of four microRNAs in cytology via US-FNAC, and in histology samples of patients with thyroid nodular disease that underwent surgery, based in clinical evaluation, imageology features and cytological criteria for suspicion of malignancy. We selected three miRNAs previously described as upregulated in TC (miR-146b, miR-221 and miR-222), and one categorized as downregulated (miR-15a), for expression profile analyses [47].

We observed a broad spectrum of variation in miRNAs expression levels, evident in both cytology and histology analyses. The study reveals a significant overexpression for miRNAs in malignant tumors when compared with benign lesions, in histology samples. Alterations of these specific miRNAs in TC have been already described in the literature, being associated with thyroid cell transformation [31,48].

In cytology samples, miRNAs showed a consistent trend for overexpression in malignant tumors compared with benign lesions, displaying higher median values across Bethesda categories I to VI. Notably, there were considerable differences in maximum values within each Bethesda category for all miRNAs, except miR-15a, despite no statistical significance have been achieved. In histology samples, all four miRNAs were overexpressed with a statistical significance between malignant tumors and benign lesions. The microRNA expression profile in cytology and in histology presented different levels with high amplitudes, which can reflect tumor heterogeneity, multifocality and/or low cellular representativity, particularly marked on cytology samples, with a lower percentage of cells (neoplastic) causing an increase of false negative results.

When we considered the values of miRNAs expression in cytology for malignant tumors under maximum levels for benign tumors (Table 2), the cytological information taken together with clinical evaluation, and imageology, could improved the diagnosis in one third of cases for miRNAs 146b and 222, almost 20% for miR-221, and even 10% for miR-15a. Those nodules could be diagnosed as malignant in a pre-surgical phase, without performing more FNACs and allowing better surgical options and health cost reduction.

In indeterminate nodules, miRNAs showed a tendency for overexpression in cytology, and with statistical significance in histology, reflecting their malignity in our series. Indeterminate nodules represented a high number of malignant cases in our series, because case selection was done based in a histology diagnosis of malignancy (or benignity in the 13 control cases), and these patients underwent surgery due to clinical reasons besides FNACs results (e.g. US features).

We found a tendency for upregulation in miR-146b, in cytology and statistically significant in histology, being in line with other studies [49,50]. MiR-146b is described to target SMADA and IRAK1, causing downregulation and increasing cells proliferation and migration in PTCs. [51].

The expression levels of miR-221 and miR-222 in cytology were higher in malignant tumors than in benign lesions, and reached statistical significance in histology, in accordance with Visone R. et al. [53]. Both miR-221 and miR-222 are described to target CDKN1B (p27Kip1 protein) and KIT. CDKN1B is an important regulator of the cell cycle, and KIT is involved in cell differentiation and growth. The overexpression of miR-221 and miR-222 induced the papillary thyroid carcinoma cell line (TPC-1) to progress into the S phase of the cell cycle. Moreover, miR-221 and miR-222 also play a role in vivo since there is an inverse correlation between their expression levels and down regulation of p27Kip1. On the contrary, the inhibition of miR-221 and miR-222 expression increases p27Kip1 protein levels. These results support the influence of these miRNAs in the control of the cell cycle through the regulation of the p27Kip1 protein levels.

The miR-15a has been described as downregulated in thyroid cancer. However, in our series, we observed an upregulation, suggesting a gain of function of this specific miRNA in PTCs. Jin J et al. [25] have reported an association between miR-15a low expression levels and human PTCs suggesting a possible influence of this miRNA in TC development by promoting proliferation, survival, and invasion. However, Lu Z et al. [35], showed that the upregulation of miR15a suppresses BCL-2 expression. In an in vivo study, it was found inhibition of cell proliferation and invasion, promoting apoptosis, when miR15a was upregulated, concluding that the overexpression of miR-15a inhibit tumor progression via regulation of AKT pathway. These discrepant results indicate that further studies are necessary to establish the role of miR-15a in thyroid cancer.

Several statistically significant associations between clinicopathological and molecular features of the PTCs were already described for BRAF mutations [54] and TERTp mutations [55], both associated with features of worse prognosis, whereas RAS mutations associated with benign or low-risk tumors [56,57]. Our data revealed a significant association between the expression levels of miR-146b, miR-221 and miR-222 and the presence of BRAF mutations in PTCs. Our findings are in accordance with Yang S. I. et al. [58] showing that the expression of miR-146b was significantly associated with the presence of BRAF mutations. Sun Y. et al. [59] showed that miR-146b, miR-221 and miR-222 together with miR-181 were upregulated in PTCs patients with BRAF mutations. These findings suggest a connection or influence of the presence of BRAF mutations with miRNAs expression in the development of TC. Our results showed a significant association between the expression levels of miR-146b, miR-221and miR15a, and the presence of RAS mutations in PTCs. The presence of RAS mutations confirms a neoplastic process but does not differentiate between benign, malignant, and low-risk neoplasm, whose analysis was beyond the scope of our study, as done by others [60]. Curiously, only miR-15a was significant associated with the presence of TERTp mutations in PTCs, suggesting that TERTp mutated PTCs can have a different repertoire of microRNA expression.

The (over)expression of the four microRNAs was significantly associated with several clinicopatological features in PTCs, namely for capsular invasion, vascular and lymphatic invasion, which were related with and an aggressive tumor behavior in accordance with others [30,34]. For example, upregulation of miR-146b was previously associated with tumor aggressiveness and poor clinicopathological features, such as extra thyroidal and capsule invasion, presence of lymph nodes or distant metastasis [52].

Our miRNAs expression analysis presents some limitations, namely the retrospective nature of the study and the small size of the series, both in cytology (lower representatives neoplastic cells), and in histology, due to the small number of non-PTCs subtypes. Both mentioned limitations plus the great variability of miRNAs expression level in cytology and histology, did not allowed the comparison in paired cyto-histological cases. Another limitation was the bias in the selection of indeterminate nodules in our series, making their study unfeasible, as done by others [61]. However, our results suggest that cytologies with overexpression levels of miRNAs have a higher risk to be malignant lesions, and miRNA evaluation may contribute to a pre-surgical diagnosis, changing the clinical and therapeutic decision.

Our results of miRNA expression profile in thyroid nodules reinforce the potential clinical utility of molecular testing in cytological assessments by US-FNAC, contributing to anticipate a diagnosis of malignancy and to predict tumor behavior. By integrating molecular information, clinicians achieve better informed decisions, and optimize treatment strategies.

5. Conclusions

Our results confirm that in TC there is a dysregulation of the expression levels of miRNAs, potentially influencing their target genes on the promotion of cell differentiation, proliferation and survival. While the miRNA profile´s ability to exclude malignancy is limited, we may suspect for malignancy in cytology samples with high expression levels of miRNAs, contributing to reduce repeated US-FNAC, and to a optimize the pre-surgical diagnosis . Moreover, the associations identified between miRNAs expression and clinicopathological features, in one hand, and genetic profile of PTCs, in the other hand, suggest that molecular analysis in US-FNAC can provide early insights into tumor genetic profile and behavior.

Current results suggest that US-FNAC molecular analysis will be useful in the early and accurate diagnosis of DTCs and, therefore, is likely to contribute to a more efficient treatment of TC. Yet, confirmation is pending from the analysis of a larger series.