Comprehensive analysis of cuproptosis-related lncRNAs in immunotherapy response and prognosis in papillary thyroid cancer

Introduction

Thyroid cancer is the most prevalent endocrine tumor, with a notable rise in incidence over the past thirty years, ranking it fourth among cancers worldwide (1-3). Papillary thyroid cancer (PTC) is the most prevalent histologic subtype, representing more than 80% of cases (4). While the majority of patients achieve favorable outcomes post-surgery with subsequent radioactive iodine and thyroid-stimulating hormone suppression, a minority exhibit dedifferentiation into aggressive variants, resulting in recurrence and unfavorable prognosis (5,6). Adverse outcomes are influenced by genomic lesions and complex regulatory networks, as illustrated by the BRAFV600E mutation, which activates MAPK and PI3K signaling pathways (7-9). Recent evidence suggests that alterations in the tumor immune microenvironment contribute to the progression of PTC, thereby complicating the prediction of tumor behavior, therapeutic response, and survival outcomes (10,11). The emergence of high-throughput profiling enables the identification of gene signatures associated with specific biological or oncogenic processes, which may enhance prognostication and identify new therapeutic targets in PTC.

Proteotoxic stress ultimately leads to cell death, and the lethality linked to copper dependency is contingent upon functional mitochondrial respiration rather than adenosine triphosphate synthesis (12). Cuproptosis is a newly identified method of programmed cell death (PCD) resulting from excessive intracellular copper buildup. Unlike ferroptosis, excessive copper interacts with lipoylated enzymes in the tricarboxylic acid (TCA) cycle, resulting in protein aggregation and instability of proteins containing iron-sulfur clusters.

Nineteen genes associated with cuproptosis have been identified (12-15). The set comprises NLRP3, NFE2L2, ATP7A, ATP7B, SLC31A1, FDX1, LIAS, LIPT1, LIPT2, DLD, DLAT, PDHA1, PDHB, MTF1, GLS, CDKN2A, DBT, GCSH, and DLST. Multiple members possess recognized functions associated with copper (13-15). Lipoylation increases sensitivity to cuproptosis; thus, proteins capable of promoting or utilizing such modifications, namely FDX1, LIPT1, LIAS, DLD, DLAT, PDHA1, and PDHB, play an indispensable role in this process (12). Copper flux regulators influence susceptibility as well. The importer SLC31A1 and the exporters ATP7A and ATP7B regulate copper homeostasis; elevated expression of SLC31A1 or the lack of ATP7B has been shown to enhance cellular susceptibility to cuproptosis (12-15). The function of cuproptosis in PTC remains inadequately elucidated, although recent progress.

Long non-coding RNAs (lncRNAs) are defined as RNA transcripts exceeding 200 nucleotides in length. They are increasingly acknowledged as essential regulators of tumor initiation and progression in various human malignancies (16,17), with several demonstrating prognostic significance in PTC. Morovat et al. identified three lncRNAs that function as reliable diagnostic markers for thyroid cancer (18). Qin et al. established a five-lncRNA model associated with ferroptosis that forecasts patient prognosis and immune response in thyroid cancer (19).

This study investigated the relationships of cuproptosis-related lncRNAs with molecular functions, the tumor immune microenvironment, and clinical outcomes in PTC. Although most patients with PTC achieve favorable prognoses, a subset develops recurrent, progressive, or treatment-refractory disease, underscoring the need for improved molecular-based risk stratification. For certain patients, biomarkers capable of simultaneously predicting clinical outcomes and estimating potential benefit from immunotherapy would be of clinical value. However, the interplay between cuproptosis-related lncRNAs, prognosis, and the immune landscape in PTC remains poorly understood. Therefore, we aimed to construct a cuproptosis-related lncRNA signature to assess progression risk and explore its association with immune characteristics and predicted responsiveness to immunotherapy in PTC. These findings will provide a comprehensive view of the role of cuproptosis-related lncRNAs in PTC pathogenesis and elucidate their potential relevance for predicting immunotherapy efficacy. We present this article in accordance with the TRIPOD reporting checklist (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-1-0042/rc).

Methods

Data preparation

RNA sequencing data, tumor mutational load, and clinical annotations for PTC have been acquired from The Cancer Genome Atlas, encompassing the THCA cohort of 551 samples, which includes 493 tumors and 58 normal tissues (https://portal.gdc.cancer.gov/). The panel of 19 genes associated with cuproptosis, as outlined by Tsvetkov et al. (12), includes FDX1, DLAT, PDHB, LIAS, LIPT2, ATP7A, GLS, CDKN2A, DBT, DLST, MTF1, SLC31A1, NLRP3, DLD, GCSH, PDHA1, ATP7B, LIPT1, and NFE2L2. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

The screening of cuproptosis-related lncRNAs

The “limma” R package was utilized to identify cuproptosis-associated lncRNAs, employing a correlation cutoff of r>0.40 and a significance threshold of P<0.001. The coexpression relationships between these lncRNAs and cuproptosis genes were illustrated as network flows using the “ggalluvial” package.

Risk score calculation based on cuproptosis-related lncRNAs

The entire PTC cohort was randomly partitioned into training and testing sets (Table 1). We employed least absolute shrinkage and selection operator (LASSO) Cox regression on the training set to establish a cuproptosis-related lncRNA signature for forecasting progression-free survival (PFS). Candidate models were assessed by receiver operating characteristic (ROC) analysis, maintaining those that met the following criteria: P<0.01 and area under the curve (AUC) >0.68 in the training set, and P<0.02 and AUC >0.65 in the testing set.

We computed a risk score for each patient by summing the product of the expression levels of lncRNAs and their corresponding Cox coefficients. Patients were categorized into low or high risk based on the cohort median cutoff. The predictive performance of the score was assessed using time-dependent ROC curves and the concordance index, employing the “timeROC” and “Hmisc” R packages, with results validated in the testing set.

Nomogram construction

Independent prognostic variables identified through multivariable analysis were utilized to create a nomogram using the “rms” R package. Calibration of the model was evaluated through calibration plots to determine the concordance between predicted and observed outcomes.

Function enrichment

The “limma” R package was utilized to detect differentially expressed genes (DEGs) between the low- and high-risk groups, applying a threshold of |log₂ fold change| >1 and an adjusted P value <0.05. Functional enrichment analysis for Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was performed utilizing “clusterProfiler,” “GOplot,” and “enrichplot,” with a significant threshold of false discovery rate (FDR) <0.05 (20,21).

Immune infiltration and immunotherapy response

The distribution of each invading immune cell type was assessed utilizing the CIBERSORT algorithm, as executed in the “CIBERSORT” R package (version 1.03) (22). The quantities of immune cells and immune-related functional scores were analyzed between the low- and high-risk groups, with statistical significance defined at P<0.05. We evaluated potential response to immunotherapy by evaluating tumor mutational burden (TMB) (23) and employing the Tumor Immune Dysfunction and Exclusion (TIDE) database (24). The anticipated sensitivities to anticancer drugs were subsequently obtained utilizing the “pRRophetic” R package (25).

RNA preparations and reverse transcription quantitative polymerase chain reaction (RT-qPCR)

Total RNA was extracted from surgical specimens and cultured cells utilizing the EZ-press RNA Purification Kit (EZBioscience, Minnesota, USA). Complementary DNA was synthesized from the extracted RNA using StarScript III All-in-One RT Mix with gDNA Remover (GenStar, Beijing, China). Quantitative PCR was conducted utilizing 2× RealStar Fast SYBR qPCR Mix (GenStar, Beijing, China), employing cDNA as the template and β-ACTIN as the internal reference to measure the expression levels of DLG3-AS1 and LINC01545. All reactions were conducted in technical triplicate, and relative abundance was assessed using the 2^−ΔΔCq method. Primer sequences were as follows: β-ACTIN, forward 5'-AACTGGGACGACATGGAGAAAA-3' and reverse 5'-GGATAGCACAGCCTGGATAGCA-3'; DLG3-AS1, forward 5'-ACAAGAAGAGGAAGCAGTTAGG-3' and reverse 5'-GCAGAAGAGTGTCTGTGAAGTA-3'; LINC01545, forward 5’-CCTCTGCAAGCTCAAAGCAAG-3' and reverse 5’-GGTGATTTGTTGATGCTGGGG-3'.

Statistical analysis

All statistical analyses were performed using R (version 4.0.3). Data visualization was conducted with the “ggplot2” and “pheatmap” tools. Comparisons were conducted using either the chi-square test or Fisher’s exact test, contingent upon the distribution of categorical data. PFS, disease-specific survival (DSS), and overall survival (OS) were assessed using Kaplan-Meier analyses and Cox proportional hazards models (both univariate and multivariable), utilizing the “survival” and “survminer” packages. A two-tailed P value of less than 0.05 was deemed statistically significant. All studies were conducted with three distinct biological replicates. Data are presented as the mean ± standard error of the mean. Differences in matched tissue samples were evaluated using paired t-tests. Correlations among variables were analyzed utilizing Spearman’s rank approach. Survival curves were generated utilizing GraphPad Prism 10, with statistical significance established at P<0.05.

Results

Identification of lncRNAs linked to cuproptosis in PTC

Figure S1 provides the overview of the study design. We identified 2,578 lncRNAs (Figure 1A) that were coexpressedly correlated with the 19 cuproptosis-related genes identified in Tsvetkov’s study (12). A total of 551 PTC patients obtained from The Cancer Genome Atlas (TCGA) database were randomly divided into training and testing groups for further investigation (Table 1). Thereafter, the cuproptosis-related lncRNAs recognized for their predictive relevance in PFS by univariate analysis were analyzed using LASSO COX regression. This process culminated in the identification of ten risk lncRNAs (AC011815.1, LINC02458, PSMG3-AS1, LINC01545, AL627095.2, AC097532.3, AC084782.3, AL158071.2, DLG3-AS1, and Z97192.4) within the training group (Figure 1B,1C). Figure 1D illustrates the coexpression heatmap of cuproptosis-related lncRNAs and genes. The calculation of a cuproptosis-based risk score utilized these ten lncRNAs.

Figure 1 The construction of risk score based on cuproptosis-related lncRNAs. (A) The identification of lncRNAs that were associated with cuproptosis in PTC. (B) The screening of risk cuproptosis-related lncRNAs based on LASSO Cox regression analysis to build a prognostic model in PTC. (C) The multivariate analysis of ten recognized cuproptosis-related lncRNAs to further build the model. (D) The coexpression heatmap of cuproptosis-related genes and the ten lncRNAs. (E) The PCA plot of certain genes between low- and high-risk groups divided by the median value of risk score based on the ten lncRNAs. CI, confidence interval; LASSO, least absolute shrinkage and selection operator; PCA, principal component analysis; PTC, papillary thyroid cancer.

To further assess biological links between candidate lncRNAs and cuproptosis, we calculated a cuproptosis pathway activity score and examined its correlation with the signature lncRNAs. LINC02458 was positively correlated with the cuproptosis score, whereas the other candidates showed negative correlations (Figure S2A). Moreover, expression of key downstream genes varied across the cuproptosis landscape (Figure S2B), and GO enrichment analysis revealed enrichment in RNA splicing, ribonucleoprotein complex biogenesis, regulation of DNA metabolism, and macromolecule localization (Figure S2C). These findings provide further pathway-level support for the biological relevance of these cuproptosis-related lncRNAs.

Risk score calculation based on the cuproptosis-related lncRNAs

A risk score was calculated for each case by summing the product of the expression levels of the signature lncRNAs and their respective coefficients. Patients were stratified into low- and high-risk groups using the cohort median as the cutoff (Figure 1E). The distribution of risk score and its association with PFS and DSS in the training group were shown in Figure 2. Of the ten lncRNAs examined, AC011815.1, LINC02458, PSMG3-AS1, LINC01545, AL627095.2, and AC097532.3 exhibited overexpression in high-risk PTC patients, whereas AC084782.3, AL158071.2, DLG3-AS1, and Z97192.4 showed downregulation in this group. Consistent results were observed in the test group and among all patients (Figure 2).

Figure 2 The distribution of risk score. (A) The correlation between risk score distribution and PFS, as well as the expression patterns of the ten cuproptosis-related lncRNAs between low- and high-risk group. (B) The correlation between risk score distribution and DSS, as well as the expression patterns of the ten cuproptosis-related lncRNAs between low- and high-risk group. DSS, disease-specific survival; PFS, progression-free survival.

Prognostic significance of risk score in PTC

In the training cohort, increased risk scores were associated with diminished PFS, a pattern that was similarly noted in the testing set and the total cohort (Figure 3A). Risk scores demonstrated inverse associations with DSS and OS in the training, testing, and combined cohorts; however, several comparisons failed to achieve statistical significance, likely due to the low event rates typical of this predominantly indolent cancer. In accordance with these findings, the high-risk category continued to be linked to poorer PFS within tumor-node-metastasis (TNM)-defined subgroups (Figure 3B).

Figure 3 Kaplan-Meier survival curves. (A) PFS, DSS and OS curves of PTC patients between low- and high-risk group in all patients, train and test group, respectively. (B) Subgroup PFS analysis according to TNM stage. DSS, disease-specific survival; OS, overall survival; PFS, progression-free survival; PTC, papillary thyroid cancer; TNM, tumor-node-metastasis.

The evaluation of predictive accuracy of risk score

The predictive performance for PFS was evaluated through time-dependent ROC analysis (Figure 4A) and the concordance index in the overall cohort (Figure 4B). The risk score yielded AUCs of 0.776, 0.763, and 0.822 for 1-, 3-, and 5-year PFS, respectively. The risk score demonstrated the highest AUC for 5-year PFS among all variables analyzed, surpassing traditional clinicopathologic indicators (Figure 4A). The C-index for PFS consistently surpassed that of other parameters (Figure 4B). Analyses for DSS and OS were not conducted due to the insufficient number of events in this condition. The results combined indicate that the risk score provides improved prognostic differentiation for PFS in individuals with PTC.

Figure 4 The evaluation of prognostic prediction values of risk score and construction of nomogram. (A) The 1-/3-/5-year ROC curves of risk score and the comparison of AUC among risk score and other clinicopathological parameters. (B) The comparison of C-index between risk score and other clinicopathological parameters. (C) The univariate analysis of PFS in PTC patients. (D) The multivariate analysis of PFS in PTC patients. (E) The nomogram to predict PFS based on risk score in all PTC patients. (F) The calibration curves to compare the predicted and actual 1-/3-5-year PFS in all patients, train and test group, respectively. *, P<0.05; ***, P<0.001. AUC, area under curve; CI, confidence interval; HR, hazard ratio; N, node; PFS, progression-free survival; PTC, papillary thyroid cancer; ROC, receiver operating characteristic curve; T, tumor.

Nomogram construction based on risk score

We conducted univariable and multivariable Cox proportional hazards analysis to identify parameters independently correlated with PFS in PTC (Figure 4C,4D). Ultimately, age, T stage, N stage, and risk score were recognized as risk factors for PFS. The identified risk factors were subsequently utilized to create a nomogram for predicting PFS of PTC (Figure 4E). The calibration curves demonstrated strong concordance between projected and actual 1-/3-/5-year PFS in the training group, testing group, and the whole patient cohort, respectively (Figure 4F).

Comparative GO and KEGG enrichment in high-risk and low-risk PTC

We used differential expression analysis to identify DEGs between the low- and high-risk groups. GO enrichment (Figure 5A,5B) indicated that these genes were primarily linked to very-low-density lipoprotein particles, triglyceride-rich plasma lipoprotein particles, nutrient availability responses, sulfur compound binding, proteoglycan binding, protease binding, and the structure, components, and functions of the extracellular matrix. KEGG analysis identified modifications in cholesterol metabolism, ECM-receptor interactions, AGE-RAGE signaling associated with diabetic complications, Wnt signaling, PI3K-Akt signaling, IL-17 signaling, and proteoglycans in cancer (Figure 5C,5D).

Figure 5 The comprehensive analysis of function enrichment, immune infiltration and TMB between low- and high- risk group. (A,B) The top 10 enriched BP, CC and MF between low- and high-risk group. (C,D) The most significantly altered KEGG pathways between low- and high-risk group. (E) The relationship between TMB level and risk score in PTC. (F) The top 15 mutated genes in low- and high-risk group, respectively. (G) The comparison of PFS, DSS and OS in subgroups divided by risk score and TMB level. (H) The correlation between risk score and RNA stemness in PTC. BP, biological process; CC, cellular component; DSS, disease-specific survival; H, high; KEGG, Kyoto Encyclopedia of Genes and Genomes; L, low; MF, molecular function; OS, overall survival; PFS, progression-free survival; PTC, papillary thyroid cancer; TMB, tumor mutational burden.

The TMB and RNA stemness comparison between low- and high-risk PTC patients

High-risk PTC patients had elevated TMB in comparison to low-risk individuals (P=0.04, Figure 5E). In both risk categories, BRAF was the most commonly changed gene, primarily exhibiting missense variants (55% in the low-risk group and 65% in the high-risk group), followed by NRAS missense mutations (10% versus 6%; Figure 5F). Figure 5G depicts the survival effect of TMB. Patients with greater TMB experienced inferior PFS, DSS, and OS relative to those with minimal TMB (All P<0.01). We subsequently categorized individuals according to their risk score and TMB. Individuals categorized as high-risk with elevated TMB had the shortest PFS, followed in order by high-risk with low TMB, low-risk with high TMB, and low-risk with low TMB. Concurrent patterns were seen for DSS and OS. The risk score exhibited a significant correlation with the RNA-based stemness index of PTC cells (P<0.001; Figure 5H).

The correlation between immune infiltration and risk score in PTC

We utilized CIBERSORT to analyze the correlation between tumor-infiltrating immune cells and the ten lncRNAs in the signature (Figure 6A). The composite risk score had a positive association with dendritic cells and resting memory CD4 T cells, while demonstrating a negative correlation with follicular helper T cells, CD8 T cells, and plasma cells.

Figure 6 The comprehensive analyses of risk score in immune infiltration, immunotherapy response and drug sensitivity in PTC. (A) The correlation between infiltrating levels of immune cells and the ten risk lncRNAs according to CIBERSORT analyses. (B) The comparison of immune functions between low- and high-risk groups. (C) The expression profile of immune blockades between low- and high-risk group. (D) The association between immunotherapy response and risk score evaluated by comparison of scores of dysfunction, exclusion and TIDE according to TIDE database. (E) The correlation between risk score and drug sensitivity to treatments of phenformin, LFM-A13, OSU-03012, GNF-2 and BI-2536, respectively. *, P<0.05; **, P<0.01; ***, P<0.001. IC₅₀, half-maximal inhibitory concentration; PTC, papillary thyroid cancer; TIDE, Tumor Immune Dysfunction and Exclusion.

We additionally evaluated immunological functioning outcomes (Figure 6B). Elevated risk scores correlated with enhanced co-inhibition and co-stimulation of antigen-presenting cells, increased major histocompatibility complex (MHC) class I activity, heightened parainflammatory signaling, and amplified human leukocyte antigen (HLA) fingerprints.

Figure 6C illustrates the correlations of checkpoints. The risk score exhibited positive coexpression with PD-L1, PD-L2, CTLA-4, IDO-1, CD276, and TNFSF4, while demonstrating negative coexpression with TNFRSF4, TNFRSF14, and TNFRSF25.

High risk score predicts better immunotherapy response in PTC

The levels of TMB, immunological infiltration, and immune blockades have been observed to correlate with responses to immunotherapy. We subsequently evaluated the correlation between the risk score and anticipated immunotherapy efficacy utilizing the TIDE database. TIDE evaluates response based on a composite score generated from T-cell exclusion and dysfunction metrics (24). Patients at high risk exhibited an elevated T-cell exclusion score, in conjunction with diminished functionality and total TIDE scores. A reduced TIDE score indicates increased sensitivity to immune-checkpoint blocking, implying better response and prognosis for the high-risk group undergoing anti-programmed cell death protein 1 (PD-1) or anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) monotherapy (Figure 6D). The reactions to alternative pharmacological therapies were also assessed utilizing the “pRRophetic” R package (25). High-risk PTC patients were anticipated to respond favorably to phenformin, LFM-A13, and OSU-03012 therapies, whereas low-risk individuals exhibited sensitivity to GNF-2 and BI-2536 treatments (Figure 6E).

To assess whether immune characteristics linked to risk score grouping are also evident at the individual transcript level, we compared selected immune checkpoint genes and TIDE scores between low- and high-expression groups of each candidate cuproptosis-related lncRNA. The results showed that multiple lncRNAs were significantly correlated with immune checkpoint expression and TIDE scores, with an overall trend resembling the immune phenotype observed in risk stratification (Figure S3). These findings further support the link between these cuproptosis-related lncRNAs and immune remodeling, and highlight their potential to predict immunotherapy response in PTC.

High TMB status predicts better immunotherapy response in PTC

Because TMB may reflect genomic instability and neoantigen burden, the correlation between TMB status and tumor immune features, as well as immunotherapy response, was further analyzed. Stromal score, immune score, and ESTIMATE score showed distinct distributions across TMB strata (Figure 7A). In addition, several immune checkpoint-related genes exhibited differential expression trends between the two groups, with IDO1 showing a significant increase in the TMB-high subgroup (Figure 7B). Correlation analyses further revealed that TMB was positively associated with activated dendritic cells and regulatory T cells (Tregs), but negatively associated with CD8⁺ T cells (Figure 7C-7E). To further evaluate the potential relevance of TMB to immunotherapy responsiveness, we compared TIDE scores between TMB-low and TMB-high groups. The TMB-high group exhibited a higher exclusion score but lower dysfunction and overall TIDE scores than the TMB-low group (Figure 7F), consistent with the trend observed in the high-risk patient subgroup (Figure 6D). These findings suggest that TMB status is associated with distinct immune microenvironment characteristics, and that high TMB status predicts better immunotherapy response in PTC.

Figure 7 Immunological relevance of tumor mutational burden in PTC. (A) Comparison of stromal score, immune score, and ESTIMATE score between the TMB-low and TMB-high groups. (B) Comparison of immune checkpoint-related gene expression between the TMB-low and TMB-high groups. (C-E) Correlations between TMB and activated dendritic cells, regulatory T cells, and CD8 T cells, respectively. (F) Comparison of dysfunction, exclusion, and TIDE scores between the TMB-low and TMB-high groups. *, P<0.05; **, P<0.01. PTC, papillary thyroid carcinoma; TIDE, Tumor Immune Dysfunction and Exclusion; TMB, tumor mutational burden.

Validation of the high-risk cuproptosis-related lncRNAs in PTC

Based on bioinformatic analysis and a focused literature scan, we prioritized two previously uncharacterized lncRNAs, DLG3-AS1 and LINC01545, for experimental confirmation. Their expression was quantified in tumor specimens and corresponding neighboring normal tissues. DLG3-AS1 demonstrated a marked decrease in tumors, consistent with the computational results (Figure 8A). LINC01545 exhibited substantial upregulation in tumor tissue relative to corresponding normal tissues (Figure 8B). These data support the results of the bioinformatic analyses and highlight DLG3-AS1 and LINC01545 as potential prognostic markers and therapeutic targets in PTC.

Figure 8 Validation of representative cuproptosis-related lncRNAs in PTC tissues. (A) The relative expression levels of DLG3-AS1 in PTC and para- cancer tissues were confirmed by qPCR. (B) The relative expression levels of LINC01545 in PTC and adjacent tissues were confirmed by qPCR. *, P<0.05; ***, P<0.001. PTC, papillary thyroid cancer; qPCR, quantitative polymerase chain reaction.

Discussion

Dysregulation of PCD contributes to human malignancy and exhibits characteristic biological and immunological signatures (26,27). In recent years, multiple PCD modalities have been delineated, including ferroptosis, pyroptosis, entosis, NETosis, necroptosis, oxeiptosis, alkaliptosis, parthanatos, autophagy, and lysosome-dependent cell death (28). Clinically, promoting apoptosis—the canonical and most extensively investigated PCD program—remains a cornerstone anticancer strategy (29). Cuproptosis is a newer PCD mechanism driven by copper metabolism (12). Copper, akin to iron, is a vital trace metal that facilitates organismal function and homeostasis; its direct interaction with lipoylated enzymes of the TCA cycle may promote protein aggregation and tumorigenesis (30). Nonetheless, the impact of cuproptosis and its related lncRNAs on PTC remains inadequately elucidated. Systematic analysis of cuproptosis-associated lncRNAs may yield prognostic indicators and uncover alternative treatment options for PTC.

This study discovered six lncRNAs associated with elevated risk and four associated with diminished risk, all related to cuproptosis in PTC. Among the high-risk set, PSMG3-AS1 has been characterized as oncogenic, enhancing proliferation, invasion, migration, and resistance to therapy in several malignancies (31-34). Consistent with these observations, Huang et al. reported that PSMG3-AS1, regulated by m6A modification, predicts unfavorable PFS in PTC based on bioinformatic evidence (35). The remaining nine cuproptosis-related lncRNAs have not yet been functionally described in human cancers. Together, these findings support both the robustness and the novelty of our cuproptosis-related lncRNA signature.

Recent studies in other tumor types have suggested that cuproptosis-related lncRNAs are not merely coexpression markers, but may be linked to broader regulatory programs involved in copper homeostasis, mitochondrial stress, tumor progression, and immune remodeling (36,37). In ovarian cancer, cuproptosis-related lncRNA signatures were associated with prognosis, immune infiltration, and immunotherapy-related indices (38,39), and one study further showed that LINC02285 promoted ovarian cancer cell proliferation and migration and reduced sensitivity to Elesclomol-CuCl2, providing functional support that at least some cancer-related lncRNAs (CRLs) may influence cuproptosis-associated phenotypes at the cellular level (39). Similar associations between cuproptosis-related lncRNA signatures, outcome, and immune characteristics have also been reported in glioma and hepatocellular carcinoma (40-42). In this context, the enrichment patterns observed in our supplementary analysis, particularly those related to RNA splicing, ribonucleoprotein complex biogenesis, and DNA metabolic regulation, are biologically plausible and suggest that the candidate lncRNAs may participate in broader transcriptional and post-transcriptional programs relevant to cuproptosis biology rather than representing only coexpression-based statistical associations. At the same time, direct mechanistic evidence remains limited for most individual transcripts, and further experimental studies will be needed to clarify how these lncRNAs functionally interact with cuproptosis-related pathways in PTC.

We assessed GO enrichment between low- and high-risk PTC cohorts utilizing the median risk score derived from the ten-lncRNA panel. The most impacted phrases encompassed very-low-density lipoprotein particles, triglyceride-rich plasma lipoprotein particles, and sulfur compound binding. These findings align with the biology of cuproptosis, which relies on lipoylated metabolites from the TCA cycle and leads to the degradation of iron-sulfur cluster proteins (12). Prior studies further support a TCA-centered metabolic shift in PTC: pyruvate carboxylase, a key anaplerotic enzyme, is upregulated, correlates with lymph node metastasis, and enhances lipogenesis through activation of the Akt/mTOR pathway (43). Proteomic profiling has likewise shown marked activation of the TCA cycle in PTC (44). Taken together, these observations implicate TCA cycle-dependent cuproptosis in PTC initiation and metastatic progression.

KEGG enrichment highlighted pathways in cholesterol metabolism, ECM-receptor interaction, AGE-RAGE signaling in diabetic complications, Wnt signaling, PI3K-Akt signaling, IL-17 signaling, and proteoglycans in cancer. Cholesterol handling influences multiple oncogenic processes, including proliferation (45), apoptosis (46), and invasive behavior (47). In addition, enhanced cholesterol efflux can drive tumor-associated macrophage-mediated progression by fostering immunosuppressive and trophic programs (48), while excess cholesterol upregulates immune checkpoints in CD8⁺ T cells, promoting T-cell exhaustion within the tumor microenvironment (49). Consistent with these observations, gut microbiome and metabolomic profiling linked steroid biosynthesis and lipid digestion to PTC development (50), and a nationwide study in Korea associated reduced high-density lipoprotein cholesterol with greater thyroid cancer risk, particularly in metabolically unhealthy individuals (51).

Extracellular matrix components such as collagen and fibronectin facilitate PTC cell growth, invasion, and migration (52), in line with our GO results that indicated alterations in ECM structure, composition, and function. IL-17 family cytokines exert strong effects on inflammatory and immune responses and contribute to metabolic disorders and malignancy, partly via cascades that include AGE-RAGE, MAPK, and NF-κB signaling (53). The primary etiological factors in thyroid cancer include the activation of MAPK and PI3K-Akt pathways due to BRAFV600E and RAS mutations, while the Wnt pathway is also crucial in the biology of PTC (54,55). Altogether, these enriched pathways outline a regulatory network connected to cuproptosis in PTC, warranting deeper investigation, with special emphasis on immune regulatory circuits.

PTC frequently co-occurs with autoimmune thyroiditis, and dense immune-cell infiltration is often observed intraoperatively even in patients without a formal autoimmune diagnosis (11,56). Consistent with prior work showing that shifts in the immune milieu can promote PTC development (10), we examined how the tumor immune microenvironment relates to cuproptosis. Patients in the high-risk stratum exhibited greater TMB than those in the low-risk group. In the context of oncology, TMB is often regarded as a proxy indicator of tumor heterogeneity, as a higher mutational burden increases the likelihood of generating neoantigens. However, recent studies have emphasized that the predictive value of TMB is context-dependent on the disease itself and should be interpreted in conjunction with tumor lineage, immune background, and the biological quality of potential mutations, rather than as an isolated observational variable (57-59). This is particularly important in PTC, as mounting evidence supports significant heterogeneity in the tumor immune microenvironment. Recent research has shown that patterns of immune cell infiltration and expression of immune-related genes are closely associated with tumor progression and prognosis in PTC (60), while other studies have indicated that some PTC may exhibit a stronger immune response phenotype (61). Furthermore, transcriptomic analysis suggests that T cell exhaustion and immune escape programs are important components of the immune landscape in PTC (62). In this context, PCD should also be considered, as recent literature reports that pathways such as immunogenic cell death (ICD), ferroptosis, and cuproptosis may affect tumor progression, immune signaling, and therapeutic response (63). Therefore, in our cohort, the association between higher TMB and unique immune phenotypes supports the notion that TMB may be involved in the remodeling of the immune microenvironment and may help predict the immunotherapy response in PTC.

The composite risk score had a positive correlation with dendritic cells and resting memory CD4 T cells, as well as with immune functional metrics such as MHC class I activity, parainflammation, and HLA markers. Due to immune escape diminishing the effectiveness of immunotherapy, we analyzed checkpoint expression across different strata and identified elevated levels of PD-L1, PD-L2, CTLA-4, IDO-1, CD276, and TNFSF4 in the high-risk group, indicating heightened vulnerability to immune-checkpoint blocking. To further estimate therapeutic benefit, we applied TIDE, which models T-cell dysfunction and exclusion (24). High-risk patients showed higher exclusion scores, lower dysfunction scores, and lower overall TIDE scores, patterns that predict better responses and improved prognosis with anti-PD-1 or anti-CTLA-4 monotherapy.

The supplementary analysis of lncRNA levels further supports this explanation. Several lncRNAs associated with cuproptosis are correlated with checkpoint expression and TIDE scores, suggesting that our model does not solely rely on risk scores to explain the correlation with immune expression, but also through post-transcriptional level analysis. Although these findings currently lack direct validation, they provide additional support for the biological link between lncRNAs associated with cuproptosis and immune phenotypes related to predicted immunotherapy response in PTC.

Beyond immunotherapy, we also assessed predicted responses to targeted agents to inform a more comprehensive treatment strategy for PTC (25). The high-risk subgroup was forecast to be more responsive to phenformin, LFM-A13, and OSU-03012, whereas the low-risk subgroup showed greater predicted sensitivity to GNF-2 and BI-2536.

Accumulating studies show that lncRNAs shape gene expression at the transcriptional, post-transcriptional, and post-translational tiers (64). Some lncRNAs base-pair with genomic DNA to modify chromatin accessibility and influence transcription. For example, cis-acting transcripts can repress nearby genes through local chromatin interactions, whereas enhancer-derived lncRNAs stimulate transcription by activating their cognate enhancer elements (65). Additionally, certain lncRNAs form complexes with proteins to post-transcriptionally regulate gene expression (66). Some operate through base-pairing with target RNAs to attract regulatory proteins or serve as competing endogenous RNAs (ceRNAs) that sequester microRNAs through complementary binding (67).

To clarify functional relevance, we combined focused literature curation with in vitro assays for the ten candidate lncRNAs. PSMG3-AS1 stood out: its expression is elevated in endometrial carcinoma and cervical squamous cell carcinoma (32,33), and higher levels predict poorer outcomes in hepatocellular carcinoma (68). The host gene PSMG3 is also overexpressed in tumors and associates with lymph node metastasis and advanced stage (69), suggesting possible cooperative oncogenic activity between PSMG3-AS1 and PSMG3.

Although DLG3-AS1 has not been characterized in PTC, its paralog DLG1 functions as a tumor suppressor and is frequently reduced in several solid malignancies, including cervical, colorectal, and breast cancers (70-72). By contrast, DLG1-AS1 has been implicated in tumor promotion across multiple settings—cervical carcinoma, triple-negative breast cancer, hepatocellular carcinoma, and PTC—where its upregulation correlates with disease progression (73-76).

Of note, the host gene DLG3 is overexpressed in breast carcinoma and has been implicated in docetaxel resistance, as well as increased cellular migration and invasion (77). Consistent with this context, we observed that DLG3-AS1 is markedly reduced in PTC, suggesting a possible tumor-suppressive role. Before this study, diminished DLG3-AS1 expression had been documented only in the central nervous system of patients with Parkinson’s disease, underscoring the novelty of its association with PTC (78).

Peripheral circulation DLG1-AS1 has been proposed as a blood-based biomarker to differentiate benign from malignant thyroid nodules (76). Our study revealed that DLG3-AS1 expression was significantly reduced in PTC tissue compared to nearby normal tissue. These observations lead us to propose that DLG3-AS1 may have prognostic utility in PTC and could be measurable in plasma.

Recent studies have also developed several prognostic models for PTC. A Surveillance, Epidemiology, and End Results (SEER)-based nomogram identified age, tumor stage, gender, and marital status as independent prognostic factors, supporting the value of clinicopathologic variables in risk stratification (79). An ultrasound-based nomogram further showed that combining radiomics features with clinicopathological characteristics achieved better discrimination than the clinical model alone (80). Likewise, a DeepSurv model based on clinical risk factors showed good performance across the SEER, MDACC, and TCGA datasets, indicating that clinical variables can still provide meaningful prognostic information when modeled appropriately (81). However, recent molecular models suggest that gene- or lncRNA-based signatures may capture additional biological heterogeneity. A 4-mRNA model remained associated with survival after adjustment for clinical parameters (82), and an ICD-related lncRNA model further linked the risk signature to immune infiltration, TMB, and TIDE (83). Taken together, these studies suggest that adding gene or lncRNA features may provide prognostic value beyond conventional clinicopathologic parameters alone. Consistently, our current study established a cuproptosis-related lncRNA risk model, and the derived risk score showed superior prognostic value compared with traditional clinicopathological characteristics.

Although in current study, the risk score showed higher time-dependent AUCs and C-index values than traditional-single clinicopathologic variable, several constraints warrant further concern (78). First, the molecular function roles of the implicated lncRNAs in PTC biological progression remain to be clarified. Second, because PTC is a predominantly indolent malignancy, the relatively low number of progression events may increase uncertainty in multivariable Cox estimates. Third, although the model was validated by an internal test subgroup, due to the relative indolent malignancy of PTC, an external validation or a longer follow-up is needed.

Conclusions

Collectively, we developed a cuproptosis-associated lncRNA signature that estimates PFS in PTC. The derived risk score achieved stronger prognostic discrimination than standard clinicopathologic features. High-risk patients had an increased TMB, enhanced immune-cell infiltration, and raised checkpoint expression, leading to predictions of greater benefit from immune-checkpoint blocking. If confirmed prospectively, this methodology could aid in the selection of candidates for anti-PD-1/PD-L1 or anti-CTLA-4 therapy.

Acknowledgments

The authors wish to thank all the researchers contributing to the public databases used in current study.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-1-0042/rc

Peer Review File: Available at https://gs.amegroups.com/article/view/10.21037/gs-2026-1-0042/prf

Funding: This study was supported in part by grants from the Seed Projects of Beijing Friendship Hospital, Capital Medical University (No. YYZZ202313), the Training Fund for Open Projects at clinical Institutes and Department of Capital Medical University (No. CCMU2024ZKYXY007), and the Natural Science Foundation of Capital Medical University (No. PYZ24066).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://gs.amegroups.com/article/view/10.21037/gs-2026-1-0042/coif). Z.L. reports receiving support from the Seed Projects of Beijing Friendship Hospital, Capital Medical University, the Training Fund for Open Projects at clinical Institutes and Department of Capital Medical University, and the Natural Science Foundation of Capital Medical University. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The clinical project was approved by the Ethics Committee of Beijing Friendship Hospital, Capital Medical University (approval No. BFHHZS20250232). Written informed consent was obtained from all participants prior to inclusion in the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Hernando J, Ros J, Arroyo A, et al. Clinical and Translational Challenges in Thyroid Cancer. Curr Med Chem 2020;27:4806-22. [Crossref] [PubMed]
2. Chen DW, Lang BHH, McLeod DSA, et al. Thyroid cancer. Lancet 2023;401:1531-44. [Crossref] [PubMed]
3. Kim J, Gosnell JE, Roman SA. Geographic influences in the global rise of thyroid cancer. Nat Rev Endocrinol 2020;16:17-29. [Crossref] [PubMed]
4. Lim H, Devesa SS, Sosa JA, et al. Trends in Thyroid Cancer Incidence and Mortality in the United States, 1974-2013. JAMA 2017;317:1338-48. [Crossref] [PubMed]
5. Feng J, Shen F, Cai W, et al. Survival of aggressive variants of papillary thyroid carcinoma in patients under 55 years old: a SEER population-based retrospective analysis. Endocrine 2018;61:499-505. [Crossref] [PubMed]
6. Carling T, Ocal IT, Udelsman R. Special variants of differentiated thyroid cancer: does it alter the extent of surgery versus well-differentiated thyroid cancer? World J Surg 2007;31:916-23. [Crossref] [PubMed]
7. Kowalska A, Walczyk A, Kowalik A, et al. Increase in Papillary Thyroid Cancer Incidence Is Accompanied by Changes in the Frequency of the BRAF V600E Mutation: A Single-Institution Study. Thyroid 2016;26:543-51. [Crossref] [PubMed]
8. Zhang J, Du Y, Zhang X, et al. Downregulation of BANCR Promotes Aggressiveness in Papillary Thyroid Cancer via the MAPK and PI3K Pathways. J Cancer 2018;9:1318-28. [Crossref] [PubMed]
9. Knippler CM, Saji M, Rajan N, et al. MAPK- and AKT-activated thyroid cancers are sensitive to group I PAK inhibition. Endocr Relat Cancer 2019;26:699-712. [Crossref] [PubMed]
10. Nicolson NG, Brown TC, Korah R, et al. Immune cell infiltrate-associated dysregulation of DNA repair machinery may predispose to papillary thyroid carcinogenesis. Surgery 2020;167:66-72. [Crossref] [PubMed]
11. Resende de Paiva C, Grønhøj C, Feldt-Rasmussen U, et al. Association between Hashimoto's Thyroiditis and Thyroid Cancer in 64,628 Patients. Front Oncol 2017;7:53. [Crossref] [PubMed]
12. Tsvetkov P, Coy S, Petrova B, et al. Copper induces cell death by targeting lipoylated TCA cycle proteins. Science 2022;375:1254-61. [Crossref] [PubMed]
13. Moriya M, Ho YH, Grana A, et al. Copper is taken up efficiently from albumin and alpha2-macroglobulin by cultured human cells by more than one mechanism. Am J Physiol Cell Physiol 2008;295:C708-C721. [Crossref] [PubMed]
14. Nevitt T, Ohrvik H, Thiele DJ. Charting the travels of copper in eukaryotes from yeast to mammals. Biochim Biophys Acta 2012;1823:1580-93. [Crossref] [PubMed]
15. Lutsenko S. Human copper homeostasis: a network of interconnected pathways. Curr Opin Chem Biol 2010;14:211-7. [Crossref] [PubMed]
16. Gong Y, Zhu W, Sun M, et al. Bioinformatics Analysis of Long Non-coding RNA and Related Diseases: An Overview. Front Genet 2021;12:813873. [Crossref] [PubMed]
17. Wu D, Li R, Liu J, et al. Long Noncoding RNA LINC00467: Role in Various Human Cancers. Front Genet 2022;13:892009. [Crossref] [PubMed]
18. Morovat P, Morovat S, Hosseinpour M, et al. Survival-based bioinformatics analysis to identify hub long non-coding RNAs along with lncRNA-miRNA-mRNA network for potential diagnosis/prognosis of thyroid cancer. J Cell Commun Signal 2023;17:639-55. [Crossref] [PubMed]
19. Qin Y, Zhang D, Zhang H, et al. Construction of a ferroptosis-related five-lncRNA signature for predicting prognosis and immune response in thyroid carcinoma. Cancer Cell Int 2022;22:296. [Crossref] [PubMed]
20. Yu G, Wang LG, Han Y, et al. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 2012;16:284-7. [Crossref] [PubMed]
21. Walter W, Sánchez-Cabo F, Ricote M. GOplot: an R package for visually combining expression data with functional analysis. Bioinformatics 2015;31:2912-4. [Crossref] [PubMed]
22. Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453-7. [Crossref] [PubMed]
23. Klempner SJ, Fabrizio D, Bane S, et al. Tumor Mutational Burden as a Predictive Biomarker for Response to Immune Checkpoint Inhibitors: A Review of Current Evidence. Oncologist 2020;25:e147-59. [Crossref] [PubMed]
24. Jiang P, Gu S, Pan D, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med 2018;24:1550-8. [Crossref] [PubMed]
25. Geeleher P, Cox N, Huang RS. pRRophetic: an R package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PLoS One 2014;9:e107468. [Crossref] [PubMed]
26. Linkermann A, Stockwell BR, Krautwald S, et al. Regulated cell death and inflammation: an auto-amplification loop causes organ failure. Nat Rev Immunol 2014;14:759-67. [Crossref] [PubMed]
27. Vanden Berghe T, Linkermann A, Jouan-Lanhouet S, et al. Regulated necrosis: the expanding network of non-apoptotic cell death pathways. Nat Rev Mol Cell Biol 2014;15:135-47. [Crossref] [PubMed]
28. Tang D, Kang R, Berghe TV, et al. The molecular machinery of regulated cell death. Cell Res 2019;29:347-64. [Crossref] [PubMed]
29. Qi X, Li Q, Che X, et al. Application of Regulatory Cell Death in Cancer: Based on Targeted Therapy and Immunotherapy. Front Immunol 2022;13:837293. [Crossref] [PubMed]
30. Li SR, Bu LL, Cai L. Cuproptosis: lipoylated TCA cycle proteins-mediated novel cell death pathway. Signal Transduct Target Ther 2022;7:158. [Crossref] [PubMed]
31. Zhou L, Huang X, Zhang Y, et al. PSMG3-AS1 enhances glioma resistance to temozolomide via stabilizing c-Myc in the nucleus. Brain Behav 2022;12:e2531. [Crossref] [PubMed]
32. Huang S, Chen J, Gao X, et al. LncRNAs PSMG3-AS1 and MEG3 negatively regulate each other to participate in endometrial carcinoma cell proliferation. Mamm Genome 2022;33:502-7. [Crossref] [PubMed]
33. Man S, Li X, Zhu W. miR-4417 targets lncRNA PSMG3-AS1 to suppress cell invasion and migration in cervical squamous cell carcinoma. Oncol Lett 2021;22:502. [Crossref] [PubMed]
34. Shen W, Hong X, Jin C, et al. LncRNA PSMG3AS1 promotes proliferation of non-small cell lung cancer cells by sponging miR-613 to upregulate SphK1. Cell Cycle 2021;20:829-38. [Crossref] [PubMed]
35. Huang Y, Li X, Chen W, et al. Analysis of the prognostic significance and potential mechanisms of lncRNAs associated with m6A methylation in papillary thyroid carcinoma. Int Immunopharmacol 2021;101:108286. [Crossref] [PubMed]
36. Bhat AA, Afzal M, Moglad E, et al. lncRNAs as prognostic markers and therapeutic targets in cuproptosis-mediated cancer. Clin Exp Med 2024;24:226. [Crossref] [PubMed]
37. Huang Y, Meng L, Wang X, et al. Cuproptosis and lncRNAs: Interplay in cancer immunity, prognosis, and chemotherapy sensitivity. Biochem Biophys Res Commun 2026;811:153461. [Crossref] [PubMed]
38. Zhou M, Tang J, Huang G, et al. Prognostic Significance and Immune Landscape of a Cuproptosis-Related LncRNA Signature in Ovarian Cancer. Biomedicines 2024;12:2640. [Crossref] [PubMed]
39. Lei H, Zhou Z, Liu C, et al. Characterization of Cuproptosis-Related LncRNAs Prognostic Signature and Identification of LINC02285 as a Novel Biomarker for Ovarian Cancer. J Inflamm Res 2025;18:7419-42. [Crossref] [PubMed]
40. Hao S, Gao M, Li Q, et al. Machine learning predicts cuproptosis-related lncRNAs and survival in glioma patients. Sci Rep 2024;14:22323. [Crossref] [PubMed]
41. Lei B, Zhan A, You G, et al. A novel cuproptosis-associated LncRNA model predicting prognostic and immunotherapy response for glioma. Discov Oncol 2025;16:1089. [Crossref] [PubMed]
42. Zhuo Z, Guo Z, Peng H, et al. Cuproptosis-Related lncRNA Risk Model for Hepatocellular Carcinoma Prognosis and Immunotherapy Response. Dig Dis 2025; Epub ahead of print. [Crossref]
43. Liu C, Zhou X, Pan Y, et al. Pyruvate carboxylase promotes thyroid cancer aggressiveness through fatty acid synthesis. BMC Cancer 2021;21:722. [Crossref] [PubMed]
44. Strickaert A, Corbet C, Spinette SA, et al. Reprogramming of Energy Metabolism: Increased Expression and Roles of Pyruvate Carboxylase in Papillary Thyroid Cancer. Thyroid 2019;29:845-57. [Crossref] [PubMed]
45. Ding X, Zhang W, Li S, et al. The role of cholesterol metabolism in cancer. Am J Cancer Res 2019;9:219-27.
46. Wang Y, Liu C, Hu L. Cholesterol regulates cell proliferation and apoptosis of colorectal cancer by modulating miR-33a-PIM3 pathway. Biochem Biophys Res Commun 2019;511:685-92. [Crossref] [PubMed]
47. Liu Z, Liu X, Liu S, et al. Cholesterol promotes the migration and invasion of renal carcinoma cells by regulating the KLF5/miR-27a/FBXW7 pathway. Biochem Biophys Res Commun 2018;502:69-75. [Crossref] [PubMed]
48. Goossens P, Rodriguez-Vita J, Etzerodt A, et al. Membrane Cholesterol Efflux Drives Tumor-Associated Macrophage Reprogramming and Tumor Progression. Cell Metab 2019;29:1376-1389.e4. [Crossref] [PubMed]
49. Ma X, Bi E, Lu Y, et al. Cholesterol Induces CD8(+) T Cell Exhaustion in the Tumor Microenvironment. Cell Metab 2019;30:143-156.e5. [Crossref] [PubMed]
50. Lu G, Yu X, Jiang W, et al. Alterations of Gut Microbiome and Metabolite Profiles Associated With Anabatic Lipid Dysmetabolism in Thyroid Cancer. Front Endocrinol (Lausanne) 2022;13:893164. [Crossref] [PubMed]
51. Kim J, Kim MK, Baek KH, et al. Repeated Low High-Density Lipoprotein Cholesterol and the Risk of Thyroid Cancer: A Nationwide Population- Based Study in Korea. Endocrinol Metab (Seoul) 2022;37:303-11. [Crossref] [PubMed]
52. Prete A, Lo AS, Sadow PM, et al. Pericytes Elicit Resistance to Vemurafenib and Sorafenib Therapy in Thyroid Carcinoma via the TSP-1/TGFβ1 Axis. Clin Cancer Res 2018;24:6078-97. [Crossref] [PubMed]
53. Meehan EV, Wang K. Interleukin-17 Family Cytokines in Metabolic Disorders and Cancer. Genes (Basel) 2022;13:1643. [Crossref] [PubMed]
54. Xing M. Molecular pathogenesis and mechanisms of thyroid cancer. Nat Rev Cancer 2013;13:184-99. [Crossref] [PubMed]
55. Sastre-Perona A, Santisteban P. Role of the wnt pathway in thyroid cancer. Front Endocrinol (Lausanne) 2012;3:31. [Crossref] [PubMed]
56. Noureldine SI, Tufano RP. Association of Hashimoto's thyroiditis and thyroid cancer. Curr Opin Oncol 2015;27:21-5. [Crossref] [PubMed]
57. Elbehi AM. The challenges and opportunities of applying tumour mutational burden analysis to precision cancer medicine. Camb Prism Precis Med 2025;3:e3. [Crossref] [PubMed]
58. Gandara DR, Agarwal N, Gupta S, et al. Tumor mutational burden and survival on immune checkpoint inhibition in >8000 patients across 24 cancer types. J Immunother Cancer 2025;13:e010311. [Crossref] [PubMed]
59. Niknafs N, Najjar M, Dennehy C, et al. Of Context, Quality, and Complexity: Fine-Combing Tumor Mutational Burden in Immunotherapy-Treated Cancers. Clin Cancer Res 2025;31:2850-63. [Crossref] [PubMed]
60. Li X, Jian J, Zhang A, et al. The role of immune cells and immune related genes in the tumor microenvironment of papillary thyroid cancer and their significance for immunotherapy. Sci Rep 2024;14:18125. [Crossref] [PubMed]
61. Mohanty A, Afkhami M, Reyes A, et al. Exploring markers of immunoresponsiveness in papillary thyroid carcinoma and future treatment strategies. J Immunother Cancer 2024;12:e008505. [Crossref] [PubMed]
62. Golestanifar A, Seyedtaghia MR, Mousavi P. Integrated bulk and single-cell transcriptomic profiling reveals NECTIN-TIGIT interaction underlies T cell exhaustion in papillary thyroid carcinoma. Clin Exp Med 2025;26:13. [Crossref] [PubMed]
63. Saadh MJ, Bishoyi AK, Rekha MM, et al. Dual roles of long non-coding RNAs in thyroid cancer: regulation of programmed cell death pathways. Med Oncol 2025;42:217. [Crossref] [PubMed]
64. Mercer TR, Dinger ME, Mattick JS. Long non-coding RNAs: insights into functions. Nat Rev Genet 2009;10:155-9. [Crossref] [PubMed]
65. Marques AC, Hughes J, Graham B, et al. Chromatin signatures at transcriptional start sites separate two equally populated yet distinct classes of intergenic long noncoding RNAs. Genome Biol 2013;14:R131. [Crossref] [PubMed]
66. Statello L, Guo CJ, Chen LL, et al. Gene regulation by long non-coding RNAs and its biological functions. Nat Rev Mol Cell Biol 2021;22:96-118. [Crossref] [PubMed]
67. Salmena L, Poliseno L, Tay Y, et al. A ceRNA hypothesis: the Rosetta Stone of a hidden RNA language? Cell 2011;146:353-8. [Crossref] [PubMed]
68. Zhang J, Huang J, Chen W, et al. miR-143-3p Targets lncRNA PSMG3-AS1 to Inhibit the Proliferation of Hepatocellular Carcinoma Cells. Cancer Manag Res 2020;12:6303-9. [Crossref] [PubMed]
69. Xuan DTM, Yeh IJ, Su CY, et al. Prognostic and Immune Infiltration Value of Proteasome Assembly Chaperone (PSMG) Family Genes in Lung Adenocarcinoma. Int J Med Sci 2023;20:87-101. [Crossref] [PubMed]
70. Fuja TJ, Lin F, Osann KE, et al. Somatic mutations and altered expression of the candidate tumor suppressors CSNK1 epsilon, DLG1, and EDD/hHYD in mammary ductal carcinoma. Cancer Res 2004;64:942-51. [Crossref] [PubMed]
71. Cavatorta AL, Fumero G, Chouhy D, et al. Differential expression of the human homologue of drosophila discs large oncosuppressor in histologic samples from human papillomavirus-associated lesions as a marker for progression to malignancy. Int J Cancer 2004;111:373-80. [Crossref] [PubMed]
72. Lin HT, Steller MA, Aish L, et al. Differential expression of human Dlg in cervical intraepithelial neoplasias. Gynecol Oncol 2004;93:422-8. [Crossref] [PubMed]
73. Min J, Jin D, Zhang F, et al. DLG1-AS1 is activated by MYC and drives the proliferation and migration of hepatocellular carcinoma cells through miR-497-5p/SSRP1 axis. Cancer Cell Int 2021;21:16. [Crossref] [PubMed]
74. Zhang H, Shi X, Ge Z, et al. PBX3-activated DLG1-AS1 can promote the proliferation, invasion, and migration of TNBC cells by sponging miR-16-5p. Mol Ther Oncolytics 2022;25:201-10. [Crossref] [PubMed]
75. Huang Y, Zhang K, Li Y, et al. The DLG1-AS1/miR-497/YAP1 axis regulates papillary thyroid cancer progression. Aging (Albany NY) 2020;12:23326-36. [Crossref] [PubMed]
76. He T, Wang H, Sun J, et al. Altered expression of DLG1-AS1 distinguished papillary thyroid carcinoma from benign thyroid nodules. BMC Endocr Disord 2019;19:122. [Crossref] [PubMed]
77. Jiang C, Zhang X, Qu T, et al. The prediction of pCR and chemosensitivity for breast cancer patients using DLG3, RADL and Pathomics signatures based on machine learning and deep learning. Transl Oncol 2024;46:101985. [Crossref] [PubMed]
78. Yang Y, Wang Y, Wang C, et al. Identification of hub genes of Parkinson's disease through bioinformatics analysis. Front Neurosci 2022;16:974838. [Crossref] [PubMed]
79. Wang D, Yang Y, He Y, et al. Natural History and Prognostic Model of Untreated Papillary Thyroid Cancer: A SEER Database Analysis. Cancer Control 2024;31:10732748241253956. [Crossref] [PubMed]
80. Zhou B, Liu J, Yang Y, et al. Ultrasound-based nomogram to predict the recurrence in papillary thyroid carcinoma using machine learning. BMC Cancer 2024;24:810. [Crossref] [PubMed]
81. Zheng G, Wei P, Li D, et al. A Deep Learning Survival Model for Evaluating the Survival Prognosis of Papillary Thyroid Cancer: A Population-Based Cohort Study. Ann Surg Oncol 2025;32:4780-9. [Crossref] [PubMed]
82. Xiang L, Zhao JH, Tang Y, et al. Prognostic prediction of patients having classical papillary thyroid carcinoma with a 4 mRNA-based risk model. Medicine (Baltimore) 2024;103:e38472. [Crossref] [PubMed]
83. Wang Y, Li X, Huang Y, et al. The Prognostic Value and Potential Immune Mechanisms of lncRNAs Related to Immunogenic Cell Death in Papillary Thyroid Carcinoma. J Inflamm Res 2024;17:1995-2008. [Crossref] [PubMed]