The role of MAPK, notch and Wnt signaling pathways in papillary thyroid cancer: Evidence from a systematic review and meta-analyzing microarray datasets employing bioinformatics knowledge and literature

Abstract

Papillary thyroid cancer (PTC) is a prevalent kind of thyroid cancer (TC), with the risk of metastasis increasing faster than any other malignancy. So, understanding the role of PTC in pathogenesis requires studying the various gene expressions to find out which particular molecular biomarkers will be helpful. The authors conducted a comprehensive search on the PubMed microarray database and a meta-analysis approach on the remaining ones to determine the differentially expressed genes between PTC and normal tissues, along with the analyses of overall survival (OS) and recurrence-free survival (RFS) rates in patients with PTC. We considered the associated genes with MAPK, Wnt, and Notch signaling pathways. Two GEO datasets have been included in this research, considering inclusion and exclusion criteria. Nineteen genes were found to have higher differences through the meta-analysis procedure. Among them, ten genes were upregulated, and nine genes were downregulated. The expression of 19 genes was examined using the GEPIA2 database, and the Kaplan-Meier plot statistics were used to analyze RFS and the OS rates. We discovered seven significant genes with the validation: PRICKLE1, KIT, RPS6KA5, GADD45B, FGFR2, FGF7, and DTX4. To further explain these findings, it was discovered that the mRNA expression levels of these seven genes and the remaining 12 genes were shown to be substantially linked with the results of the experimental literature investigations on the PTC. Our research found nineteen panels of genes that could be involved in the PTC progression and metastasis and the immune system infiltration of these cancers.

Highlights

• •The associated genes with MAPK, Wnt, and Notch signaling pathways were considered.
• •Nineteen genes were found to have higher differences through the meta-analysis procedure.
• •Seven significant genes were PRICKLE1, KIT, RPS6KA5, GADD45B, FGFR2, FGF7, and DTX4.
• •Nineteen panels of genes could be involved in the PTC progression and metastasis and the immune system infiltration of these cancers based on corresponding analyses.

Abbreviations:

ANOVA

One-way analysis of variance

AP-1

Activator protein 1

ATC

Anaplastic thyroid carcinoma

BLBC

Basal-like breast cancer

CCND1

Cyclin D1

CTNNB1

Catenin beta 1

DTX4

Deltex E3 ubiquitin ligase 4

DUSP

Dual specificity phosphatase

EGFR

Epidermal growth factor receptor

ERBB3

Erb-b2 receptor tyrosine kinase 3

ERK

Extracellular signal-regulated kinase

FDR

False discovery rate

FGF7

Fibroblast growth factor 7

FGFR2

Fibroblast growth factor receptor 2

FTC

Follicular thyroid carcinoma

GADD45B

Growth arrest and DNA damage-inducible beta

GEO

Gene expression omnibus

GEPIA2

Gene expression profiling interactive analysis v.2

IGF2

Insulin-like growth factor 2

IL1RAP

Interleukin 1 receptor accessory protein

Jag

Jagged

JNK

c-Jun N-terminal kinases

HSPA1B

Heat shock protein family A (Hsp70) member 1B

KEGG

Kyoto encyclopedia of genes and genomes

KIT

c-KIT

KM

Kaplan-Meier

LEF/TCF

Lymphoid enhancer factor/T cell factor

MAPK

Mitogen-activated protein kinase

MTC

Medullary thyroid carcinoma

NCBI

National center for biotechnology information

OS

Overall survival

PDGF

Platelet-derived growth factor

PTC

Papillary thyroid cancer

qRT-PCR

Quantitative reverse transcriptase-polymerase chain reaction

RFS

Recurrence/disease/relapse-free survival

RPS6KA5

Ribosomal protein S6 kinase A5

RT-PCR

Real time polymerase chain reaction

MAPK13

Mitogen-activated protein kinase 13

RAC2

Ras-related C3 botulinum toxin substrate 2

PRICKLE1

Prickle homolog 1

TC

Thyroid cancer

TCF7L1

Transcription Factor 7-Like 1

TCGA

The cancer genome atlas

TNF

Tumor necrosis factor

HEY2

Hairy/enhancer-of-split related with YRPW motif protein 2

1. Background

Thyroid cancer (TC) is the most common endocrine malignancy, ranking 9th with 586,000 cases globally; it has received increased attention in recent decades due to its rising incidence rate in many countries [[1], [2], [3]]. TC is usually associated with several significant risk factors, including age, specific genetic abnormalities, gender, environmental conditions, and lifestyle [4,5]. Medullary thyroid carcinoma (MTC), follicular thyroid carcinoma (FTC), anaplastic thyroid carcinoma (ATC), and papillary thyroid carcinoma (PTC) are some of the histological forms of TC [6]. PTC has the highest rate of occurrence among these groups, contributing to more than 80 % of all TC cases, with incidence and mortality of 3.1 % and 0.4 %, respectively. Eastern Asia, Australia/New Zealand, Northern America, and Micronesia/Polynesia have the highest national incidence rates, according to the GLOBOCAN 2020 statistics [3].

Because most patients of PTC have a better condition following multiple therapies such as chemotherapy, radiotherapy, and surgical resection, lymph node metastasis and local relapse are still common in PTC patients, causing increased mortality [[7], [8], [9]]. Despite enormous achievements in the PTC prognosis, a discrepancy exists in the knowledge of the disease's molecular and functional processes and the involved signaling pathways [10]. The development of the PTC has recently demonstrated the involvement of corresponding genes of the mTOR and PI3K signaling pathways [11,12]. According to the findings, Three gene biomarkers were downregulated in this disorder. So, the low levels of expression of these genes may be helpful in drug development and delivery systems in PTC patients. In the present study, the authors have been encouraged to research the role of other critical signaling pathways, including Wnt, MAPK, and Notch, in PTC development. The Wnt/β-catenin is a well-known pathway linked to some essential cell functions [13,14]. This pathway is involved in initiating and progressing some different malignant tumors.

Moreover, it is activated in several human cancers, and this activation contributes to tumor growth and recurrence by regulating cancer cell proliferation, apoptosis, and the epithelial-mesenchymal transition (EMT) processes [[15], [16], [17]]. This pathway's activation promotes tumor cell migration and invasion [18]. Mitogen-activated protein kinase (MAPK) is another pathway that the genetic changes in its signaling components such as RET/PTC, RAS, and BRAF have been well explored in the PTC and lead to constitutive MAPK signaling pathway activation [[19], [20], [21], [22]]. Targeted expression of the RET/PTC or BRAF^T1799A oncogenes in transgenic mice showed that they are involved in the TC, indicating mutations in MAPK signaling components contribute to PTC development [23,24]. The notch signaling pathway is needed for tissue homeostasis development and maintenance [25]. The Notch signaling pathway currently consists of four mammalian receptors and at least five ligands. A large-scale gene expression study in the PTC revealed many Notch signaling components [26].

Since cancers are categorized as complex diseases, and the PTC is not an exception, we have studied three signaling pathways (Notch, Wnt, and MAPK) and their corresponding genes to expound further on the critical roles of significant genes between the PTC and normal tissues. To satisfy the research aim, we have systematically searched the National Center for Biotechnology Information-Gene Expression Omnibus (NCBI-GEO) database to extract the potential GEO microarray datasets by excluding non-relevant datasets from further analysis. Finally, as mentioned above, the selected datasets were the target for the meta-analysis approach to determine the essential biomarkers affecting the signaling pathways. The determined differentially expressed genes will also be analyzed for different survival rates and validity through the literature's experiments.

2. Methods

2.1. Data sources and search strategies

The publicly available microarray database (i.e., NCBI-GEO) was selected for the thorough search using a Boolean query to satisfy the research aim. General terms employed in the query included either "papillary thyroid cancer" or PTC. Several inclusion and exclusion criteria were considered to filter out the search results obtained from the NCBI-GEO database. The remained GEO datasets, in terms of the webpage contents and their corresponding published articles, should provide those extracted from tumor tissue of Homo sapiens organism and with experiment type of "Expression profiling by array" by including both normal and cancer mRNA samples. Additionally, all platform types were included for not losing any potential datasets. Finally, the GEO datasets considering the inclusion criteria were used to study the Wnt, MAPK, and Notch signaling pathways through meta-analysis.

2.2. Identification of potential common genes

The list of genes involved in Wnt (hsa04310), MAPK (hsa04010), and Notch (hsa04330) signaling pathways was available online in The Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Then, the associated genes of the three signaling pathways mentioned above were the targets for the input for the meta-analysis procedure.

2.3. Meta-analysis approach

We used the meta-analysis tool for gene expression datasets, ExAtlas (i.e., http://alexei.nfshost.com/exatlas/) for performing multi-platform of GEO datasets [27]. The main characteristics of the ExAtlas used for the GEO datasets, including only the gene expressions of Wnt, MAPK, and Notch signaling pathways complying with the inclusion criteria, were based on random-effects model (because of heterogeneous GEO microarray datasets), z-score as well as Fisher's methodologies. Before performing the meta-analysis approach, all GEO datasets were log2 transformed, quantile normalized and analyzed through t-test one-way analysis of variance (ANOVA). The standard deviation values of less than 0.3, the false discovery rate (FDR) of 0.05, and the fold change of 2 were the criteria for inspecting the quality of gene expression datasets and the meta-analysis procedure, respectively.

2.4. Post-processing approaches

Generally, the analysis of the survival rates, through plotting the Kaplan-Meier (KM) estimate, comprised statistical interpretation of clinical or experimental data and the follow-up time of about five years where the health status of the patients is of interest in terms of mortality or recurrence rates. We used the Gene Expression Profiling Interactive Analysis v.2 (GEPIA2) (http://gepia2.cancer-pku.cn/#index) free tool to obtain both overall survival (OS) and relapse-free survival (RFS) rates using The Cancer Genome Atlas (TCGA)-THCA database where the number of samples for the TC and control were 512 and n = 59, respectively [28,29]. The default values for the p-value and confidence interval of the hazard ratio were 0.05 and 95 %, respectively.

The National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) and PanCanAtlas (version 20190101) provided molecular and clinical information on 11263 individuals with 33 different kinds of cancer to be utilized in CVCDAP's release 2019 [30]. Using CVCDAP, researchers may easily construct a virtual cohort using samples from a single study or a collection of studies with similar molecular or clinical features. A virtual group may be created using any combination of tissue, clinical, and molecular characteristics. These may include disease type, disease stage, age, somatic mutations, and degree of mRNA or protein expression, to mention a few. It is possible to store cohorts with another saved set to perform operations such as union, intersection, or subtraction. CVCDAP's enhanced characteristics may make it possible to identify molecular processes underlying biological or clinical topics of interest much more quickly than possible. We used the multivariate Cox regression tool from CVCDAP to investigate the possible correlations between significant genes and the union of BRAF-, KRAS-, HRAS-, and NRAS-like mutations based on hazard ratio measurement.

It is customary to create risk categories using the prognostic index (PI), often known as the "risk score." As the Cox model's linear component, the PI is denoted by the equation PI = b₁x₁+b₂x₂+ … +b_px_p, where xi is the gene expression value and b_i, the risk coefficient, is derived via the Cox fitting computational procedure. To calculate the b coefficients, we used the SurvExpress web server tool [31]. The Cox model includes all imported genes in a single model to create risk categories regarding the concordance index.

The TCGA thyroid carcinoma study (TCGA-THCA) data was collected from 507 patients with papillary thyroid cancer (PTC) and 59 controls. Further investigation of significant DEGs in Firebrowse (http://firebrowse.org) was based on identifying associations between the level 3 mRNA sequencing data and clinical information (i.e., extrathyroidal extension(ETE) and tumor TNM stages) [32]. The False Discovery Rate (FDR) counterpart of the P-value is the Q value for multiple hypothesis correction, defined as the lowest FDR at which the test may be considered significant [33]. The website utilized R's 'p.adjust' function using the 'Benjamini and Hochberg' technique to get Q values from P values.

3. Results

3.1. Identified GEO datasets through a systematic review

The workflow of systematically querying GEO datasets and GEO profiles is depicted in Fig. 1, along with the reasons for including and excluding a specific GEO dataset. The search procedure identified 1916 items, of which 323 non-Homo sapiens datasets and 1831 datasets using other profiling platforms were excluded (a total of 1865). After carefully inspecting the contents of 51 GEO datasets, 47 were from different sources. Samples types (i.e., non-mRNA), the microarray data for four remaining GEO datasets (viz., GSE29265, GSE97001, GSE3678, and GSE138198 with a total of 34 healthy and 37 PTC samples) were downloaded for further analysis and quality check before the meta-analysis procedure.

Fig. 1.

The flow chart diagram for systematic search and determined eligible GEO datasets.

3.2. Meta-analysis outcomes

Before importing the GEO datasets in the ExAtlas website, the in-common genes between the GEO datasets and the genes associated with three signaling pathways separately will result in the identification of significant differentially expressed genes of 59 genes related to the Notch signaling pathway, 166 genes related to Wnt signaling pathway, and 294 genes associated with MAPK signaling pathway between two tissue types. The initial quality control of the samples showed that they passed the first assessment stage of GEO datasets; after performing the meta-analysis procedure on the two GEO datasets for three signaling pathways, various numbers of genes were differentially expressed, employing the p-value and FDR measurements.

3.3. Identification of DEGs in notch, MAPK, and Wnt signaling pathways

For the Notch signaling pathway, two upregulated genes deltex E3 ubiquitin ligase 4 (DTX4, p-value = 8.53E-10) and hes-related family bHLH transcription factor with YRPW motif 2 (HEY2, p-value = 0.000745) were identified. Furthermore, for the Wnt signaling pathway, two upregulated Cyclin D1 (CCND1, p-value = 0) and prickle homolog 1 (PRICKLE1, p-value = 3.91E-07) as well as one downregulated gene including transcription factor 7-like 1, (TCF7L1, p-value = 9.22E-07) was identified. Finally, for the MAPK signaling pathway, six upregulated and eight downregulated significant genes were found, and the details are listed in Table 1.

Table 1.

Differentially expressed genes were identified through a meta-analysis between the PTC and healthy samples for Notch, Wnt, and MAPK signaling pathways.

No	Signaling pathway	Gene symbol	Fold Change	p-value	FDR	Current study	Literature evidence	Methods	Ref
1	Notch	DTX4	3.886	8.53E-10	4.01E-08	Up	Up	●RT-qPCR	[34]
2		HEY2	5.767	0.000745	0.0175	Up	Up	●Immunocytochemistry ●Western blot	[35]
							Up	●In silico	[36]
3	Wnt	CCND1	2.672	0	0	Up	Up	●In silico	[37]
							Up	●RT-qPCR	[38]
4		PRICKLE1	2.349	3.91E-07	2.36E-05	Up	Not stated	●Immunocytochemistry ●qRT-PCR	[39]
							Not stated	●In silico	[40]
5		TCF7L1	2.305	2.305	2.305	Down	Down	●Immunocytochemistry ●Western blot	[35]
							Down	●In silico	[41]
6	MAPK	IL1RAP	4.176	3.48e-12	1.68e-10	Up	Up	●RT-qPCR	[34]
							Up	●NanoString ●RNA-sequencing	[42]
7		DUSP4	4.167	0	0	Up	Up	●RT-qPCR	[43]
							Up	●PCR	[44]
							Up	●qRT-PCR	[45]
8		RAC2	2.466	0.000957	0.0163	Up	Down	●qRT-PCR ●In silico	[46]
							Up	●FISH and RT-PCR ●In silico	[47]
9		ERBB3	3.192	4.92E-09	1.82E-07	Up	Up	●RT-qPCR	[48]
10		MAPK13	2.044	2.19E-07	6.07E-06	Up	Up	●In silico	[49]
11		DUSP5	2.659	2.95e-05	0.000536	Up	Up	●qRT-PCR	[45]
12		KIT	6.835	4.72E-08	1.50E-06	Down	Down	●Immunocytochemistry	[50]
							Up	●Immunocytochemistry	[51]
							Down	●NanoString nCounter	[52]
13		PDGFRA	3.459	1.65E-06	4.08E-05	Down	Up	●cDNA microarray analysis ●Western blot	[53]
							Up	●TaqMan qRT-PCR ●Microarray hybridization	[54]
							Up	●Western blot	[55]
							Down	●miRNA microarray analysis	[56]
							Up	●Immunohistochemistry	[57]
14		IGF2	3.112	0.000101	0.001862	Down	Down	●Microarray analysis ●RT-PCR	[58]
15		RPS6KA5	2.074	6.99E-06	0.000155	Down	Down	●Western blot ●qRT-PCR	[59]
16		GADD45B	2.47	3.75E-13	1.67E-11	Down	Down	●qRT-PCR	[60]
17		HSPA1B	2.308	1.71E-05	0.000345	Down	Down/Up	●Review of Cancers	[61]
18		FGFR2	2.122	1.82E-14	1.01E-12	Down	Down	●qRT-PCR ●Western blot	[62]
19		FGF7	2.556	0.003146	0.0499	Down	Down	●miRNA microarray analysis	[56]

3.4. Post-processing outcomes

The OS and RFS analyses identified two upregulated and five downregulated genes, as illustrated in Fig. 2, Fig. 3. Among upregulated genes, PRICKLE1 and DTX4 were differentially expressed in OS rate with p-values 0.032 and 0.025, respectively. Moreover, by considering the downregulated genes, three genes growth arrest and DNA damage-inducible beta, protein kinase C alpha, and fibroblast growth factor 7 (i.e., GADD45B, p-value = 0.034; KIT, p-value = 0.012; FGF7, p-value = 0.014) were found to be significant in terms of OS rate; however, four genes c-KIT (KIT, p-value = 0.012), ribosomal protein S6 kinase A5 (RPS6KA5, p-value = 0.026), GADD45B (p-value = 0.025), and fibroblast growth factor receptor 2 (FGFR2, p-value = 0.029) were significant in terms of RFS rate.

Fig. 2.

KM plots of the OS results obtained from the identified genes involved in the PTC.

Fig. 3.

KM plots of the RFS results obtained from the identified genes involved in the PTC.

V-RAF murine sarcoma viral oncogene homolog B1 (BRAF) and rat sarcoma (RAS), specifically KRAS, NRAS, and HRAS, are essential for DNA repair or damage mutations in BRAF-like or RAS-like mutations across various cancers. The oncoplot tool of CVCDAP identified that thyroid cancer (THCA) has the mutation frequency for BRAF (59 %) and RAS (12 %), with patients harboring BRAF/RAS mutations (Fig. 4). Hence, the clinical implications of BRAF and the clinical significance of RAS mutations are important. Thus, we utilized CVCDAP to perform a multivariate Cox analysis of THCA patients using a union cohort of BRAF, HRAS, KRAS, and NRAS mutations. The results demonstrated a significant association with disease-free interval (DFI), as shown in Fig. 5. After adjusting for age and stages II and III, the association analyses were also carried out, and p-values less than 0.05 were statistically significant. The outcomes indicate almost substantial correlations among the cohort of BRAF/RAS mutations and age and stages without considering the obtained DEGs. Besides, the downregulation of GADD45B and FGFR2 directly correlates with the united cohort of BRAF/RAS mutations (p-value = 0.046 and 0.01, respectively). However, the correlation between age and BRAF/RAS mutations is insignificant for IGF2, IL1RAP, and PRICKLE1.

Fig. 4.

Frequency of BRAF-like and RAS-like mutations in TCGA-THCA using CVCDAP oncoplot tool.

Fig. 5.

Forest presentation of the 19 identified genes involved in the PTC correlated with clinical attributes (i.e., age and cancer stage) regarding the hazard ratio.

According to Table 2, fifteen out of nineteen DEGs were significant in extrathyroidal extension. Moreover, considering the pathological TNM stages of tumors, three, four, and none of the DEGs were identified as statistically significant for T, N, and M tumor stages.

Table 2.

Associations of identified DEGs with extrathyroidal extension and tumor pathological TNM stages with significant Q values less than 0.3 (in green color).

Gene Symbol	PATHOLOGY_T_STAGE (Q)	PATHOLOGY_N_STAGE (Q)	PATHOLOGY_M_STAGE (Q)	EXTRATHYROIDAL_EXTENSION (Q)
KIT	0.000901	3.08E-05	0.876	0.000897
PDGFRA	0.461	6.48E-05	0.879	0.000433
IGF2	0.224	0.00716	1	0.00329
FGF7	0.424	0.688	0.501	0.523
GADD45B	0.219	0.895	0.761	0.681
HSPA1B	0.139	0.108	0.769	0.0102
FGFR2	0.00104	0.00383	0.692	0.000726
RPS6KA5	0.00969	0.392	0.559	0.0794
IL1RAP	0.0365	3.30E-09	0.905	0.000471
DUSP5	0.00111	5.79E-10	0.886	5.59E-06
ERBB3	3.66E-05	1.47E-11	0.927	4.29E-07
DUSP4	0.00885	4.31E-06	0.777	0.00287
RAC2	0.256	4.17E-07	0.695	0.00791
MAPK13	0.00367	0.000264	0.607	5.04E-06
TCF7L1	0.00014	4.27E-08	0.996	1.36E-06
CCND1	0.346	0.549	0.992	0.351
PRICKLE1	0.0953	1.31E-07	0.9	0.00273
HEY2	0.671	0.0063	0.663	0.623
DTX4	0.163	2.68E-06	0.925	0.000507

The findings indicate that these biomarkers (19 differentially expressed genes among four datasets) can distinguish between groups of people at risk due to variations in their gene expression levels. Despite this, the p-value of the risk group separation and the concordance index were all statistically significant and higher (CI > 0.5) for these biomarkers (i.e., CI = 0.781, p-value = 0.0169554). The risk model formula (1) based on high and low-risk categories for nineteen DEGs is shown below:

$$PI=-0.69765\times KIT+0.33186\times PDGFRA+0.69527\times IGF2+1.44974\times FGF7-0.88824\times GADD45B+0.81813\times HSPA1B+0.92940\times FGFR2-0.32-16\times RPS6KA5+1.33539\times IL1RAP+0.04208\times DUSP5-0.64415\times ERBB3+0.45028\times DUSP4-0.65345\times RAC2+5.21839\times MAPK13+0.39954\times TCF7L1-0.95618\times CCND1-0.72011\times PRICKLE1-0.83411\times HEY2-1.32696\times DTX4$$ (1)

4. Discussion

The detailed information on the Notch, Wnt, and MAPK signaling pathways involved in the PTC at the cellular level and their corresponding identified genes are displayed in Fig. 6. Additionally, 19 significant gene biomarkers are comprehensively discussed regarding their critical roles, and gene crosstalks according to the three signaling pathways are provided in the following subsections. The required practical information and data are gathered to further validate these genes by quantitatively measuring the target gene expression levels (see Table 1 for more details).

Fig. 6.

Schematic illustration of the main signaling pathways and the associated significant genes involved in the PTC at the cellular level. The downregulated and upregulated genes are presented in red and green colors. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

4.1. Notch signaling pathway

Only one gene (DTX4) was important in the Notch signaling pathway. It is described that five Dx-related proteins containing highly conserved C-terminal E3 ubiquitin ligase and interacting with the Notch signaling pathway in mammalians include DTX1, DTX2, DTX3, DTX3L, DTX4, among which, DTX3 and DTX3L don't have the N terminus region [63,64]. The DTX4, a membrane-associated positive regulator of Notch signaling, involves early Notch activation and ubiquitylation [63]. Zhang, Liu, Li, Peng, Li and Li [34] considered the critical role of DTX4 upregulation. A computational approach was conducted on microarray datasets of PTC samples, and finally, the significant genes were validated through experimental tests.

Researchers utilized the PockDrug-Server in conjunction with the PPI network to find genes that may target thyroid cancer therapy. They searched TCGA thyroid cancer mRNA profiles and DNA methylation levels. Three genes, HEY2, TNIK, and LRP4, were selected for further analysis, considering the interactions with significant hub genes. This mechanism, when engaged, slows the spread of medullary thyroid carcinoma by targeting HEY2. To conclude, HEY2, TNIK, and LRP4 may have a role in developing thyroid cancer [36]. Thyroid tumors had an increased level of the NOTCH downstream effectors HEY2, identified as PROX1 negative regulators [65], with HEY2 showing combined statistical significance. Previous investigations have shown that the NOTCH pathway is active in thyroid malignancies [[66], [67], [68], [69]], supporting our result. Earlier experiments using thyroid gene expression profiling have shown a valid increase in HEY2 expression. Also, the qRT-PCR has confirmed that the HEY2 gene was overexpressed in the clinical samples.

4.2. Wnt signaling pathway

We identified four critical genes for the Wnt signaling pathway, including CCND1, TCF7L1, SFRP1, and SERPINF1. Among various proteins involved in the cell cycle process, the cyclin family is a highly conserved protein whose levels fluctuate rapidly in the four phases: G1, S, G2, and M. In these family members, the overexpression of CCND1 plays a functional role in the initiation and development of various human cancers and tumorigenesis via encoding the CCND1 protein [[70], [71], [72]], and the Involvement of oncogenes including CCND1 and catenin beta 1 (CTNNB1) has been well evidenced in the PTC [37,71]. Lymphoid enhancer factor/T cell factor (LEF/TCF), as a group of transcription factor proteins, binds specifically and highly to conserved regions of DNA. TCF7L1, a well-known member of the LEF/TCF family interacting with the Wnt/β-catenin signaling pathway, is associated with poor prognosis by overexpression in some high-grade cancers such as breast and colorectal [73]. In the case of PTC, Ray et al. [41] and Choi et al. [35] concluded that TCF7L1 had been strongly downregulated in tumor tissues.

PRICKLE1, a candidate gene expected to regulate PTC pathogenesis, was active in a recent study. Two genes, CCND1 and PRICKLE1, have been linked to PTC via strong correlation coefficients [40]. A key molecule in the Wnt signaling pathway is PRICKLE1, which regulates the proliferation and migration of tumor cells [[74], [75], [76], [77]]. Additionally, qRT-PCR and immunohistochemistry (IHC) showed that PRICKLE1 was overexpressed in PTC samples.

4.3. MAPK signaling pathway

The 19 genes have been determined to be critical in the MAPK signaling pathway. Interleukin 1 receptor accessory protein (IL1RAP) or IL1R3, as a co-receptor of IL1R1, plays a vital role in the transmission function of IL-1 signaling and is accounted as a biomarker for chronic myeloid leukemia stem cells [78]. Although IL1RAP has not been labeled for the biomarkers list in the TC, its involvement in acute myelogenous leukemia stem cells is reported elsewhere [78]. However, the significant upregulation of IL1RAP gene expression level in PTC disease has been identified in several reports [34,42,79].

It has been reported that dual-specificity phosphatase (DUSP) proteins mitigate the extracellular signal-regulated kinase (ERK) pathway with a slow growth rate in the PTC. In this way, DUSP4, known as MAPK phosphatase two, induces a nuclear phosphatase in regulating cell proliferation and differentiation by suppressing ERK1/2, P38, and c-Jun N-terminal kinases (JNK). In other types of cancers, such as breast, glioma, and lung, the dysregulation of DUSP4 is associated with carcinogenesis [80]. In the case of PTC disease, elevated expression of DUSP4 has been previously studied. For example, Delys et al. [45], Lee et al. [44], and Ma et al. [43] concluded that DUSP4, a PTC biomarker, is an upregulated gene. DUSP5 is also the target of interest among DUSP proteins for the scientific cancer community. DUSP5 is one of the most critical genes with a high rate prognosis in patients with paclitaxel resistance to basal-like breast cancer (BLBC). Only Delys et al. [45] reported that the expression level of DUSP5 has been significantly increased among other differentially expressed genes.

Erb-b2 receptor tyrosine kinase 3 (ERBB3), sometimes referred to as the epidermal growth factor receptor (EGFR), is one of the four members in the ERBB family, and its role has not been well studied in human-related cancers [81]. Several reports have evidenced overexpression of ERBB3 in PTC that has also been confirmed independently by the TCGA database [48,82,83]. Moreover, the Involvement of ERBB3 overexpression in thyroid cancer in those significantly mutated by BRAF^V600E showed treatment resistance with its corresponding inhibitors (for example, RAF/MEK) [84]. Previous studies investigated the role of c-KIT (KIT) or CD117 as a transmembrane tyrosine kinase receptor in the PTC [[50], [51], [52]]. It has been shown that downregulation of the c-KIT receptor is related to developing malignant tumors through activation of the signal transduction cascades using immunocytochemical tests [50].

For over two decades, the platelet-derived growth factor receptor (PDGF) family consisting of four polypeptide ligands, i.e., PDGFA, B, C, and D, has been frequently investigated [54]. In a publication reported by Chen et al. [53], the authors concluded that PDGFA and PDGFRA are upregulated in both FTC and PTC cell lines, suggesting a crucial role of PDGFA in thyroid cell carcinogenesis. In another study, the authors have proposed clinical evidence for the downregulation of PDGFA in more than 500 patients with the possible capability of being regarded as PTC biomarkers. Insulin-like growth factor 2 (IGF2), with therapeutic potential in wound healing and epigenetically similar expression patterns in different body tissues, also plays a pivotal role in developing various cancer types. IGF2 activity can regulate its binding to receptors associated with varying expression levels, frequently disrupted in cancer [85]. There is only one study showing the Involvement of IGF2 in the pathogenesis of PTC where the same IGF2 downregulation was observed for using cortisol, resulting in the decline of bone mass [58,86]. RPS6KA5, also known as MSK1/MSPK, is an essential substrate of the MAPK-activated protein kinase family, which regulates protein synthesis, cell cycle, and migration in response to hormones [87]. In colorectal cancer, RPS6KA5 upregulation may not be accounted for in the accurate diagnosis of the disease [88]. A recent study on PTC revealed that the downregulation of RPS6KA5 may be an important biomarker in developing this type of TC [59]. GADD45B is a member of the GADD45 family involved in cell proliferation induced against stress signals, including inflammatory cytokines and mitogen stimulation [89]. Dysregulation of the GADD45B gene have been reported in some cancer types related to Homo sapiens organism in which the tissues of lymphoma, thyroid, breast, cervical, lung, and esophageal are get involved [[90], [91], [92], [93], [94]]; moreover, the same situation can be imagined for PTC as well [60].

FGFR2 is crucial in different cellular processes through ligand binding, like cell division, growth, differentiation, embryonic development, tissue repair, and angiogenesis. According to the current research findings, we speculated that the downregulation of FGFR2, also known as CD332, is potentially a significantly differentially expressed gene between PTC and normal tissues. Its role in the PTC has not been well understood. However, the high expression of FGFR2 in different human cancers, such as lung and breast cancers, has been reported elsewhere. Finally, Fu et al. confirmed the FGFR2 as a downregulated target of miR-1266 using quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) and Western blot procedures [62]. FGF7, as a source of mitogen mesenchyme, acts on FGFR2 and induces the proliferation, differentiation, and migration of epithelial cells. It occurs upon FGFS binding through the MAPK signaling pathway activated during wound and mucosal healing [95]. In other cases of human cancers, such as colorectal cancer, the contribution of FGF7 and FGFR2 signaling remains unclear [96]. Among the list of genes Cong et al. [56] identified in PTC, FGF7 downregulation was evidenced as a confirmation of the outcomes in our study.

As an example, it has been shown that the small GTPase, RAC2, has a role in integrin and immune receptor signaling, polarization to M2 macrophages, and ROS production during host defense [[97], [98], [99]]. It was discovered that animals like mice lacking the RAC2 gene had significantly reduced tumor development, angiogenesis, and metastasis. RAC2 was shown to be downregulated in a PTC with/without HT RNA-seq data set study, indicating that the genes encoding these proteins are affected by ROS. Moreover, It was discovered that the tumor tissues had a reduced level of RAC2 expression [47].

MAPK13 is one distinct gene of the four encoded p38 MAPKs, attracting fewer research studies [100,101]. They also included genes such as MAPK13 and DUSP5 from the MAPK pathway in the case of BRAF-related PTC [102].

In some cancer types, including melanoma, pancreatic cancer, and papillary thyroid cancer, HSPA1B, a member of the HSP70 network (Heat shock proteins), was one of the genes expressed in predominantly plasma and serum samples using the ultracentrifugation (UC) isolation method [[103], [104], [105]]. Considering the HSPA1B target gene, several micro-RNAs such as miR-495–3p, miR-615–3p, miR-326, miR-331–3p, miR-130b-5p, miR-15a-5p and miR-125b-5p were involved in samples of plasma extracellular vesicles derived from cancer patients (e.g., lung cancer, clear cell renal cell carcinoma, prostate cancer, and diffuse large B-cell lymphoma) [[106], [107], [108], [109]].

4.4. BRAF-like and RAS-like mutations

Generally speaking, the association between BRAF mutation and age was found in the literature without any significant relationship among the other clinical features. Also, no clinical attributes were reported in PTC patients regarding RAS mutation [110]. In another study, three out of twenty-four PTC patients harbored the BRAF mutation H/K-RAS mutations, whereas the others lacked somatic mutation [111]. This study shows that the union cohort of BRAF/RAS-like mutations is almost related to age and clinical attributes based on the identified DEGs, which aligns with the literature findings.

4.5. Clinical features associated with ETE

Although ETE has been recognized as a significant pathogenic characteristic indicating negative consequences, very little research has focused on its molecular basis. The combined analysis of genomic and clinical data revealed substantial genetic components of PTC associated with ETE in this research. In a study, the downregulation of the WNT4 gene was directly correlated with ETE [112]; however, in this research, except for four genes (i.e., FGF7, GADD45B, CCND1, and HEY2) from Wnt, MAPK, and Notch signaling pathways, the other remaining fifteen significant DEGs were significantly associated with ETE.

4.6. Target therapy, resistance, and dynamic changes of mutation status

Integrating current targeted therapy and resistance mechanisms in PTC involving MAPK, Notch, and Wnt signaling pathways is crucial for improving clinical outcomes. The BRAFV600E mutation, found in a significant percentage of PTCs, presents an opportunity for targeted drug therapy [113]. However, resistance mechanisms against BRAF-targeted treatments have been identified, emphasizing the need for a comprehensive understanding of resistance mechanisms [114]. Additionally, the stimulation of the MAPK pathway, particularly the activation of MAPK signaling, is actively explored as a therapeutic target for thyroid cancer [115]. The importance of targeting the MAPK signaling pathway, which includes mutations in RET, RAS, and BRAF, is further emphasized because these genetic changes are often seen in PTC [116]. Thyroid cancer is not the only kind of cancer in which Notch signaling has been shown to have a role; papillary carcinomas have also been linked to this pathway [117]. Furthermore, a correlation exists between the methylation of DACT2 and the stimulation of the Wnt signaling pathway, which facilitates the spread of PTC [118]. To overcome resistance and maximize the effectiveness of targeted treatment, new studies have shown that combined targeting of MAPK and PI3K pathways may enable the re-differentiation of Braf-mutated thyroid cancer organoids [119]. Furthermore, adenyl cyclase activators have shown growth suppression of thyroid cancer cells, indicating the potential for alternative targeted therapeutic approaches [120]. These findings underscore the importance of exploring diverse targeted therapy options to overcome resistance and improve treatment efficacy. Clinical progress in PTC requires more investigation of the interplay between the MAPK, Notch, and Wnt signaling pathways and the integration of current targeted therapies. To improve the efficacy of treatment options and address resistance mechanisms in PTC, targeting the revealed genetic abnormalities and signaling networks may be fruitful, while also investigating innovative therapeutic approaches.

On the other hand, the dynamic changes in the mutational status and consequences of the signal transmission chain in PTC are critical considerations in understanding the disease progression and developing effective therapeutic strategies. The activation of MEK and ERK is a key contributor to the oncogenic phenotype, and mutations in RAS and Raf, which are upstream pathway components, are common and include the BRAFV600E mutant [121]. These mutations, along with genetic alterations like BRAF and RET/PTC rearrangement, are responsible for PTC's onset, progression, and dedifferentiation, primarily through the activation of the MAPK and PI3K signaling cascades [122]. Moreover, an essential research subject is the impact of MAPK activation on additional signaling pathways implicated in oncogenic transformation, including Notch. It is vital for comprehending the dynamic alterations in the mutational status of PTC and the repercussions of the signal transmission chain [123]. Also, many types of cancer have genetic mutations that change signaling pathways. For example, mutated BRAF is common in thyroid cancer. It shows how important it is to understand how mutational status changes and how those changes affect signaling pathways [113]. Also, knowing how the BRAFV600E mutation affects the disease's signaling cascade is crucial since it is the most common genetic change in PTC [124]. Therefore, integrating the knowledge of dynamic changes in mutational status and the consequences of the signal transmission chain is essential for advancing the understanding and management of PTC.

4.7. Limitations and projections of a meta-analysis on GEO datasets in PTC disease

Meta-analyses utilizing Gene Expression Omnibus (GEO) datasets have provided valuable insights into the molecular landscape of PTC. However, several limitations and future projections should be considered to enhance the utility and reliability of such analyses.

Limitations.

• 1.Data Heterogeneity: Heterogeneity may be introduced into the meta-analysis due to including several GEO datasets with varying sample sizes, platforms, and experimental procedures.
• 2.Data Quality: The reliability and repeatability of the meta-analysis findings may be affected by differences in data quality between GEO datasets, such as batch effects and technological biases.
• 3.Sample Size and Power: Statistical power and generalizability of results may be compromised due to small sample numbers in certain GEO datasets.
• 4.Publication Bias: Publication bias is a potential issue for meta-analyses that depend on published papers within GEO databases since non-significant or negative findings may be underestimated.
• 5.Data Integration Challenges: Data consistency and comparability are difficult to guarantee due to the difficulties in integrating and harmonizing data from disparate GEO databases.

Projections.

• 1.Standardized Protocols: To improve data harmonization and reduce technical heterogeneity, future meta-analyses should emphasize the implementation of standardized techniques for data preparation, standardization, and quality control.

• 2.Advanced Statistical Methods: More refined insights into the molecular landscape of PTC may be gained by using sophisticated statistical approaches like meta-regression and network meta-analysis to overcome data heterogeneity.
• 3.External Validation: To increase the robustness and trustworthiness of discovered molecular markers, future meta-analyses are anticipated to use external validation cohorts to validate results from GEO datasets.
• 4.Longitudinal Studies: Dynamic variations in gene expression patterns are related to PTC development and recurrence and may be better understood by longitudinal analysis using longitudinal GEO datasets.
• 5.Multi-Omics Integration: By combining GEO gene expression data with other omics information (such as proteomics and epigenomics), we may get a more complete picture of PTC etiology and better pinpoint possible treatment targets.

Although meta-analyses of GEO datasets have made important contributions to our molecular knowledge of PTC, it will be crucial to advance such studies' reliability and translational value if the aforementioned constraints are overcome and future forecasts are embraced.

5. Conclusion

We have compiled four gene expression profiles of PTC and normal tissues. We performed a meta-analysis, ultimately yielding the list of essential genes within this disease's three MAPK, Wnt, and Notch signaling pathways. All 19 significant genes may engage in all stages of the development and genesis, advancement, and spread of PTC. The research results are proposed as a list of molecular markers and pathways for further investigation. Moreover, the correlations between DEGs and clinical features were demonstrated. Possible associations among BRAF/RAS-like mutations in PTC and the significant DEGs were investigated. In addition to the present meta-analysis study, other corroborative experimental research is required to ensure the correct functions of the current computational and statistical findings; however, all nineteen panels of identified significant genes were evidenced through experimental literature studies and a robust model with a reasonably good concordance index.

Ethical Approval and Consent to participate

Not applicable.

Consent for publication

Not applicable.

Availability of supporting data

The datasets used for the current study are publicly available in the NCBI-GEO repository, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE29265 https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE97001

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE138198

https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE3678.

Funding

No financial supports were available for this research.

Authors' information

Biotechnology Research Center, and Kidney Research Center, Tabriz University of Medical Sciences, Tabriz, Iran.

CRediT authorship contribution statement

Elham Amjad: Data curation, Formal analysis, Resources, Writing – original draft, Investigation, Validation, Visualization. Solmaz Asnaashari: Methodology, Writing – original draft, Formal analysis, Validation. Ali Jahanban-Esfahlan: Formal analysis, Investigation, Project administration, Writing – original draft, Writing – review & editing, Supervision. Babak Sokouti: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper