In search of new stratification strategies: tissue proteomic profiling of papillary thyroid microcarcinoma in patients with localized disease and lateral neck metastases

Papillary thyroid carcinomas (PTC) are the most common thyroid malignancies that are often diagnosed as microcarcinomas when the tumor is less than one centimetre in diameter. Currently, there are no valid stratification strategies that would reliably assess the risk of lateral neck metastases and optimize surgical treatment. Aiming to find potential tissue biomarkers of metastatic potential, we conducted a cross-sectional proteomic pilot study on formalin-fixed paraffin-embedded tissues of metastatic (N = 10) and non-metastatic (N = 10) papillary thyroid microcarcinoma patients. Samples were analysed individually using liquid chromatography/mass spectrometry, and the differentially expressed proteins (DEP) were functionally annotated. We identified five overexpressed DEPs in the metastatic group (EPB41L2, CSE1L, GLIPR2, FGA and FGG) with a known association to tumour biology. Using bioinformatic-based tools, we found markedly different profiles of significantly enriched biological processes between the two groups. The identified DEPs might have a role as potential tissue biomarkers for PTC metastases. However, further prospective research is needed to confirm our findings.


Introduction
Papillary thyroid carcinomas (PTC) are the most common thyroid malignancies with one of the highest worldwide incidences-they occur in 14.9 out of 100,000 people (Kitahara et al. 2020;Limaiem et al. 2022).Most patients are diagnosed when the tumour is under one centimetre in diameter (papillary microcarcinoma), and although PTCs are slowgrowing and are generally considered to be less aggressive carcinomas, patient mortality increases in advanced disease stages (Wiltshire et al. 2016;Lim et al. 2017;Alok Pathak et al. 2013).Microcarcinomas are less aggressive than larger tumours, but they disseminate to other parts of the body in roughly a third of cases (McConahey et al. 1986;Moreno et al. 2012;Wada et al. 2008;Hughes et al. 1996;Harwood et al. 1978;Zaydfudim et al. 2008;Scheumann et al. 1994).Even though a majority of these tumours spread to neighbouring tissues, in 22% of cases "skip" metastases can bypass the primary neck region of thyroid lymph drainage and metastasize directly to lateral neck regions, thus complicating patient treatment (Chung et al. 2009;Zhao et al. 2019;Lim and Koo 2012;Park et al. 2012;Hartl et al. 2012).There are multiple risk factors associated with skip metastasis, including female sex, tumor multifocality, extracapsular spread, age over 55 years, and tumor location in the superior portions of the thyroid, due to lymphatic flow directed to the lateral neck (Wu et al. 2023).The inherent heterogeneity in these metastases makes them difficult to stage accurately, making uniform patient treatment impossible (Yang et al. 2023).Finally, occult neck metastases, that are commonly discovered after elective neck dissection (as opposed to physical examination and preoperative imaging studies), occur in up to 71.4% of papillary thyroid carcinoma patients, and are more frequent in younger patients (Frazell and Foote 1955;Wada et al. 2003;Lee et al. 2011;Hartl et al. 2012;Moreno et al. 2012).
According to the American thyroid association guidelines, the gold standard in the treatment of metastatic PTC is neck dissection, accompanied by a near-total or total thyroidectomy and the removal of other affected tissues (Haugen et al. 2016).However, after surgical treatment, metastatic recurrences happen in about 7-26% of cases of PTC, thus affecting patient disease-specific survival.To plan for the optimal extent of surgical treatment, reduce disease burden and the associated treatment morbidity, a highly specific and sensitive screening method is necessary to differentiate patients with aggressive forms of PTC (Giordano et al. 2015;Simon et al. 1996).Timely preoperative stratification of PTCs according to biological aggressiveness, paired with consequential appropriate surgical treatment could significantly reduce the morbidity and mortality of this disease.
Several studies have addressed the feasibility of proteomic profiling of papillary (micro)carcinomas to assess their metastatic potential.Lin et al. compared the proteomic profiles of papillary microcarcinoma aspirates with and without central metastases where they found an increase in the expression levels of 24 proteins and decreased expression levels of 89 proteins in metastatic carcinomas.Among these, the interferon-stimulated gene 15 was proposed as a prognostic biomarker of PTC patients with lymph node metastasis (Lin et al. 2019).Zhan et al. (2019) compared different degrees of PTC metastasis (no metastases-N0, central compartment metastases-N1a, lateral neck metastases-N1b) to healthy thyroid tissues in patients over 45 years, and reported that an increased expression levels of myosin 1 g and laminin subunit gamma mRNA were associated with tumour size, higher disease grade, and BRAFV600E with mutation and unfavourable prognosis.In a tissue-based proteomic study, Cao et al. compared papillary microcarcinoma tissues depending on metastases, although not specifying whether they were central or lateral, and found significant differences in differentially expressed proteins (DEP), namely, downregulation of Solute Carrier Family 25 Member 15, DIRAS Family GTPase 2, Phospholipase A2 receptor 1 and Mitochondrial Amidoxime Reducing Component 1, while these DEPs were enriched in mitochondrial dysfunction, possibly promoting tumorigenesis by activating the PI3K/ AKT signalling pathways.Interestingly, this research has also found significant differences in proteomic profiles depending on sex (Cao et al. 2022).
Even though some progress in the field has been made, to the best of our knowledge, there are no studies that compare tissue proteomes of papillary microcarcinomas, depending on the presence of lateral neck metastases.This is important because these metastases are often occult during primary surgery, and thus carry a higher morbidity and mortality rate.Considering their incidence and limited treatment options, a timely preoperative stratification of microcarcinomas would optimize surgical treatment and significantly reduce disease burden.To investigate this, we conducted a cross-sectional shotgun-based proteomic pilot study analysing the differences between proteomic profiles of papillary thyroid microcarcinoma tissue in patients with localized disease and lateral neck metastases.Furthermore, our aim was to identify potential target molecules which might serve as biomarkers that could be correlated to disease aggressiveness and metastatic potential.

Participants and study outline
This study was designed as a non-randomized cross-sectional pilot study on patients with pathohistologically confirmed papillary thyroid microcarcinoma.We included a total of 20 participants, which were divided into 2 research groups depending on the presence of lateral neck metastases at the time of surgery (N = 10 per group).The study was approved by the Institutional Ethics Committees of the Sestre milosrdnice University Hospital Center, Zagreb (EP-17295/18-1) and of the School of Medicine, University of Zagreb (case number 380-59-10,106-22-111/13).The study outline is presented in Fig. 1.
Written informed consent was obtained from all participants.Inclusion criteria were: (i) patients over 18 years of age; (ii) patients with pathohistologically confirmed micro-PTC that underwent lateral neck dissection (due to cytologically confirmed lateral neck metastatic disease); and (iii) signed informed consent.Exclusion criteria were: (i) prior oncological or surgical treatment; (ii) incomplete medical history; (iii) other thyroid malignancies; and (iv) distant metastases.

Sample collection
Patient data were collected from electronic medical records.Participants' formalin-fixed paraffin-embedded (FFPE) microcarcinoma tissue samples were collected from the Department of Pathology's bioarchive at the Sestre milosrdnice University Hospital Center, Zagreb.FFPE sample storage range was between 3 and 12 years, at the time when the research was conducted.Initially, all sample tissues were postoperatively fixed in 10% formalin and embedded in paraffin blocks.Standard pathohistological analysis was used to determine the tumour type, size, extra thyroidal extension, intraglandular dissemination, calcification, lymphangioinvasion, and in the case of dissection, the number of lymph nodes and the number of metastases dissected.In metastatic lymph nodes, the size of metastases and extranodal extension was also noted.In cases of carcinoma multifocality, the primary lesion determined by the pathologist and the one which was used in our analysis was the one with the largest diameter, while smaller clusters of carcinoma cells (according to the pathology reports none larger than 1 mm in diameter) were deemed as intraglandular dissemination of the largest lesion, thus rendering the carcinoma as multifocal.

Protein extraction and digestion
Proteins were extracted from FFPE tissues using a commercially available FFPE-FASP Protein Digestion Kit (ab270543) from Abcam (Cambridge, UK).Tissue was deparaffinised using xylene and ethanol, homogenized with UPX Universal Protein Extraction buffer, and heated to 99 °C for 45 min to release the formalin cross-linked proteins.The concentration of the extracted proteins was determined by Lowry method using the Bio-Rad RC DC Protein Assay Kit II.The extracted proteins were prepared for liquid chromatography-mass spectrometry (LC-MS) using 10 kDa molecular weight cutoff spin-filters.Briefly, after denaturation in 8 M urea, and alkylation using 55 mM iodoacetamide, the proteins were digested over-night using trypsin.The peptides were desalted and concentrated using C18-based Stage (Stop and Go Extraction) tips (Rappsilber et al. 2007).

Liquid chromatography-mass spectrometry
Purified peptides from individual patients were analysed in two technical replicates using nanoLC EASY-nLC™ 1200 System (Thermo Fisher Scientific) coupled to Q Exactive™ Plus Hybrid Quadrupole-Orbitrap™ Mass Spectrometer (Thermo Fisher Scientific) through a nano-electrospray LC-MS interface.Peptides were separated in a 60 min gradient of acetonitrile (0-80%) in 0.1% formic acid using 75 μm × 250 mm reversed-phase chromatography column (Thermo Fisher Scientific).Automated mass spectrometer measurement cycles consisted of full-scan MS and MS/ MS scans of up to 12 most intense ions (data dependent acquisition mode), using 20.0 s dynamic exclusion to avoid over-fragmentation (analysis) of the highest concentration peptides.Full MS scans (m/z from 350 to 1800) were obtained in an Orbitrap analyser (Thermo Fisher Scientific) at a resolution of 70,000 with internal calibration using Fig. 1 Study outline depicting subject groups and methodological approach.Proteins were isolated from formalin-fixed paraffin-embedded (FFPE) tissues of metastatic/nonmetastatic papillary thyroid microcarcinoma patients.After tryptic digestion, the peptides were sequenced by liquid chromatography-mass spectrometry (LC-MS).Experimental data were then analysed by appropriate bioinformatics software.(Created with BioRender.com)the "lock mass" setting.Top 12 ions were selected for fragmentation (MS2), recorded at resolution 17,500.

Data analysis
The MS raw files were processed using the Max Quant software 2.3.1.0.with integrated Andromeda search engine against the complete human proteome UniProt database (released 9th March 2023, Proteome ID: UP000005640) and internal contaminants database (Cox et al. 2011;Cox and Mann 2008).Cysteine carboxymethylation was set as fixed modification.Oxidation of methionine and N-terminal acetylation were set as variable modifications.Digestion mode was set to Trypsin/P with maximum two missed cleavages.Label-free quantification (LFQ) was used, and to increase the number of identified features, match between runs was enabled (Supplementary Fig. 1).LFQ intensities generated by MaxQuant were further analysed using the Perseus 2.0.9.0.software (Tyanova et al. 2016).Reverse hits, contaminants and proteins only identified by site were removed from further analysis.Biological replicates (patient samples) were grouped by condition (metastasis/ no metastasis) and LFQ intensities log transformed.Protein group entries with less than 70% valid values of log transformed LFQ intensity in each group were removed from further analysis.The difference between metastasis and no metastasis samples was determined by a bothsided, two sample t test.Proteins with p value ≤ 0.01 and mean log-transformed ratio metastasis vs. no metastasis higher than 1 or lower than − 1 were termed significantly DEPs.Furthermore, principal components analysis (PCA) was performed on the data with prior imputation of missing values with random values from normal distribution (MinProb method), with parameters: width 0.3 and down shift 1.8 (Elias and Gygi 2007).The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al. 2022) partner repository with the dataset identifier PXD042415.
To include in the bioinformatics analysis only those proteins that were characteristic of the metastatic or nonmetastatic state, a certain protein had to be expressed in a minimum of 70% of patient samples within its experimental group (Supplementary Table 1).Functional enrichment of the identified proteins was performed using SR plot, an online platform for data analysis and visualization (SR plot 2023).Proteins with a twofold ratio of overexpression/ underexpression between the experimental groups were included, as were proteins expressed exclusively in a certain patient group.Identified proteins from metastatic and nonmetastatic patients were enriched for the biological processes they are involved in.To plot the detected proteins on a volcano plot, proteins with (i) p value ≤ 0.01 determined by a both-sided, two-sample t test and (ii) mean log-transformed ratio metastasis vs. no metastasis higher than 1 or lower than − 1 were termed as significantly DEPs.The heatmap was plotted using the Morpheus tool (Morpheus 2023).Protein interactions and gene ontology analysis within the interactomes was done using the String 11.5 software (Database 2023).Literature search was then conducted in the MEDLINE database for all proteins found in both groups, to elucidate their potential role in PTC aggressiveness and metastasizing.Pathophysiological significance was determined and reported by manual curation.

Demographic and sample data
Patient demographics and clinical characteristics, as well as tumour sample specifications are presented in Tables 1  and 2. Pathohistological analysis of the patient group without metastases has shown that none of the tumours had signs of calcification, extra thyroidal extension, multifocality or lymphangioinvasion.However, six cases of extraglandular extension and multifocality of the primary tumour were found in the metastases group, with one case with lymphangioinvasion and calcification.Average nodal yield was 19.8 ± 10.3 lymph nodes, with an average of 6.7 metastatic lymph nodes in the dissection specimen, resulting in a lymph node ratio of 0.34, which is close to the preferable lymph nodal yield and ratio for minimizing cancer recurrence (Weitzman et al. 2022;Nam et al. 2018;Ji et al. 2019).

Protein identification
Isolation of proteins from the tissues of non-metastatic and metastatic PTC patients and further analysis of their composition yielded similar amounts of detected proteins.Proteins isolated from the tissues of PTC patients with and without metastasis (ten patients per sample group) were individually identified and quantified using LC-MS.Technical duplicates showed a high consistency in the number of sequenced peptides/identified proteins across the experimental samples (Supplementary Fig. 1).In total, 2289 of proteins were identified and quantified that correspond to the proteome of papillary thyroid microcarcinoma patients.A majority of identified proteins (73%) were identified in both non-metastatic and metastatic patients; however, the non-metastatic patients had a sizeable amount of specific proteins expressed (24%), as compared to patients with lateral neck metastases (3%).To estimate the similarity between the individual samples in both groups, we examined their clustering using PCA.The analysis showed an overlap in protein expression between individual patients in the two experimental groups.In addition, protein clusters isolated from patients without metastasis showed a higher degree of correlation compared to metastatic patients (Fig. 2).

Gene enrichment
We used gene enrichment to identify and rank those biological processes that are overrepresented in a group of genes more than would be expected by chance.Using the SR plot software, proteins identified from the tissues of non-metastatic and metastatic PTC patients were associated with the processes that were designated as statistically significant.Gene enrichment results show that, in the metastatic group, biological processes of post-translational protein modification and protein remodeling are more expressed (namely, protein-lipid complex remodeling, plasma lipoprotein particle remodeling, protein-containing complex remodeling), as well as substrate adhesion-dependent cell spreading.Of the differentially expressed proteins, FGA and FGG were associated with post-translation protein modification processes and  Each group consisted of ten patients.The samples of the patients without metastasis (blue) group on the left-hand side of the PCA plot (circled), while the samples of the patients with metastasis (orange) are not clustered and mostly fall outside the circled area substrate adhesion-dependent cell spreading.On the other hand, processes more expressed in the non-metastatic group are more related to certain immune-response processes, as shown in Fig. 3.One of the DEPs in the non-metastatic group, RPS15A, is involved in RNA and mRNA catabolic processes, which were more expressed in that group.In general, gene enrichment showed that processes affecting protein structure and function are more prevalent in the metastatic group.

Differences in expression
To visualise the distribution and identity of proteins that are most differentially expressed between the non-metastatic and metastatic PTC patients we plotted the quantified proteins that were expressed in a minimum of 70% of patients in both experimental groups on a volcano plot (Fig. 4).Proteins found to be significantly overexpressed (p ≤ 0.01 and at least twofold upregulated) in the metastasis group are band 4.1-like protein 2 (EPB41L2; p = 0.0068), fibrinogen alpha/gamma chain (FGA/FGG; p = 0.0061 and 0.0037, respectively), exportin-2 (CSE1L; p = 0.0053), Golgi-associated plant pathogenesis-related protein 1 (GLIPR2; p = 0.0005), while those with significantly lower expression (p ≤ 0.01 and at least twofold downregulated) are cytochrome c oxidase subunit 6C (COX6C; p = 0.0029), ferredoxin reductase (FDXR; p = 0.0085), 40S ribosomal protein S15a (RPS15A; p = 0.0046).To visualize the differences between the statistically significant upregulated and downregulated proteins on an individual patient level of both groups, we used heatmaps and panels shown in Fig. 5.Although there exists a degree of variability of DEP expression between patients, the heatmap shows a general pattern of DEP Fig. 3 Comparative analysis of the biological processes associated with the identified proteins of the non-metastatic (A) and metastatic (B) patient group revealed key metabolic differences between the groups.Individual dot size corresponds to the number of annotated proteins within a group.Biological processes are ranked and colour coded according to their respective Enrichment Score, i.e., statistical significance.MHC major histocompatibility complex, TAP Transporter associated with Antigen Processing, ER endoplasmatic reticulum over-or under-expression across the two groups of patients for the individual proteins.Proteins that were found to be expressed only in the tissues of patients with metastatic PTC were plotted on an interaction network using the String 11.5 software (Fig. 6).Clustering occurred in two distinct groups, one being associated with endoplasmatic reticulum targeting, exosome, vesicle and cytoplasm/cytosol components, while the other was predominantly associated with lipoprotein components.The strongest clustering was observed for proteins attributed to intermediate, very low and high-density lipoprotein particles and chylomicron cellular components, and endoplasmic reticulum targeting.The physiologic roles of the two clustering groups we observed are accordant to the overly expressed processes identified after gene enrichment, as shown in Fig. 3.

Discussion
Due to their small size and subtle clinical presentation, papillary thyroid microcarcinomas are challenging to detect.Optimal risk stratification, treatment and follow-up strategies are still being discussed.Although the prognosis for patients with these tumors is usually favourable, a particular subset of patients develops aggressive forms of the disease leading to regional and distal metastases, negatively impacting the disease prognosis.An early stratification method could be used to drive optimal diagnostic, therapeutic and follow-up strategies, but current prognostic factors for PTC are unable to account for differing biological aggressiveness of certain carcinomas.As patients would benefit from a tailored surgical or adjuvant treatment, additional individual data are required to complement currently used clinical risk assessment criteria.Therefore, the potential use of biomarkers to identify PTCs with a higher metastatic potential prior to treatment and post-treatment follow-up, could significantly reduce disease burden.
DEPs found in this research are linked to carcinogenesis of many tumors, but, to the best of our knowledge, this is the first study linking EPB41L2, CSE1L, GLIPR2 and FGG to papillary thyroid (micro)carcinomas.The upregulated EPB41L2 is a member of the band 4.1 family of proteins that have a role in cell cycle, division and transport, but their role in cancer biology appears to be dichotomous.On one hand, they inhibit tumor cell growth, however they also suppress cAMP production, which can promote the growth of PTC cells (Wang et al. 2014;Goto et al. 2013;Kamma et al. 2008).Furthermore, CSE1L (Exportin-2) is a receptor that mediates importin-alpha export form the nucleus into the cytoplasm (Kutay et al. 1997).Along with other members this family, it has a role in carcinogenesis: affecting apoptosis, tumor suppressor responses, oncogene activity and transduction pathways.In addition, it is highly expressed in proliferating human cells and various cancers, especially of higher grade and stage (Tanaka et al. 2007;Nachmias and Schimmer 2020;Tai et al. 2010) It is potentially useful as a biomarker of metastasis, since a higher prevalence of secretory Exportin-2 in patients' sera with metastatic cancer, has also been noted; while identifying miRNArelated single-nucleotide polymorphisms (miR-SNPs) of the exportin-associated genes has been suggested as a biomarker in thyroid cancer prediction (Wen et al. 2017;Tung et al. 2009).GLIPR2, which is also one of the overexpressed DEPs, is known to play a role in autophagy and has, interestingly, been reported to also have a role in promoting epithelial-to-mesenchymal transition in kidney fibrosis and hepatocellular carcinoma (Baxter et al. 2007;Gibbs et al. 2008).
Fibrinogen, a large protein associated with coagulation was also implicated in tumor stroma formation and promotion of neovascularization (Palumbo et al. 2000;Staton et al. 2003).The pathophysiological role of fibrinogen in tumor metastasis is implied by its positive regulation of cell spreading, promoting epithelial-to-mesenchymal transition and augmenting the effects of fibroblast growth factor-2 (Zhang et al. 2019;Sahni et al. 2008) Although FGA and FGG have been found to be potential biomarkers of several cancers (Song et al. 2022;Ward et al. 2010;Liu et al. 2012), of special interest is a paper by Lu et.Al, which found that FGA might be a potential serum biomarker for PTC (Lu et al. 2018).These findings implicate the potential significance of altered fibrinolysis in thyroid cancer patients with distant metastases.
An underexpressed protein in the metastatic group was COX6C, a component of the Cytochrome C oxidase complex (Tian et al. 2021).The knockdown of COX6C may inhibit apoptosis and lead to cancer development and, on a similar note, recently it was also found that a somatic alteration of COX6C/DERL2 is present in follicular thyroid tumors (Lange et al. 2016;Chen et al. 2022).Another downregulated apoptosis modulator detected was FDXR, ferredoxin reductase, a target of p53 and p73 that has an effect on tumor suppression by inducing apoptosis (Zhang et al. 2017(Zhang et al. , 2020;;Liu and Chen 2002).Finally, although the high expression of Ribosomal protein S15a (RPS15a) plays a role in carcinogenesis of many cancers, we have found that its lower levels in thyroid cancer are linked to metastasis (Liu et al. 2022;Zhao et al. 2015;Wan et al. 2021).
Gene enrichment analysis showed that among the enriched biological processes, posttranslational protein modification and several protein remodeling processes were especially evident in the metastatic group.Posttranslational modifications play important roles in thyroid tumor biology-they are known to be involved in regulation of proliferation, tumorigenesis, invasion and metastatic development, and may thus dictate disease progression development (Broekhuis et al. 2022;Pan and Chen 2022).In addition, positive regulation of substrate adhesion-dependent cell spreading was enriched in the metastatic group.The process of adhesion-dependent spreading has been suggested to modulate cancer metastasis (Morales et al. 2022).Interestingly, the differentially expressed FGA and FGG, have been associated with substrate adhesion-dependent cell spreading and posttranslational protein modification-suggesting their role in facilitating the spread of PTC and, therefore, their potential use as predictors of PTC metastasis.
The aim of this pilot study was to identify proteins of interest that would be valid biomarker candidates for further research.We did not find a DEP that could be exclusively attributed to metastatic or non-metastatic PTC patients.Due to the nature of our results, a single biomarker candidate seems unlikely, and instead research should be aimed at using a panel of over-and under-expressed DEPs that might be used to predict the metastatic potential of these microcarcinomas.The main limitations of this study is lie in the fact that it was designed as a single centre pilot study, which likely cannot fully represent the variability of this disease.In addition, the sample size is rather small and our findings are based solely on LC-MS without the use of complimentary methodology such as, for example, immunohistochemistry.However, we believe our study opens new horizons in micro-PTC research: namely, although some research exists on thyroid cancer proteomics, our work focuses on discriminating between the presence or absence of lateral neck metastases, which may have an important clinical application (Lin et al. 2019;Zhan et al. 2019;Cao et al. 2022).It is currently impossible to confidently differentiate between indolent and aggressive forms of recurrent metastatic disease.The only clinically useful diagnostic test is the serum thyroglobulin levels, and up to some extent, BRAFV600 mutation status (Lamartina et al. 2019;Viola et al. 2023).Thus, the overexpressed proteins in the metastatic group have previously not been associated with PTC metastasis, and are thus candidates for further validation as biomarkers.In addition, identification of FGA and FGG overexpression implicates the potential importance fibrinogen metabolism alteration might play in thyroid tumor biology.To confirm our results, larger focused longitudinal prospective studies are needed, preferably aimed to validate and quantify the overexpressed proteins in metastatic PTC patients.
Fig. 5 General overview of the study principle and the differentially expressed proteins (DEPs) (A), presented as a heat map (B) showing the log 2 transformed label free quantification values of the top upregulated and downregulated proteins (p < 0.01) as compared between metastatic and non-metastatic papillary thyroid microcarcinoma patients.C Panels show variability in the relative expression levels (REL) of discrete proteins in thyroid cancer tissues of individual patients.EPB41L2 band 4.1-like protein 2, FGA/FGG fibrinogen alpha/gamma chain, CSE1L exportin-2, GLIPR2 Golgi-associated plant pathogenesis-related protein 1, COX6C cytochrome c oxidase subunit 6C, FDXR Ferredoxin reductase, RPS15A 40S ribosomal protein S15a ◂

Fig. 2 A
Fig. 2 A Venn diagram showing the overlap in proteins expressed in papillary thyroid microcarcinoma patients with (orange) and without (blue) metastasis.B Principal component analysis (PCA) plot representing proteomics data from the comparative analysis of papillary thyroid cancer patients with (orange) and without metastasis (blue).

Fig. 4
Fig. 4 Volcano plot comparing protein expression levels in thyroid tissue in two groups of papillary thyroid microcarcinoma patients: with and without metastasis.Up-and down-regulated proteins are shown in green and red circles, respectively.The dashed lines on the graph indicate thresholds (p value = 0.01 and mean log-transformed ratio metastasis vs. no metastasis equal to 1 or − 1)

Table 1
Patient demographics and clinical characteristicsNumbers are expressed as mean and range or as number

Table 2
Lymph node and metastases characteristicsNumbers are expressed as mean and range or as number