Synergistic role of molecular imaging and genomics in thyroid cancer management: from diagnosis to personalized treatment

Abstract

The management of thyroid cancer has advanced through the integration of molecular imaging, genomic profiling, and bioinformatics, supporting precision oncology across diverse subtypes. Although radioiodine imaging and therapy remain standard for differentiated thyroid cancers, efficacy can be limited in refractory cases due to heterogeneous iodine uptake. Advances including dosimetry-guided radioiodine therapy, hybrid imaging modalities (SPECT/CT, PET/CT, PET/MRI), and novel radiotracers (FDG, NaF) may improve diagnostic accuracy and therapeutic decision-making, particularly for aggressive tumors. Genomic profiling has reshaped tumor classification, prognosis, and therapy selection by identifying key alterations in BRAF, RAS, RET/PTC, PAX8/PPARγ, TERT, TP53, ALK, and NTRK. Targeted inhibitors against BRAF, RET, and NTRK demonstrate clinical benefit, while synergistic mutations such as BRAF V600E with TERT promoter highlight tumor complexity and may support investigation of combination strategies. Recent studies have utilized bioinformatics and multi-omics analyses for high-throughput mutation mapping, pathway analysis, and biomarker discovery, leveraging patient genomic data from TCGA and cBioPortal with visualization via R packages such as ggplot2, dplyr, ggrepel, and tidyr. Pathway analyses using GO, KEGG, and Reactome databases revealed involvement in extracellular matrix organization, cell junction assembly, PI3K-Akt signaling, and collagen formation. Integrating insights from molecular imaging, genomics, and computational biology enhances understanding of tumor biology, supports risk stratification, and informs the design of personalized therapies. Future directions include machine learning–driven data integration and expanded clinical application of next-generation sequencing and hybrid imaging to improve patient stratification and management.

Introduction

The management of thyroid cancer has undergone significant evolution over recent decades, primarily driven by advances in molecular imaging and genomic technologies [1, 2]. Our deepened understanding of the molecular basis underlying various thyroid cancer subtypes has facilitated improvements in diagnosis, treatment planning, and personalized patient care. Papillary thyroid carcinoma (PTC), the most common form, typically exhibits a favorable prognosis, whereas follicular thyroid carcinoma (FTC) and medullary thyroid carcinoma (MTC) demonstrate varying degrees of aggressiveness and metastatic potential. In stark contrast, anaplastic thyroid carcinoma (ATC) is characterized by rapid progression and poor clinical outcome [3]. Molecular imaging modalities such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT) provide comprehensive assessments of tumor metabolic activity and functional characteristics, enabling early malignancy detection and evaluation of therapeutic response [4, 5]. Concurrently, genomic profiling has become integral to precision oncology in thyroid cancer, allowing the identification of key genetic alterations, most notably alterations in BRAF, RAS (NRAS, HRAS, KRAS), the RET gene, which can guide targeted therapy decisions [2, 6]. The integration of molecular imaging and genomics enhances the discrimination between indolent and aggressive tumor forms, optimizing individualized management strategies. This synergy holds promise for improving patient survival and quality of life by enabling more accurate diagnostics and tailored treatments [2, 7].

Complementary bioinformatics and multi-omics analyses, leveraging high-throughput datasets and pathway databases such as GO, KEGG, and Reactome, enable systematic mapping of genetic alterations, network analysis, and biomarker discovery, supporting integration of molecular insights with clinical management [8]. The convergence of molecular imaging, genomics, and computational approaches enhances discrimination between indolent and aggressive tumor forms, optimizing individualized management strategies. Despite substantial advances in molecular imaging and genomics, their integrated role in thyroid cancer remains incompletely understood. This narrative review examines the complementary roles of molecular imaging, genomics, and bioinformatics in thyroid cancer and is supported by secondary bioinformatics analyses using TCGA-derived data from cBioPortal and GEPIA3, including enrichment, transcription factor, and PPI network analyses, to provide additional translational insights.

Methods

Literature search strategy

This study was conducted as a narrative review with integrated secondary bioinformatics analyses to provide a comprehensive and translational perspective on the synergistic role of molecular imaging and genomics in thyroid cancer management. A structured literature search was performed in PubMed and Google Scholar to identify relevant English-language publications. The search strategy combined the following key terms: thyroid cancer, molecular imaging, positron emission tomography (PET), single-photon emission computed tomography (SPECT), PET/CT, PET/MRI, radioiodine, genomics, molecular biomarkers, targeted therapy, precision medicine, radiogenomics, and bioinformatics.

We included original research articles, clinical studies, systematic reviews, meta-analyses, and high-quality narrative reviews that addressed the integration of molecular imaging and genomics in thyroid cancer. Additional articles were identified through manual screening of reference lists from eligible publications. Non-English studies, conference abstracts lacking sufficient methodological details, duplicate publications, and articles not directly focused on the interplay between molecular imaging and genomics were excluded. The selected literature was critically appraised and synthesized with particular emphasis on distinguishing established clinical applications from emerging and investigational approaches.

Secondary bioinformatics analysis

Secondary bioinformatics analyses were conducted to characterize genomic alterations, biological functions, and prognostic associations of thyroid cancer-related genes using publicly available datasets and established computational resources. Genomic alteration profiles of selected thyroid cancer-associated genes, including BRAF, RAS (NRAS, HRAS, and KRAS), RET, PAX8/PPARG, TERT, TP53, ALK, NTRK1, and NTRK3, were retrieved from the TCGA Pan Cancer Atlas Thyroid Carcinoma (THCA) cohort via the cBioPortal platform. To determine the functional relevance of altered genes, enrichment analyses were performed using Enrichr in the R statistical environment. Functional annotation and pathway enrichment were assessed using the Gene Ontology (GO) Biological Process, KEGG, Reactome, and TRANSFAC/JASPAR gene-set libraries. The prognostic significance of candidate genes was evaluated using the GEPIA3 platform based on TCGA-THCA transcriptomic data. Genes demonstrating both genomic alterations and prognostic relevance were identified through comparative analysis of altered gene sets and survival-associated genes using the jvenn web tool. Protein–protein interaction (PPI) networks were constructed using the STRING database (version 12.0) to investigate potential functional relationships among candidate genes. Network generation was performed using a medium-confidence interaction score (≥ 0.400) and integrated evidence from experimental studies, curated databases, co-expression analysis, text mining, gene neighborhood, gene fusion, and co-occurrence data. The generated networks were subsequently visualized and analyzed using Cytoscape software (version 3.10.3). Network visualization was optimized with the yFiles Organic Layout algorithm, while key interacting genes were identified using the CytoHubba plugin based on degree centrality. Functional associations among selected genes were further explored using the GeneMANIA plugin. Detailed information regarding software versions, analytical parameters, filtering criteria, and statistical thresholds applied in each bioinformatics analysis is provided in the corresponding sections of the manuscript.

Radioiodine imaging and therapy: from tradition to innovation

Radioactive iodine (RAI) or Radioiodine therapy has been a foundational treatment modality for differentiated thyroid cancers, particularly papillary and follicular types. This approach leverages the unique ability of thyroid tissue and differentiated thyroid cancer cells to uptake iodine, enabling selective targeting through radioactive iodine-131 administration [9]. Traditionally, RAI is used to ablate residual thyroid tissue after surgery and to treat metastatic disease. Imaging with RAI scans aids clinicians in assessing iodine uptake and mapping disease extent to guide therapy effectiveness. However, challenges persist, including heterogeneous and often reduced iodine uptake in some tumors, leading to treatment resistance and suboptimal outcomes [10]. In response, advances such as dosimetry-guided RAI therapy, which adjusts the radioactive dose to individual patient and tumor characteristics, have emerged to enhance efficacy while minimizing side effects. Moreover, integration of hybrid imaging modalities like SPECT/CT and PET/CT improves tumor localization and therapy monitoring, offering more precise disease assessment and treatment adaptation [11, 12]. These innovations mark a transition from traditional to more personalized radioiodine-based management of thyroid cancer.

Advanced molecular imaging beyond iodine

As understanding of thyroid cancer biology deepens, molecular imaging techniques beyond iodine-based methods have gained prominence. PET/CT utilizes specific radiotracers to target different disease processes. 18 F-fluorodeoxyglucose (FDG) visualizes heightened tumor metabolic activity, while 18 F-sodium fluoride (18 F-NaF) targets areas of osteoblastic activity to detect bone metastases. This ability to find non-iodine-avid disease is crucial, as it allows clinicians to identify the aggressive or treatment-resistant forms of thyroid cancer that conventional RAI imaging cannot detect [1, 13]. These functional imaging approaches allow for enhanced tumor characterization, improved staging, and more informed therapeutic decisions. Targeted molecular imaging agents that bind specific tumor markers, including somatostatin analogs for neuroendocrine thyroid tumors, offer additional diagnostic precision. Furthermore, hybrid imaging technologies such as PET/MRI provide simultaneous anatomical and functional data, facilitating comprehensive assessment of tumor progression and therapeutic response [14, 15]. Together, these advances provide a multifaceted view of the disease. These detailed diagnostic findings enable treatment plans to be precisely individualized for each patient, which ultimately leads to better clinical outcomes.

Genomic pathways in thyroid cancer and their impact on management

Fundamental role of genetic alterations in thyroid carcinogenesis

Carcinogenesis in thyroid tissue is a multistep process characterized by genetic and epigenetic alterations that disrupt normal cellular homeostasis. Central to thyroid tumor development are mutations in key genes such as BRAF, RAS (NRAS, HRAS, KRAS), and RET, as well as chromosomal rearrangements and DNA methylation changes. These genetic modifications activate critical signaling pathways—including the MAPK and PI3K/AKT/mTOR cascades—that regulate cell proliferation, differentiation, and survival [16–20]. The interplay of environmental factors like ionizing radiation with these genetic changes further promotes tumor initiation and progression. Advances in next-generation sequencing (NGS) have enhanced the precision of molecular profiling, allowing for the identification of common and rare mutations and gene fusions, which serve as diagnostic, prognostic, and therapeutic biomarkers. Importantly, the clinical evidence supporting the implementation of these genomic biomarkers is not uniform and varies according to the specific alteration, tumor subtype, and disease context. As summarized in Table 1, BRAFV600E mutations, RET rearrangements, ALK rearrangements, and NTRK fusions represent clinically actionable alterations with direct therapeutic implications, particularly in advanced or treatment-refractory thyroid cancers, as they can guide the selection of approved or clinically validated targeted therapies, including BRAF, RET, and TRK inhibitors. In contrast, RAS mutations, PAX8::PPARG rearrangements, TERT promoter mutations, and TP53 alterations are primarily used for molecular classification, diagnostic refinement, and risk stratification, while their predictive value remains dependent on histologic subtype, disease stage, and co-occurring molecular alterations [21–25]. This comprehensive overview underscores the heterogeneity of thyroid cancers and the critical role of molecular profiling in guiding personalized management.

Table 1.

Evidence-based Summary of reported genomic alterations in thyroid cancer subtypes, including prevalence and clinical relevance

Cancer Type	Altered Gene Name	Type of Alteration	Approximate Prevalence (%)	Impact on prognosis	Impact on treatment	Relevant targeted
PTC	BRAF	Mutation-V600E	40–50%	Higher recurrence risk	Almost resistance to RAI, Candidate for targeted therapies	Dabrafenib, aggressive in PTC Vemurafenib, Selumetinib, (off-label for BRAF)
PTC, FTC, HTC	RAS	Mutation-NRAS, HRAS, KRAS	10–20%	RAS (Mutations)	Almost resistance to RAI, Responds to MEK inhibitors	Dabrafenib, aggressive in PTC Vemurafenib, Selumetinib, (off-label for BRAF)
PTC	RET/PTC	Rearrangement	5–10%	More aggressive than PTC without this alteration	Candidate for RET inhibitor therapies	Selpercatinib Pralsetinib
FTC	PAX8/PPARY	Rearrangement	30–40%	Relatively good prognosis in FTC	Usually responsive to RAI	NA
PTC, FTC, ATC	TERT	Promoter Mutation	10–20% (higher in advanced cases)	Higher risk of recurrence and mortality, highly aggressive	Associated with RAI resistance A marker for advanced disease	Under investigation
ATC, PDTC	TP53	Mutation	ATC: 60–80% PDTC: 20–30%	TP53 (Mutations)	Associated with highly aggressive cancers (ATC)	NA
PTC	ALK	Rearrangement	< 1%	NA	Candidate for ALK inhibitor therapies	Alectinib Lorlatinib
PTC	NTRK	Rearrangement	< 1%	NA	Candidate for NTRK inhibitor therapies	Larotrectinib Entrectinib

Activation of the MAPK pathway and the role of BRAF mutations

The mitogen-activated protein kinase (MAPK) pathway is a central driver of thyroid tumorigenesis and regulates cellular proliferation, differentiation, survival, and migration through the RAF–MEK–ERK cascade [26–28]. Among MAPK-related alterations, BRAF V600E is one of the most frequent driver events in papillary thyroid carcinoma (PTC), occurring in approximately 45% of cases although its prevalence varies across populations and histologic variants [29]. This mutation induces constitutive activation of BRAF kinase by mimicking phosphorylation at residues T599/S602, leading to persistent MAPK pathway signaling independent of upstream stimuli [30]. Such activation drives uncontrolled cellular proliferation and confers an aggressive phenotype, often associated with resistance to RAI therapy. Other oncogenic events that activate MAPK signaling include mutations in RAS-family genes (NRAS, HRAS, and KRAS) and rearrangements involving RET. However, these alterations define biologically distinct tumor groups and do not have equivalent prognostic or therapeutic implications [31]. Importantly, co-occurrence of BRAF V600E with TERT promoter mutations has been shown to synergistically increase tumor aggressiveness, recurrence risk, and mortality [32, 33]. Clinically, the therapeutic relevance of MAPK-pathway alterations varies by tumor subtype and treatment setting. BRAF-directed therapy, often in combination with MEK inhibition, has an established role in selected patients with BRAF V600E-mutant advanced thyroid cancers, particularly anaplastic thyroid carcinoma. In contrast, MEK inhibition as a redifferentiation strategy to restore radioiodine uptake remains more context-dependent and requires careful patient selection. Therefore, MAPK alterations should not be treated as a uniform therapeutic category across all thyroid cancer subtypes [34]. Figure 1 summarizes the principal components of MAPK signaling and the major genomic alterations implicated in thyroid carcinogens.

Fig. 1.

Schematic representation of MAPK signaling pathway and associated mutations in thyroid cancer. The figure highlights frequently mutated genes including BRAF, RAS (NRAS, HRAS, KRAS), and RET, which lead to constitutive activation of the pathway, promoting tumor initiation, progression, and radioiodine resistance. These alterations serve as important therapeutic targets for kinase inhibitors in precision medicine

Alterations in the PI3K/AKT/mTOR pathway

The PI3K/AKT/mTOR signaling axis regulates vital cellular functions, including metabolism, growth, differentiation, and survival. Aberrations in this pathway through mutations in PIK3CA, AKT, or loss of the tumor suppressor PTEN promote oncogenic signaling and tumor progression, especially in aggressive thyroid cancers like ATC [35–38]. Notably, PIK3CA mutations and copy number gains are more prevalent in FTC, ATC, and frequently co-occur with MAPK pathway mutations, suggesting complex molecular crosstalk contributing to tumorigenesis and therapy resistance [39, 40]. Mouse model studies indicate that isolated PTEN loss induces thyrocyte proliferation but is insufficient for malignant transformation, emphasizing the multigenic nature of thyroid carcinogenesis [41]. Schematic outline highlighting the role of PI3K/AKT/mTOR Pathway, in tumor progression in thyroid cancer, along with potential targets for imaging and genomic analysis, is shown in Fig. 2. The sequential molecular events underlying the PI3K/AKT and MAPK pathway alterations that drive the progression spectrum of thyroid tumorigenesis are illustrated in Fig. 3.

Fig. 2.

Schematic overview of major signaling pathways in thyroid cancer. The figure depicts the PI3K/AKT/mTOR, MAPK (RAS/RAF/MEK/ERK), and NF-κB pathways and their crosstalk. Activation of these cascades by growth factors and recurrent mutations (e.g., BRAF and RAS) contributes to tumorigenesis, invasion, and therapeutic resistance, highlighting opportunities for integrated molecular imaging and genomics-driven targeted therapies

Fig. 3.

Progression of thyroid cancer subtypes through PI3K–AKT and MAPK pathway alterations. This diagram illustrates the molecular evolution of thyroid cancers from normal follicular thyroid cells via dysregulation of the PI3K–AKT and MAPK (Mitogen-Activated Protein Kinase) pathways. Mutations activating PI3K–AKT (e.g., RAS, PTEN, PIK3CA) and MAPK (e.g., BRAF V600E) signaling lead to distinct tumor subtypes: follicular thyroid adenoma (FTA), follicular thyroid carcinoma (FTC), and papillary thyroid carcinoma (PTC). Progressive activation of these pathways contributes to more invasive phenotypes, including poorly differentiated thyroid carcinoma (PDTC) and anaplastic thyroid carcinoma (ATC), reflecting increasing molecular complexity and aggressiveness

TERT promoter mutations

Mutations within the promoter region of the TERT gene, commonly at hotspots C228T and C250T, lead to upregulated telomerase expression, granting tumor cells replicative immortality [42, 43]. These mutations are enriched in aggressive thyroid cancer subtypes including anaplastic thyroid carcinoma and some differentiated thyroid carcinomas, correlating with worse prognosis, higher recurrence, and decreased survival [44–46]. TERT promoter mutations often co-exist with MAPK pathway alterations (e.g., BRAF V600E), resulting in synergistic effects on tumor progression. Their detection via NGS facilitates risk stratification and may inform novel therapeutic avenues targeting telomerase activity [47–49].

Gene fusions

Chromosomal rearrangements resulting in gene fusions are hallmark features in papillary and follicular thyroid cancers. Fusions involving RET/PTC, PAX8/PPARγ, ALK, and NTRK activate oncogenic signaling pathways such as MAPK and PI3K/AKT, disrupting normal cellular differentiation and driving tumor proliferation [50, 51]. Targeted inhibitors against these fusion proteins, including selpercatinib (RET inhibitors) and larotrectinib (NTRK inhibitors), have shown promising therapeutic results, underscoring their clinical significance.

Mutational collaboration in thyroid cancer and its impact on molecular classification and treatment

Thyroid cancer progression is frequently associated with the accumulation of multiple genomic alterations, whose combined effects may exceed those of individual driver events. However, biological synergy has been established more convincingly for some alteration pairs than for others. Notably, the co-occurrence of BRAF V600E and TERT promoter mutations significantly increases cellular proliferation, invasion, metastasis, and resistance to radioiodine therapy [32, 33].

Combinations of mutations activating parallel pathways such as BRAF V600E alongside PI3K/AKT/mTOR alterations (e.g., PTEN, PIK3CA) contribute to aggressive phenotypes and therapeutic resistance. Similarly, RAS-family mutations may coexist with alterations in tumor-suppressor or differentiation-associated genes, including TP53 and DICER1, particularly in poorly differentiated or follicular-patterned thyroid tumors. These additional events are associated with loss of differentiation and adverse clinicopathologic characteristics, although their prognostic effects vary according to histologic and molecular context. These molecular interactions complicate treatment strategies, as monotherapies targeting single pathways may be insufficient. Comprehensive mutational profiling to identify synergistic mutation patterns may support the design of rational combination strategies and improving clinical outcomes [52, 53].

Convergence of genomics and imaging/therapy for precision medicine

Precision management of thyroid cancer can be enhanced by the complementary use of genomic profiling and molecular imaging, although the extent of their clinical integration varies across tumor subtypes and treatment settings. Genomic analysis uncovers critical mutations, gene fusions, and expression alterations that define tumor behavior and potential therapeutic targets. Concurrently, imaging techniques such as PET, functional magnetic resonance imaging (fMRI), and hybrid methods like PET/CT and PET/MRI provide spatial and functional information on tumor metabolism, iodine avidity, disease distribution, and response [54, 55]. Although both approaches have established clinical applications, fully integrated imaging–genomic workflows remain emerging and require further clinical validation.

Integration of molecular imaging and genomics

The integration of molecular imaging and genomic profiling has improved selected aspects of diagnosis and treatment planning of thyroid cancer. While conventional imaging modalities such as ultrasound, CT, and PET provide crucial information regarding the anatomical and functional characteristics of thyroid nodules, they are limited in their ability to definitively distinguish between benign and malignant lesions. Genomic data, on the other hand, offer a molecular-level understanding of tumor behavior and can complement imaging findings by identifying specific mutations or gene fusions that correlate with malignancy and treatment responsiveness.

In parallel with these developments, emerging electrochemical and optical biosensors, designed for highly sensitive and rapid detection of biological analytes, are gaining attention as complementary tools that can further enhance diagnostic precision, as highlighted in recent reviews on hormone-targeted biosensing platforms [56]. Combining these two approaches enables a more accurate and patient-specific risk stratification, especially in indeterminate nodules where cytology and imaging alone are insufficient. For example, a nodule with suspicious ultrasound features harboring a BRAF V600E mutation strongly suggests PTC and may guide guideline-based surgical and therapeutic decision-making. Conversely, nodules with benign imaging and absence of high-risk genetic alterations are more likely to be indolent and may qualify for active surveillance rather than immediate intervention.

One practical illustration of this synergy is in the differentiation between benign and malignant thyroid nodules. The Thyroid Imaging Reporting and Data System (TIRADS) stratifies the malignancy risk of thyroid nodules based on their ultrasound appearance [57]. Imaging features such as composition, shape, margin, echogenicity, and vascularity offer morphological clues, while the presence or absence of specific gene mutations such as BRAF, TERT, or TP53 provides strong molecular evidence for malignancy. To reduce the number of unnecessary fine-needle biopsies and thyroid surgeries, it is vital to add thyroid scintigraphy (using ^99mTc-pertechnetate or ¹²³I) to the TIRADS model [58]. The integration of these features improves diagnostic accuracy and guides personalized treatment strategies. Table 2 summarizes the distinguishing imaging [59–61] and genomic characteristics between benign and malignant thyroid nodules, highlighting how their combination supports clinical decision-making.

Table 2.

Imaging features and genetic findings of benign vs. malignant thyroid nodules

Feature	Benign Nodules (Non-Cancerous)	Malignant Nodules (Cancerous)
1.Imaging Features (e.g., Ultrasound)
Shape	Usually round or oval	Often irregular, taller than wide
Margins	Smooth	irregular spiculated or microlobulated and poorly defined
Internal composition	Commonly cystic (fluid-filled) or spongiform (multiple tiny cysts)	Frequently solid (dense tissue)
Echogenicity	Isoechoic or hyperechoic	Often hypoechoic (darker than the surrounding thyroid tissue)
Classification	May have macro calcifications (large, coarse classifications)	May have micro classifications (tiny, punctate bright spots, often with shadowing)
2. Genetic finding
Common Gene Mutations	Typically lack specific cancer-associated mutations or have mutations less strongly linked to malignancy (e.g. some RAS mutations)	BRAF V600E (highly indicative of papillary thyroid cancer) RAS (can be seen in both, but specific variants are more indicative of malignancy)
		TERT promoter (associated with aggressive disease)
		TP53 (common in anaplastic thyroid)
Gene fusions	Less commonly observed	May involve gene fusions such as RET fusions or NTRK fusions

Guiding targeted therapy selection via genomics

A comprehensive understanding of tumor genomics enables the identification of actionable mutations that guide personalized targeted therapies. For instance, the presence of the BRAF V600E mutation indicates sensitivity to BRAF inhibitors such as dabrafenib and vemurafenib, which have demonstrated significant improvements in progression-free survival [62, 63]. Similarly, detection of RET fusions allows for the application of selective RET inhibitors like selpercatinib and pralsetinib. The identification of additional molecular alterations also facilitates the design of combination regimens that can overcome drug resistance mechanisms, enhancing overall treatment efficacy and patient outcomes [64, 65].

The presence of the BRAF V600E mutation, associated with decreased RAI avidity, indicates sensitivity to BRAF inhibitors like dabrafenib and vemurafenib, often combined with MEK inhibitors (trametinib) to improve efficacy and overcome resistance [66, 67] Similarly, in medullary thyroid cancer, RET mutations are targeted with highly selective RET inhibitors such as selpercatinib and pralsetinib, which have demonstrated superior outcomes compared to multikinase inhibitors [68, 69]. Beyond these common alterations, the identification of rarer NTRK and ALK fusions provides additional actionable targets. Next-generation sequencing is therefore essential to map this genomic landscape, facilitating the choice between various targeted agents and the design of intelligent combination regimens that can prevent or overcome drug resistance mechanisms [62, 63]. This genomics-driven paradigm ensures systemic therapy is precisely aligned with the tumor’s molecular drivers, significantly enhancing treatment efficacy and patient outcomes.

Personalized treatment planning

Personalized treatment strategies consider the patient’s unique genetic, clinical, and demographic profile to optimize therapy selection, dosing, and follow-up. Ongoing monitoring potentially via liquid biopsies assessing circulating tumor DNA allows adaptive modifications of treatment in response to tumor evolution and resistance [70, 71]. The molecular characterization of thyroid nodules is essential both for clarifying uncertain biopsy results and for personalizing treatment in advanced cancers. The identification of specific genetic alterations determines the application of matched targeted therapies. This is particularly relevant in managing RAI-refractory DTC and MTC. Kinase inhibitors such as larotrectinib, vemurafenib, and lenvatinib demonstrate efficacy by targeting actionable pathways. Furthermore, genomic data are enabling new strategies to make resistant cancers sensitive to radioactive iodine again. Consequently, tumor genomics is established as the cornerstone of precision medicine in contemporary thyroid oncology [72]. Figure 4 illustrates the synergistic connection between molecular imaging, genomic data, and targeted therapy, highlighting their convergence as the foundation for optimal personalized cancer treatment.

Fig. 4.

Integrative framework for personalized treatment in thyroid cancer. This Venn diagram illustrates the synergistic convergence of three core domains in precision oncology: molecular imaging (providing functional and spatial tumor characterization), genomic profiling (revealing tumor-specific molecular alterations), and targeted therapy (directed against defined molecular drivers). The central overlapping area represents the optimal integration of these approaches, enabling highly individualized diagnostic and therapeutic strategies that improve patient outcomes

Monitoring treatment response and detecting recurrence

Effective post-treatment surveillance involves a combination of serum thyroglobulin measurements, conventional imaging, and advanced molecular diagnostics. Liquid biopsy technologies enhance early detection of residual or recurrent disease, particularly in high-risk patients requiring more intensive follow-up. Risk-adapted surveillance protocols balance the need for timely detection with avoidance of unnecessary procedures and associated burdens [73–75].

Thyroid cancer encompasses a range of histological subtypes with markedly different clinical behaviors. The majority of cases are well-differentiated forms that typically have an excellent prognosis and are amenable to curative therapy with total thyroidectomy, adjuvant radioiodine ablation using 131-I, and TSH suppression; this approach also facilitates post-operative monitoring using thyroglobulin as a sensitive biomarker. However, a minor subset of well-differentiated thyroid cancer (WDTC) displays aggressive disease, and for patients with extensive locoregional involvement, external beam radiotherapy may be a valuable adjuvant option. Medullary thyroid carcinoma (MTC) is characterized by a tendency for early metastasis and an inherent resistance to radioiodine therapy, shifting modern treatment paradigms towards molecularly targeted systemic agents. In sharp contrast, anaplastic thyroid cancer (ATC) represents one of the most lethal human malignancies, where outcomes remain poor despite aggressive multimodal therapy such as the combination of EBRT with chemotherapy, using agents such as a taxane, anthracycline, or platin. However, due to the high toxicity of this regimen, many radiation oncologists rarely employ concurrent chemoradiation, underscoring the critical need for novel targeted and personalized therapeutics [64, 76].

Side effects and risks of radioactive iodine (I-131) therapy

The administration of I-131 therapeutically results in systemic exposure to ionizing radiation, creating a risk of injury to any bodily tissue and the subsequent development of diverse tumors. Several factors determine the toxicity risk in thyroid cancer treatment, including the prescribed dose, patient-specific physiology for iodine metabolism, inherent radiosensitivity, patient sex and age, overall tumor burden, specific sites of metastasis, and the functional capacity of the cancer cells to uptake iodine [77–79]. A summary of the major clinical adverse outcomes associated with I-131 therapy is provided in Table 3.

Table 3.

Clinical adverse end points associated with radioactive iodine (I-131) therapy

Side Effect / Risk	Duration	Key Symptoms / Consequences	Contributing Factors / Notes
Common Temporary Side Effects
Swelling of Salivary Glands	3–5 days	Mild pain and swelling near the jaw.	Usually resolves on its own.
Taste Changes	Up to 3 weeks	Altered sense of taste.	Typically returns to normal.
Nausea	1–3 days	Nausea, possibly with vomiting.	Anti-nausea medication is provided and should be used as needed.
Serious & Potentially Permanent Risks
Dry Mouth & Dry Eyes	Permanent	Decreased saliva and tears, leading to dry mouth, increased tooth decay, and dry eyes.	Primary risk factors are the I-131 dose and pre-existing gland function.
Bone Marrow Damage	Varies	Increased risk of infection, bleeding, and anemia.	Risk depends on the I-131 dose and bone marrow status.
Future Secondary Cancer	Long-term	I-131 can increase the risk of a new cancer developing later in life.	Risk can be estimated using tools like the RADRAT calculator.
Gonadal Damage	Potentially Permanent	Women: Risk of early menopause and decreased fertility. Men: Risk of impotence and other issues.	Risk depends on patient age and I-131 dose.
Impaired Fertility	At least 6 months	Decreased ability to conceive or produce a healthy baby.	Pregnancy must be avoided by both female patients and partners of male patients for at least 6 months after therapy.

RADRAT: Radiation Risk Assessment Tool

External beam irradiation for thyroid carcinoma

External beam radiation therapy (EBRT) serves as an effective palliative and adjuvant treatment for unresectable or residual thyroid cancer. However, its role as an adjuvant therapy remains controversial due to a lack of prospective studies and inconsistent survival outcomes, despite evidence showing it can reduce local recurrence in WDTC and is crucial for tumors unresponsive to RAI. In aggressive ATC, EBRT is a critical component of multimodal therapy and is vital for palliative local control to prevent life-threatening complications such as asphyxiation. Nevertheless, the optimal timing and dosage for ATC are not well-established, but some evidence favoring hyperfractionated regimens (e.g., 30 Gy in 10 fractions) delivered via three-dimensional conformal radiotherapy (3D-CRT) [64, 80, 81]. Intensity-modulated radiation therapy (IMRT) has largely replaced 3D-CRT as the modern standard of care for treating the primary site and regional lymphatics. Advances in hybrid imaging with CT, MRI, SPECT and PET are crucial for defining the target volume and organs at risk, thereby enabling IMRT’s principal advantage: the delivery of highly conformal dose coverage to complex geometries while effectively sparing critical structures such as the salivary glands, spinal cord, esophagus, and larynx. A typical definitive dose for IMRT is 70 Gy delivered in 35 fractions. For small-volume disease (e.g., < 4 cm³) where irradiating a larger volume is deemed unacceptable, stereotactic body radiation therapy (SBRT) with regimens such as 50 Gy in 10 fractions or 35 Gy in 5 fractions presents a suitable alternative [64, 81–83].

Secondary computational bioinformatics analysis

Bioinformatics tools have significantly improved precision oncology by recognizing key molecular targets for personalized therapy [84]. High-throughput sequencing and multiomics technologies develop complex datasets that require powerful computational techniques for analysis and research. However, applying multiomics data information in clinical settings remains challenging. Advanced bioinformatics tools are required for discovering cancer biomarkers, supported by cloud services like Galaxy and DNAnexus, as well as single-cell analysis software such as Seurat. The integration of artificial intelligence and machine learning enhances modeling and diagnostic precision. Spatial omics technologies connect molecular signatures to tumor microenvironments, informing treatment decisions [85]. Collaborative efforts, like those from The Cancer Genome Atlas (TCGA) and cBioPortal, facilitate advancement through open data sharing and standardization [86]. Prospective work should focus on merging multiomics with innovative computational techniques to improve biomarker discovery and applications in precision treatment [87].

In this study, we used various bioinformatics tools and databases to effectively visualize data pertaining to the mutations in the genes summarized in Table 1, BRAF, RAS, RET/PTC, PAX8/PPARγ, TERT, TP53, ALK, and NTRK genes, as described below:

The ggplot2 (3.5.2) (https://cran.r-project.org/web/packages/ggplot2/index.html) [88], dplyr (1.1.4) (https://cran.r-project.org/web/packages/dplyr/index.html) [89], ggrepel (0.9.6) (https://cran.r-project.org/web/packages/ggrepel/index.html) [90], and tidyr (1.3.1) (https://cran.r-project.org/web/packages/tidyr/index.html) [91] were employed to create infographic curves. These curves illustrate data related to Table 1, which includes the prevalence of mutations in the BRAF, RAS, RET/PTC, PAX8/PPARγ, TERT, TP53, ALK, and NTRK genes across various thyroid cancer types, including PTC, FTC, HTC, and PDTC Fig. 5A. Additionally, the analysis addresses the impact of these mutations on prognosis. Figure 5B, treatment options Fig. 5C, and targeted therapy Fig. 5D for these cancer types. Together, they emphasize how molecular changes are guiding clinical management and impacting patient outcomes.

Fig. 5.

Integrated overview of key genetic alterations, prognostic implications, and therapeutic strategies in thyroid cancer. This infographic summarizes the distribution and clinical significance of major mutations across different histological subtypes (PTC, FTC, HCC, PDTC, and ATC), including BRAF, RAS (NRAS, HRAS, KRAS), RET, PAX8/PPARγ, TERT, TP53, ALK, and NTRK (NTRK1/NTRK3). (A) Frequency of key genetic alterations across thyroid cancer subtypes, (B) Prognostic implications of these mutations, (C) Targeted therapeutic strategies associated with specific molecular alterations, (D) Current treatment approaches stratified by tumor subtype and molecular profile. Collectively, these panels underscore the critical role of genomic profiling in risk stratification and precision medicine decision-making in thyroid cancer management

Genetic alteration assessment in the BRAF, RAS, RET, PAX8/PPARγ, TERT, TP53, ALK, and NTRK genes

Genetic alteration analysis was conducted using the TCGA-THCA PanCancer Atlas cohort available in cBioPortal (https://www.cbioportal.org), comprising 493 primary thyroid carcinoma samples with available somatic mutation data [92]. Genomic alteration data for predefined thyroid cancer-related genes, including BRAF, RAS (NRAS, HRAS, KRAS), RET, PAX8/PPARG, TERT, TP53, ALK, NTRK1, and NTRK3, were retrieved from the TCGA-PanCancer Atlas Studies in cBioPortal using the Mutations module in the “Quick Search” section. Only protein-coding alterations were included in the analysis. Synonymous variants were excluded, while missense, nonsense, frameshift, splice-site, in-frame insertion/deletion, copy number alterations, and gene fusion events were retained. Mutation annotations were obtained from the OncoKB and ClinVar resources available within cBioPortal. Additional variables included protein change, mutation type, variant type, copy number status, reference/variant status, disease stage according to the American Joint Committee on Cancer (AJCC) staging system, annotation, ClinVar information, and personal neoplasm cancer status. Data were retrieved from cBioPortal and no additional filtering was applied beyond the default TCGA PanCancer Atlas quality control pipeline.

Analysis of the cBioPortal THCA PanCancer Atlas dataset identified genetic alterations in BRAF, TERT, TP53, RET, KRAS, NRAS, HRAS, PAX8, ALK, NTRK1, and NTRK3. The detailed distribution of these alterations is presented in Table 4. Among the analyzed genes, BRAF exhibited the greatest diversity of alterations, with a total of 13 alterations, including two missense mutations (V600E and K601E), one in-frame deletion (P490_Q494del), and ten fusion events (BRAF-AP3B1, FAM114A2-BRAF, MACF1-BRAF, MKRN1-BRAF, ZC3HAV1-BRAF, BCL2L11-BRAF, BRAF-FAM114A2, BRAF-MACF1, BRAF-SND1, and BRAF-SUGCT). Most BRAF alterations were annotated as oncogenic or likely oncogenic and were associated with gain-of-function. Four alterations were identified in TERT, consisting of two missense mutations (R470H and S602L), one frameshift deletion (T1113Lfs*62), and one fusion event (MTMR12-TERT). All TERT alterations were classified as having unknown functional significance. Two nonsense mutations (Q192* and Q375*) were identified in TP53, both annotated as oncogenic or likely oncogenic and associated with likely loss-of-function. A total of 11 alterations were identified in RET, comprising one missense mutation (V945M) and ten fusion events (CCDC6-RET, ERC1-RET, NCOA4-RET, RET-MRLN, TBL1XR1-RET, TRIM27-RET, AKAP13-RET, DLG5-RET, FKBP15-RET, and SPECC1L-RET). Most RET fusion events were annotated as oncogenic or likely oncogenic with gain-of-function. The RAS gene family showed only missense mutations. KRAS exhibited three missense mutations (G12V, Q61K, and Q61R), whereas NRAS harbored two missense mutations (Q61K and Q61R), and HRAS also showed two missense mutations (Q61K and Q61R). These alterations were predominantly annotated as oncogenic or likely oncogenic and were recurrent hotspot gain-of-function variants. PAX8 comprised five alterations, including one missense mutation (P235Q) and four fusion events (PAX8/PPARG, PAX8-GLIS1, PAX8-NFE2L2, and LRP8-PAX8). The PAX8/PPARG fusion was annotated as oncogenic or likely oncogenic, whereas the remaining fusion events were classified as having unknown functional significance. ALK harbored five alterations, consisting of one missense mutation (P693S) and four fusion events (EML4-ALK, STRN-ALK, GTF2IRD1-ALK, and MALAT1-ALK). All ALK fusion events were annotated as oncogenic or likely oncogenic with gain-of-function. All NTRK1 alterations consisted of five fusion events (SQSTM1-NTRK1, TFG-NTRK1, TPM3-NTRK1, IRF2BP2-NTRK1, and SSBP2-NTRK1), which were annotated as oncogenic or likely oncogenic and associated with likely gain-of-function. NTRK3 exhibited five alterations, including two missense mutations (N294T and R153L) and three fusion events (ETV6-NTRK3, NTRK3-ETV6, and NTRK3-RBPMS). The fusion events were annotated as oncogenic or likely oncogenic with gain-of-function, whereas the missense variants were classified as having unknown functional significance. The spectrum of genomic alterations identified across the analyzed genes included missense, nonsense (truncating), frameshift, splice-site, in-frame insertion/deletion, and fusion mutations, as illustrated in Fig. 6. The complete distribution and characteristics of these alterations are summarized in Table 4.

Table 4.

Summary of essential genomic details of BRAF, RAS (NRAS, HRAS, KRAS), RET, PAX8/PPARγ, TERT, TP53, ALK and NTRK (NTRK1, NTRK3) genes. Data provided include protein changes, mutation type, variant classification, copy number variation, reference/variant status, disease stage (as defined by the American Joint Committee on Cancer - AJCC), functional annotation, ClinVar interpretation and individual neoplasm (cancer) status. All information was retrieved and compiled from the cBioPortal database

	Protein Change	Mutation Type	Variant Type	Copy #	REF/VAR	Neoplasm Disease Stage American Joint Committee on Cancer Code	Annotation	ClinVar	Person Neoplasm Cancer Status	Mutation Status
BRAF	V600E	Missense	SNP	Diploid	A/T	STAGE I, STAGE II, STAGE III, STAGE IVA, STAGE IVC	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Pathogenic/Likely_pathogenic	With Tumor/Tumor Free
	BRAF-AP3B1 Fusion	fusion	NA	Diploid	----	STAGE III	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	With Tumor	Somatic
	FAM114A2-BRAF Fusion	fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	MACF1-BRAF Fusion	fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	MKRN1-BRAF Fusion	fusion	NA	Gain	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	ZC3HAV1-BRAF Fusion	fusion	NA	Amp	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	BCL2L11-BRAF Fusion	fusion	NA	DeepDel	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	BRAF-FAM114A2 Fusion	fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	BRAF-MACF1 Fusion	fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	BRAF-SND1 Fusion	fusion	NA	Diploid	----	STAGE I, STAGE III	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	BRAF-SUGCT Fusion	fusion	NA	DeepDel	----	STAGE I	Oncogenic/Likely oncogenic/Resistance, Likely Gain-of-function	----	Tumor Free	Somatic
	K601E	Missense	SNP	Diploid	T/C	STAGE I, STAGE II	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Likely Gain-of-function	Pathogenic	Tumor Free
	P490_Q494del	In_Frame_Del	DEL	Diploid	TGCTGAGGTGTAGGT/----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Is in CIViC with oncogenic activity information, Recurrent hotspot or recurrent + 3D hotspot	----	Tumor Free
TERT	MTMR12-TERT Fusion	Fusion	SNP	ShallowDel	----	STAGE II	Unknown	----	Tumor Free	Somatic
	R470H	Missense	SNP	Diploid	C/T	STAGE III	Unknown	Conflicting_classifications_of_pathogenicity (Uncertain_significance(2)|Likely_benign(1))	Tumor Free
	S602L	Missense	DEL	Diploid	G/A	STAGE I	Unknown	Uncertain_significance	Tumor Free
	T1113Lfs*62	FS del	NA	Diploid	T/---	STAGE I	Unknown	----	Tumor Free
TP53	Q192*	Nonsense	SNP	Diploid	G/A	STAGE II	Oncogenic/Likely oncogenic/Resistance, Likely Loss-of-function	Pathogenic	Tumor Free
	Q375*	Nonsense	SNP	Diploid	G/A	STAGE III	Oncogenic/Likely oncogenic/Resistance, Likely Loss-of-function	Uncertain_significance
RET	V945M	Missense	SNP	Diploid	G/A	STAGE III	Unknown	Uncertain_significance	Tumor Free
	CCDC6-RET Fusion	Fusion	NA	Diploid/ DeepDel	----	STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	With Tumor/Tumor Free	Somatic
	ERC1-RET Fusion	Fusion	NA	Diploid	----	STAGE I, STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	With Tumor	Somatic
	NCOA4-RET Fusion	Fusion	NA	Diploid/ DeepDel	----	STAGE I, STAGE III, STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
	RET-MRLN Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
	TBL1XR1-RET Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
	TRIM27-RET Fusion	Fusion	NA	Diploid	----	STAGE I, STAGE III, STAGE IVA, STAGE II	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	With Tumor/Tumor Free	Somatic
	AKAP13-RET Fusion	Fusion	NA		----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
	DLG5-RET Fusion	Fusion	NA		----	STAGE III	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	With Tumor	Somatic
	FKBP15-RET Fusion	Fusion	NA		----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
	SPECC1L-RET Fusion	Fusion	NA		----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain-of-function	----	Tumor Free	Somatic
KRAS	G12V	Missense	SNP	Diploid	C/A	STAGE I	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Pathogenic	Tumor Free
	Q61K	Missense	SNP	Diploid	G/T	STAGE I	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Pathogenic	Tumor Free
	Q61R	Missense	SNP	Diploid, Gain/Amp	T/C	STAGE III	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Conflicting_classifications_of_pathogenicity(Pathogenic(1)|Uncertain_significance(2))	Tumor Free
NRAS	Q61K	Missense	SNP	Diploid	G/T	STAGE I, STAGE II, STAGE III, STAGE IV	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Pathogenic/Likely_pathogenic	Tumor Free
	Q61R	Missense	SNP	Diploid/Gain	T/C	STAGE I, STAGE II, STAGE III, STAGE IVA, STAGE IVC	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Likely_pathogenic	With Tumor/Tumor Free
HRAS	Q61K	Missense	SNP	Diploid	G/T	STAGE I, STAGE III, STAGE IVA	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Pathogenic	Tumor Free
	Q61R	Missense	SNP	Diploid, Gain/Amp	T/C	STAGE I, STAGE II, STAGE III, STAGE IVA, STAGE IVC	Oncogenic/Likely oncogenic/Resistance, is an Clvic with oncogenic activity information, Recurrent hotspot or recurrent+3D hotspot, Gain-of-function	Conflicting_classifications_of_pathogenicity(Pathogenic (1)|Uncertain_significance (2))	Tumor Free
PAX8	P235Q	Missense	SNP	Diploid	G/T	STAGE I		----	Tumor Free
	PAX8/PPARỾ	Fusion	NA	Diploid, Gain/ DeepDel	----	STAGE I, STAGE II, STAGE III	Oncogenic/Likely Oncogenic/Resistance	----	Tumor Free	Somatic
	PAX8-GLIS1 Fusion	Fusion	NA	Diploid	----	STAGE I	Unknown	----	Tumor Free	Somatic
	PAX8-NFE2L2 Fusion	Fusion	NA	Diploid	----	STAGE I	Unknown	----	Tumor Free	Somatic
	LRP8-PAX8 Fusion	Fusion	NA	Diploid	----	STAGE I	Unknown	----	Tumor Free	Somatic
ALK	P693S	Missense	SNP	Diploid	G/A	STAGE I	Unknown	----	Tumor Free
	EML4-ALK Fusion	Fusion	NA	Diploid	----	STAGE II	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
	STRN-ALK Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
	GTF2IRD1-ALK Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
	MALAT1-ALK Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
NTRK1	SQSTM1-NTRK1 Fusion	Fusion	NA		----	STAGE II	Oncogenic/Likely Oncogenic/Resistance, Likely Gain of Function	----	Tumor Free	Somatic
	TFG-NTRK1 Fusion	Fusion	NA		----	STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Likely Gain of Function	----	Tumor Free	Somatic
	TPM3-NTRK1 Fusion	Fusion	NA		----	STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Likely Gain of Function	----	Tumor Free	Somatic
	IRF2BP2-NTRK1 Fusion	Fusion	NA		----	STAGE I, STAGE IVA	Oncogenic/Likely Oncogenic/Resistance, Likely Gain of Function	----	With Tumor/Tumor Free	Somatic
	SSBP2-NTRK1 Fusion	Fusion	NA		----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Likely Gain of Function	----	Tumor Free	Somatic
NTRK3	N294T	Missense	SNP	Diploid		STAGE IV	Unknown	----	With Tumor
	R153L	Missense	SNP	Diploid		STAGE I	Unknown	----	Tumor Free
	ETV6-NTRK3 Fusion	Fusion	NA	Diploid/DeepDel	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
	NTRK3-ETV6 Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic
	NTRK3-RBPMS Fusion	Fusion	NA	Diploid	----	STAGE I	Oncogenic/Likely Oncogenic/Resistance, Gain of Function	----	Tumor Free	Somatic

REF/VAR: Reference/Variant status, FS del: Frame_shift_Deletion, In_Frame_Del: Inframe deletion, DeepDel: Deep Deletion, Amp: Amplification, ShallowDel: Shallow Deletion

Fig. 6.

Spectrum of mutation types in key thyroid cancer driver genes from the TCGA PanCancer Atlas Thyroid Carcinoma cohort via cBioPortal. This schematic illustrates the distribution of various genomic alteration types, including missense, nonsense (truncating), in-frame insertions/deletions, splice-site, and fusion mutations in clinically important genes such as BRAF, TERT, TP53, and RET in thyroid cancer. These mutation patterns highlight the molecular heterogeneity of thyroid cancer and their relevance to tumor biology and precision oncology

Pathway enrichment analysis

Pathway enrichment analysis was performed using the enrichR package (version 3.4) in R (version 4.5.1) (https://cran.r-project.org/web/packages/enrichR/index.html) [93]. A total of 5,153 unique genes harboring somatic alterations in the TCGA -THCA PanCancer Atlas cohort were retrieved from cBioPortal and used as the input gene set. Functional enrichment analysis was conducted using the “GO Biological Process 2025” (http://www.geneontology.org/), [94] KEGG 2021 Human“(http://www.kegg.jp/kegg/download/)), [95] Reactome Pathways 2024 ((https://reactome.org/download-data), [96] and TRANSFAC/JASPAR PWMs gene-set libraries (https://jaspar.elixir.no/) [97] available in Enrichr. For Gene Ontology analysis, enriched biological processes were prioritized based on the adjusted P-values reported by Enrichr. For KEGG, Reactome, and TRANSFAC/JASPAR analyses, enriched pathways and transcription factors were ranked according to the Combined Score provided by Enrichr, which integrates the enrichment P-value and Z-score to prioritize biologically relevant terms. Bubble plots were generated using the ggplot2 (version 3.5.2) (https://cran.r-project.org/web/packages/ggplot2/index.html) [88], and viridis (version 0.6.5) (https://cran.r-project.org/web/packages/viridis/index.html) [98], and patchwork (version 1.3.2) packages (https://cran.r-project.org/web/packages/patchwork/index.html) [99] in R.

Gene Ontology (GO) Biological Process enrichment analysis identified several significantly enriched terms, including neuron projection morphogenesis (GO:0048812), negative regulation of immune response (GO:0050777), cell junction assembly (GO:0034329), calcium ion transmembrane import into cytosol (GO:0097553), organelle organization (GO:0006996), positive regulation of excitatory postsynaptic potential (GO:2000463), extracellular structure organization (GO:0043062), cardiac muscle cell action potential (GO:0086001), cellular component assembly (GO:0022607), and modulation of excitatory postsynaptic potential (GO:0098815) (Fig. 7A). KEGG pathway enrichment analysis revealed significant enrichment of pathways related to ECM–receptor interaction, human papillomavirus infection, focal adhesion, thyroid hormone signaling, PI3K–Akt signaling, Rap1 signaling, carbohydrate digestion and absorption, spinocerebellar ataxia, long-term depression, and amoebiasis (Fig. 7B). Reactome pathway analysis highlighted extracellular matrix organization, ECM proteoglycans, signaling by PDGF, the RHO GTPase cycle, non-integrin membrane–ECM interactions, assembly of collagen fibrils and other multimeric structures, degradation of the extracellular matrix, collagen chain trimerization, axon guidance, and collagen formation among the most enriched pathways (Fig. 7C). Transcription factor enrichment analysis using the TRANSFAC/JASPAR PWMs library identified POU3F1, MZF1, MTF1, ZNF148, TCFAP2A, SMAD4, NR2C2, MIR138, MIR133B, KLF11, GABPA, EGR1, CEBPA, AHR, UBTF, POU2F1, KLF4, ETS2, TEAD4, and HOXD9 as the top enriched regulatory factors associated with mutated THCA genes (Fig. 8).

Fig. 7.

Functional pathway enrichment analysis of mutated genes in thyroid carcinoma (THCA). The mutated genes were analyzed using the cBioPortal and Enrichr platform. The analysis included 5153 mutated genes identified in 493 THCA patients in the TCGA PanCancer Atlas cohort. (A) Top 10 Gene Ontology (GO) biological process terms, (B) Top 10 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, (C) Top 10 Reactome pathways were identified. The analysis demonstrated that mutated genes were substantially enriched in a variety of biological processes, including cell proliferation, signal transduction, extracellular matrix organization, and tumor-associated pathways

Fig. 8.

Top 20 transcription factors (TFs) potentially regulating mutated genes in thyroid carcinoma (THCA). Putative regulatory transcription factors were identified from 5,153 mutated genes in the TCGA PanCancer Atlas THCA cohort (via cBioPortal) using motif enrichment analysis from the JASPAR and TRANSFAC^® databases. These TFs are implicated in the control of oncogenic signaling, cell proliferation, and tumor progression pathways, highlighting their potential role in thyroid cancer gene expression programs. Abbreviations: TF, transcription factor; THCA, thyroid carcinoma

Identification of prognostic- mutated genes associated with thyroid carcinoma (THCA)

The GEPIA3 database (https://gepia3.bioinfoliu.com/) is a comprehensive, freely accessible web-based platform that integrates RNA-sequencing data from the TCGA and GTEx projects for gene expression and survival analyses [100]. In this study, prognostic genes associated with THCA were identified using the GEPIA3 survival analysis module based on TCGA-THCA RNA-sequencing data. Overall survival (OS) was assessed using Kaplan–Meier survival analysis with the default median expression cutoff (50% high-expression vs. 50% low-expression groups), and hazard ratios (HRs) were estimated using the Cox proportional hazards model implemented in GEPIA3. The analysis was performed using the following parameters: Dataset Selection = TCGA Tumor, BioType = Protein Coding, Group Cutoff = Median, Time Response = Overall Survival, and Gene/Isoform = Gene. Using a significance threshold of P < 0.05, 1,164 prognostic genes were identified. These genes were subsequently intersected with 5,153 genes harboring at least one somatic alteration in the TCGA-THCA cohort (493 samples) retrieved from cBioPortal using the jvenn web tool (https://jvenn.toulouse.inrae.fr/app/index.html) [101]. The intersection analysis was performed using the default settings of jvenn without additional filtering or preprocessing, resulting in the identification of 341 mutated prognostic genes in THCA (Fig. 9A). The 341 mutated prognostic genes identified through the intersection analysis were subsequently subjected to protein–protein interaction (PPI) analysis using the STRING database (version 12.0; https://string-db.org/) [102]. The following parameters were applied: a minimum required interaction score of 0.400 (medium confidence); network edges representing confidence-based interactions; and active interaction sources including text mining, experimental evidence, curated databases, co-expression, neighborhood, gene fusion, and co-occurrence. Network visualization parameters in STRING were adjusted by disabling structure previews within network bubbles and hiding disconnected nodes.

Fig. 9.

Integrated network analysis of prognostic-mutated genes in thyroid carcinoma (THCA). This figure presents a multi-step bioinformatics analysis integrating mutated and prognostic genes from TCGA PanCancer Atlas data. (A) Venn diagram showing the overlap between mutated genes (cBioPortal) and prognostic genes (GEPIA3). Protein–protein interaction (PPI) network of the overlapping genes. The PPI network was generated using the STRING database, clustered using the MCL algorithm (inflation = 2.5) in Cytoscape, and visualized with the Degree Sorted Circle Layout, (C) Top 50 hub genes identified by network centrality analysis using the CytoHubba plugin. The identified hub genes represent critical regulatory nodes in thyroid cancer progression and prognosis, highlighting promising biomarkers and potential therapeutic targets for precision oncology

The resulting STRING-generated PPI network consisted of 337 nodes and 362 edges, with an average node degree of 2.15 and an average local clustering coefficient of 0.335. The expected number of edges was 284, and the PPI network showed a statistically significant enrichment P-value of 4.64 × 10⁻⁶, indicating that the observed interactions were significantly greater than those expected by chance. Subsequently, the network was imported into Cytoscape software (version 3.10.3; https://cytoscape.org/) [103] for network visualization and topological analysis (Fig. 9B). The Cytoscape-rendered network contained 236 nodes and 362 edges, Network clustering was performed using the Markov Cluster Algorithm (MCL) plugin with an inflation parameter of 2.5, and the resulting network was visualized using the Degree Sorted Circle Layout (https://apps.cytoscape.org/apps/clustermaker2) [104]. Hub genes were identified using the CytoHubba plugin (version 0.1; https://apps.cytoscape.org/apps/cytohubba) [105]. Nodes were ranked according to the Degree algorithm, and the top 50 hub genes among the mutated prognostic genes were selected. The selected hub gene network was subsequently visualized using the GeneMANIA plugin (version 3.5.3; https://apps.cytoscape.org/apps/genemania) [106] (Fig. 9C; Table 5).

Table 5.

The 50 top-defined hub mutated-prognostic genes ranked by Degree method

Rank	Name	Score	Rank	Name	Score	Rank	Name	Score
1	CREBBP	17	17	HEATR1	7	32	UBA1	5
2	CHD4	15	17	PTK2B	7	32	ITCH	5
3	NRXN2	12	17	RBBP8	7	32	DHX37	5
3	COL4A2	12	17	SP1	7	32	KCND3	5
5	MAPK8	11	17	PSMB10	7	32	GOLGB1	5
6	HSPG2	9	17	MRE11	7	32	LARS1	5
6	FH	9	24	P4HB	6	32	PC	5
6	FBN1	9	24	GLI3	6	32	USO1	5
6	ANK2	9	24	UNC13A	6	32	SUPT6H	5
6	MAP1B	9	24	POMGNT1	6	32	PLCB1	5
6	GRIN2A	9	24	WRN	6	32	SF3A1	5
12	ROBO2	8	24	CCNT1	6	32	SDAD1	5
12	HSPD1	8	24	CTCF	6	47	LRRTM2	4
12	SLC17A7	8	24	RB1CC1	6	47	ITGB6	4
12	GRIK2	8	32	PLOD1	5	47	LAMA1	4
12	MMP2	8	32	COL5A3	5	47	RINT1	4
17	KIF1A	7	32	KDM6B	5	-	-	-

Conclusion and future perspective

The integration of molecular imaging and genomics has contributed to more personalized thyroid cancer management, improving diagnostic assessment, therapeutic decision-making, and patient outcomes across the spectrum of disease, from indolent papillary carcinoma to highly aggressive anaplastic thyroid cancer. Advanced imaging modalities such as PET/CT, SPECT/CT, and PET/MRI offer detailed characterization of tumor biology, while novel radiotracers (e.g., FDG, NaF) provide complementary visualization of metabolic alterations, particularly in radioiodine-refractory cancers.

Genomic profiling of key driver genes, including BRAF, RAS (NRAS, HRAS, KRAS), RET, and TERT, offers critical insights into tumor aggressiveness and enables the selection of targeted therapies such as BRAF and RET inhibitors, which have improved response rates and progression-free survival. A combined imaging–genomics approach supports personalized treatment planning, allowing clinicians to distinguish indolent from aggressive disease and tailor therapies to each patient’s unique molecular and imaging profile. Bioinformatics tools now play a central role in precision oncology, enabling integration and interpretation of multi-omics data through platforms such as TCGA, cBioPortal, and databases including GO, KEGG, and Reactome, which elucidate affected pathways such as extracellular matrix remodeling and PI3K–Akt signaling. Incorporating bioinformatics into clinical workflows may support prognostication, informs targeted therapy selection, and complements imaging and genomic profiling, strengthening the foundation of an integrated precision-medicine framework. This narrative review has several limitations. As a non-systematic review, the literature search was mainly conducted in PubMed and Google Scholar, which may have introduced selection bias despite manual screening of reference lists to minimize this issue. The secondary bioinformatics analyses were based exclusively on publicly available datasets, particularly the TCGA PanCancer Atlas Thyroid Carcinoma cohort through cBioPortal and GEPIA3. These resources have inherent limitations, including over-representation of papillary thyroid carcinoma, limited ethnic and geographic diversity, and potential batch effects, which may restrict the generalizability of our findings. Furthermore, significant tumor heterogeneity across histological subtypes and disease stages means that population-level integration of molecular imaging and genomic data may not fully capture individual tumor complexity. Finally, translating these multi-omics insights into routine clinical practice remains challenging due to issues of accessibility, cost, standardization, and the need for prospective validation studies. Future progress requires the standardization of protocols that combine imaging and genomic data, ensuring broader clinical accessibility, feasibility, and cost-effectiveness. Machine learning and artificial intelligence may further support a transformative role, enabling prediction of treatment responses, discovery of biomarkers, and improved interpretation of multidimensional datasets. Prospective clinical trials are necessary to validate emerging theranostic strategies, including, redifferentiation therapies for radioiodine-refractory disease, and combination treatments targeting synergistic co-mutations such as concurrent BRAF and TERT alterations or co-mutated RAS and TP53 mutations.

Abbreviations

ALK

Anaplastic Lymphoma Kinase

ATC

Anaplastic Thyroid Carcinoma

BRAF

v-Raf Murine Sarcoma Viral Oncogene Homolog B

DICER1

Endoribonuclease Dicer

ECM

Extracellular Matrix

ERK

Extracellular Signal-Regulated Kinase

FDG

Fluorodeoxyglucose

fMRI

Functional Magnetic Resonance Imaging

FS del

Frame_shift_Deletion

FTC

Follicular Thyroid Carcinoma

GO

Gene Ontology

HRAS

Harvey Rat Sarcoma Viral Oncogene Homolog

In_Frame_Del

Inframe deletion

KEGG

Kyoto Encyclopedia of Genes and Genomes

KRAS

Kirsten Rat Sarcoma Viral Oncogene Homolog

MAPK

Mitogen-Activated Protein Kinase

MEK

Mitogen-Activated Protein Kinase Kinase

MTC

Medullary Thyroid Carcinoma

NaF

Sodium Fluoride

NGS

Next-Generation Sequencing

NRAS

Neuroblastoma RAS Viral Oncogene Homolog

NTRK

Neurotrophic Tropomyosin Receptor Kinase

PAX8/PPARγ

Paired Box Gene 8/Peroxisome Proliferator-Activated Receptor Gamma

PDGF

Platelet-derived growth factor

PET

Positron Emission Tomography

PIK3CA

Phosphatidylinositol-4,5-Bisphosphate 3-Kinase Catalytic Subunit Alpha

PTC

Papillary Thyroid Carcinoma

PTEN

Phosphatase and Tensin Homolog

RAF

Rapidly Accelerated Fibrosarcoma

RAI

Radioactive Iodine

RET/PTC

Rearranged during Transfection/Papillary Thyroid Carcinoma

SPECT

Single-Photon Emission Computed Tomography

TCGA

The Cancer Genome Atlas

TERT

Telomerase Reverse Transcriptase

VUS

Variant of Uncertain Significance

Author contribution

Samaneh Ahmadi: Conceptualization (lead), Writing – Review & Editing (lead); Mehdi Hoseini: Writing, Search; Table Design; Mehrnaz Sadat Ravari: Writing, Figure Design; Mahboobeh Zarei: Writing, Search; Pejman Shahrokhi: Writing- Review & Editing, Supervision (lead); Pegah Mousavi: Conceptualization (lead), Supervision (lead), Writing – Review & Editing (lead), Project Administration. All authors read and approved the final manuscript.

Data availability

All genomic datasets referenced in this review are publicly available through open-access platforms including TCGA (The Cancer Genome Atlas), cBioPortal, GEPIA3, STRING, and Enrichr. No new datasets were generated for this study.

Declarations

Ethical approval

This review article did not involve human participants, animal studies, or the use of patient data; therefore, ethical approval and consent to participate were not required.

Consent for publication

All co-authors agreed to the submission of the current form of the manuscript.

Competing interests

The author(s) declare that there are no conflicts of interest.