A Unique Plasma Protein Signature Characterizes Squamous Cell Carcinoma of the Oral Tongue in Young Adults

ABSTRACT

Background

The incidence of squamous cell carcinoma of the oral tongue (SCCOT) among young adults is increasing in several regions of the world. Age‐dependent differences in the biology of SCCOT have been suspected.

Methods

We used the Olink Explore 3072 high‐throughput platform to comprehensively quantify plasma proteins in 24 young (≤ 40 years of age) and 50 old (> 50 years of age) individuals. Eight young and 20 old individuals were diagnosed with SCCOT, four young and nine old individuals with SCC at other oral subsites (SCCOO), and the remaining 12 young and 21 old individuals were healthy controls. Dimension reduction analysis, differential expression analysis, and functional enrichment analysis were performed to characterize young patient‐specific biological signatures.

Results

Plasma levels of 2923 proteins were obtained. Principal component analysis indicated age‐related expression patterns. Comparing young patients to young controls/old patients/old controls, differential abundance analysis showed that increases in protein levels of Peroxiredoxin 2 (PRDX2) and C‐C motif chemokine ligand 26 (CCL26) and a decrease in Kallikrein related peptidase 4 (KLK4) were young patient‐specific. Reactome pathway enrichment analysis identified “Cellular response to chemical stress,” “Detoxification of reactive oxygen species” and “Cellular responses to stimuli” as the top altered pathways in young patients with SCCOT.

Conclusions

Abnormal cellular stress and aberrant immune regulation could thus be linked to cancer development in young patients. The unique plasma proteomic signature observed in young patients with SCCOT suggests that they constitute a specific group with distinct underlying pathophysiological processes.

1. Background

Oral cancer represented approximately 2% of all new global cancer cases in 2020 [1], and the incidence varies by sex, age, geographic regions, and sociocultural lifestyles. The disease is most prevalent in male patients over 50 years of age [1] and correlates with chronic exposure to risk factors such as tobacco smoking, alcohol consumption, and betel nut chewing [2]. Among the 377 731 new cases worldwide in 2020, 65.8% were reported in Asia, followed by Europe (17.3%) and North America (7.3%). In these regions, patients diagnosed before the age of 40 accounted for 13.4%, 2.2%, and 3.1% of the cases, respectively [1].

Most oral cancers are derived from the epithelium of the oral tongue, floor of mouth, buccal mucosa, gingiva, retromolar trigone, and hard palate, and are collectively named oral squamous cell carcinomas (OSCC). SCC of the oral tongue (SCCOT) is associated with higher mortality than SCC in other oral subsites (SCCOO) [3]. In contrast to the overall decline/stabilization in the incidence of SCCOO over the last few decades, there has been a general increase in SCCOT incidence worldwide [4, 5, 6]. In certain regions, such as Australia, Sweden, and the United States of America, the annual increase in incidence rate was higher in young adults than in the elderly [6]. The changing epidemiology may reflect changes in underlying exposures that are still to be defined [5].

Proteins circulating in the blood are secreted by tissues in the body or leaked from cell damage and death, harboring information on the systemic physiological state, future disease risk, and lifestyle behaviors [7, 8]. Although the role of many proteins in the circulation remains to be ascribed, a growing number of studies have applied circulating protein‐based proteomic assays to identify signatures of aging, cancer, and cardiometabolic diseases [7, 8, 9]. It is worth noting that aging is a non‐linear process characterized by waves of changes in plasma proteins reflecting complex shifts in biological processes [9].

In this study, we speculated that plasma protein levels differ between young and old patients due to factors besides aging, reflecting underlying differences in the nature of the tumor in young versus old patients. By comparing plasma protein profiles among four groups of participants (young patients, old patients, young controls and old controls), we aimed to capture young patient‐specific imprints that could yield insights into disease biology, especially in those patients with SCCOT.

2. Materials and Methods

2.1. Patient Recruitment

A total of 41 patients diagnosed with OSCC at Norrland's University Hospital in Sweden were recruited for this study. Twelve patients aged 27–39 years were grouped as young patients (YouPat) and 29 patients aged 50–86 as old patients (OldPat). Patients aged 41 to 49 were excluded to provide a clear‐cut distinction for age classification. Among the YouPat, eight had tumors located in the oral tongue (SCCOT), and the remaining four were grouped as SCCOO (two gingiva, one floor of mouth and one buccal mucosa). Among the OldPat, 21 had SCCOT and 10 had SCCOO (two in the floor of mouth and eight in the gingiva). As controls, 12 young (20–40 years, YouCon) and 21 old healthy individuals (51–83 years, OldCon) were included (Table S1). This study was approved by the Regional Ethics Review Board, Umeå, Sweden (Dnr 08‐003M and 2012‐131‐33M) and performed in accordance with the Declaration of Helsinki. Written informed consent to participate was obtained from all study subjects.

2.2. Plasma Protein Quantification and Data Preprocessing

Heparin plasma samples from 41 patients and 33 controls were sent to the Clinical biomarker facility, Science for Life Laboratory (Uppsala, Sweden) for protein quantification using the Olink Explore 3072 high‐throughput proteomics platform (Olink Proteomics, Uppsala, Sweden). A total of 2944 protein assays were carried out on each sample based on Olink's Proximity Extension Assay (PEA) technology coupled with next‐generation sequencing. The Olink Explore 3072 consists of eight 384‐plex protein panels. Six proteins are repeatedly measured by four different panels. Results are presented as normalized protein expression (NPX), an arbitrary unit on a log2 scale. Internal and external controls were included in each run. The limit of detection (LOD) was calculated separately for each Olink assay and sample plate. Values below the LOD were set to the LOD.

2.3. Dimension Reduction Analysis

Unsupervised principal component analysis (PCA) was performed using Simca 17 (MKS Data analytics Solutions, Umea, Sweden) to overview sample distributions based on the levels of all measured proteins. Sample grouping patterns related to different phenotypic characteristics, such as age, sex, disease, anatomic subsite, tumor, node, metastasis (TNM) staging, and fasting status were examined by a PCA score plot.

2.4. Differential Abundance and Functional Enrichment Analysis

Differences among groups were assessed using the nonparametric Kruskal–Wallis test. When significant differences were detected, Dunn's post hoc test with Benjamini‐Hochberg correction was applied to identify group differences. All analyses were performed using R (version 4.5.0), utilizing the dunn.test and kruskal.test functions. Top differentially abundant proteins were selected according to adjusted p‐value (< 0.05) and fold change (≥ 1.5 or ≤ 0.67). Functional enrichment analysis was performed using g:Profiler [10].

2.5. The Cancer Genome Atlas (TCGA) Data Analysis

Gene expression data from The Cancer Genome Atlas (TCGA) OSCC cohort were accessed via the Xena Functional Genomics Explorer (www.xenabrowser.net). Differential expression analyses between tumor and normal tissue samples were conducted using the platform's default settings. Since the platform provides unadjusted p‐values, these were further adjusted using the Bonferroni correction.

2.6. Global Gene Set Enrichment Analysis

Global functional enrichment was performed using the Gene Set Enrichment Analysis software (GSEA, v4.3.2, Broad institute) that determines whether a priori defined gene set shows statistically significant concordant differences between two biological states [11]. In this study, a total of 1654 gene sets derived from the canonical Reactome pathway database were selected for functional enrichment analysis. Gene sets with size < 15 were excluded from analysis. Proteins measured by Olink served as the background against which to test for significance. False discovery rate (FDR) was estimated using the Benjamini–Hochberg approach. Single‐sample GSEA (ssGSEA) is an extension of GSEA [12] that calculates a gene set enrichment score per sample to facilitate comparisons of biological features. We performed ssGSEA through the GenePattern web‐tool [13].

2.7. Statistical Analysis

Clinicopathological characteristics between YouPat and OldPat were compared using Fisher's exact test. The Kaplan–Meier method with log‐rank test was used to estimate the impact of age at diagnosis on patient survival. Correlations between continuous variables were assessed using nonparametric Spearman correlation analysis. The correlation coefficient (rho) was calculated to evaluate the strength of correlation. All tests were conducted in IBM SPSS Statistics 28 (IBM Corp., Armonk, NY, USA). A two‐sided p‐value < 0.05 was considered significant.

3. Results

3.1. Clinical Features

A detailed description of all the 74 participants is provided in Table S1. Comparing clinicopathological characteristics between YouPat and OldPat, no significant difference was identified except for treatment (p < 0.05, Table 1). Kaplan–Meier survival curves showed that YouPat and OldPat had similar survival outcomes (Figure S1).

TABLE 1.

Clinicopathological characteristics of patients with oral squamous cell carcinoma.

Characteristics		Young patients (n = 12)	Old patients (n = 29)	p (Fisher's exact test)
Smoking status	Never, party or former smoker	8	18	1.000
	Current smoker	4	11
Alcohol consumption	No	1	7	0.394
	Yes	11	20
Primary tumor site	Oral tongue	8	20	1.000
	Other sites in oral cavity	4	9
Tumor size (pathologic)	T1, T2	9	20	1.000
	T3, T4	3	9
Lymph node involvement (pathologic)	Negative	7	19	0.730
	Positive	5	10
TNM stage (pathologic)	I, II	6	16	1.000
	III, IVa, IVb	6	13
Treatment	Surgery	0	9	0.039
	Radiotherapy or Surgery + radiotherapy/chemotherapy	12	20

3.2. Plasma Proteins Showed an Age‐Related Expression Profile

Among the 2944 assays included in Olink Explore 3072, three assays (KNG1, SMAD1 and ARHGAP25) did not meet Olink's batch release quality control criteria and were therefore excluded. Based on the levels of the 2923 proteins measured by the remaining 2941 assays, PCA showed that YouPat and YouCon had similar plasma protein profiles. Protein profiles in OldPat and OldCon were also similar (Figure 1A). However, we found that young individuals were clustered in the upper half part of the plot, whereas old individuals were more commonly grouped in the lower half, indicating age‐related differences (Figure 1B). No grouping pattern was found for other factors such as sex, tumor anatomic subsite, clinical stage, or fasting status (Figure S2A–D).

FIGURE 1.

Plasma protein profiles in 74 individuals. PCA plots (two components, R2X = 0.256, Q2X = 0.187) visualizing grouping patterns related to disease (A) and age (B). t [1] and t [2] are scores of the first two principal components. YouPat, Young patients; YouCon, Young controls; OldPat, Old patients; OldCon, Old controls.

3.3. Differentially Abundant Proteins Associated With Aging

Since an age‐related profile was identified by PCA, we first investigated differentially abundant proteins between young and old individuals. Proteins showing significant difference between YouCon and OldCon, as well as between YouPat and OldPat (Dunn's post hoc test, adjusted p‐value < 0.05, fold change ≥ 1.5 or ≤ 0.67) were regarded as age‐associated proteins. A total of 39 proteins were identified (Table S2) and gene ontology enrichment analysis revealed that plasma levels of proteins involved in “Regulation of multicellular organismal process,” “Signaling,” “Developmental process,” “Follicle‐stimulating hormone signaling pathway,” “Regulation of the force of heart contraction,” “Circulatory system development,” and “Blood circulation” increased with age (Table S3).

3.4. Young Patient‐Specific Differentially Abundant Proteins

Proteins showing significant differences (adjusted p‐value < 0.05, fold change ≥ 1.5 or ≤ 0.67) between YouPat and all the other three groups of samples were considered as YouPat‐specific alterations. Three proteins were identified (Table 2): PRDX2 (Peroxiredoxin 2), CCL26 (C‐C motif chemokine ligand 26) and KLK4 (Kallikrein related peptidase 4). PRDX2 and CCL26 were increased, whereas KLK4 was decreased in YouPat compared to the other three groups (Figure 2A).

TABLE 2.

List of the young patient‐specific differentially expressed proteins.

Protein	YouPat versus YouCon		YouPat versus OldPat		YouPat versus OldCon
	Fold change	Adjusted p	Fold change	Adjusted p	Fold change	Adjusted p
PRDX2	2.034	0.001	1.663	0.032	1.956	0.002
CCL26	1.648	0.006	1.661	0.001	1.759	0.003
KLK4	0.465	0.002	0.411	0.000	0.371	0.000

Abbreviations: OldCon, old controls; OldPat, old patients; YouCon, young controls; YouPat, young patients.

FIGURE 2.

Young patient‐specific alterations in plasma proteins. (A) Boxplots showing protein levels across four groups of samples. (B) Boxplots showing ssGSEA scores of the Reactome pathways across four groups. YouPat, Young patients; YouCon, Young controls; OldPat, Old patients; OldCon, Old controls; SCCOT, Young patients with squamous cell carcinoma of the oral tongue.

3.5. The Cancer Genome Atlas (TCGA) Data

In the TCGA OSCC cohort, RNA sequencing data were available for 312 tumor samples and 30 non‐tumor controls. Compared to controls, PRDX2 was significantly downregulated in tumors (adjusted p < 0.05). No significant differences were observed for CCL26 and KLK4. When comparing tumors from young versus old patients, none of the genes showed significant differential expression (Figure S3).

3.6. Young Patient‐Specific Functional Changes in Plasma Proteins

Global functional differences in plasma proteins across all group pairs were investigated using GSEA (Table S4). When comparing YouPat to the other three groups, only two Reactome pathways, “Molecules associated with elastic fibers” and “Elastic fiber formation,” showed YouPat‐specific enrichment (FDR < 0.05). However, when excluding patients with SCCOO, a total of 16 Reactome pathways demonstrated YouPat‐specific enrichment (FDR < 0.05), with 14 upregulated and two downregulated in YouPat (Table 3). The top three pathways were “Cellular response to chemical stress,” “Detoxification of reactive oxygen species” and “Cellular responses to stimuli,” followed by “Molecules associated with elastic fibers.” The stress response pathways have several contributing proteins in common, such as PRDX2, CCS (Copper Chaperone for Superoxide Dismutase), CAT (Catalase) and SOD1 (Superoxide Dismutase 1), all involved in oxidative stress responses (Table 4). According to the ssGSEA score (Table S5), it is visualized that the top three pathways were activated in Young SCCOT patients compared to the other three groups, while the elastic fiber pathway was suppressed (Figure 2B). In patients with SCCOT (both young and old), Spearman correlation analysis showed that each of the three stress response pathways were negatively correlated with the elastic fiber pathway (p < 0.01, n = 28), with rho values ranging from −0.541 to −0.676. The strongest correlation was seen between “Detoxification of reactive oxygen species” and “Molecules associated with elastic fibers” (rho = −0.676, p < 0.001).

TABLE 3.

List of the reactome pathways showing young SCCOT patient‐specific enrichment.

Gene set	YouPat versus YouCon		YouPat versus OldPat		YouPat versus OldCon
	NES	FDR	NES	FDR	NES	FDR
Cellular response to chemical stress	2.296	0.000	2.229	0.001	2.500	0.000
Detoxification of reactive oxygen species	2.138	0.001	2.248	0.001	2.160	0.003
Cellular responses to stimuli	2.038	0.004	1.918	0.019	2.176	0.002
Molecules associated with elastic fibers	−2.105	0.004	−2.160	0.004	−2.095	0.000
Elastic fiber formation	−2.002	0.007	−2.148	0.002	−2.116	0.001
Cell cycle	1.928	0.012	1.796	0.033	1.941	0.013
Deubiquitination	1.938	0.013	2.166	0.000	2.134	0.004
Cell cycle mitotic	1.905	0.013	1.864	0.018	1.774	0.027
Antigen processing ubiquitination proteasome degradation	1.909	0.014	2.004	0.008	1.966	0.012
KEAP1 NFE2L2 pathway	1.891	0.015	1.877	0.017	2.046	0.010
UB specific processing proteases	1.793	0.040	2.018	0.008	2.183	0.002
m phase	1.759	0.042	1.722	0.042	1.739	0.032
SARS‐CoV‐1‐host interactions	1.760	0.044	1.965	0.013	1.958	0.012
HSP90 chaperone cycle for steroid hormone receptors shr in the presence of ligand	1.720	0.045	1.896	0.020	1.875	0.015
Mitotic metaphase and anaphase	1.762	0.047	1.775	0.037	1.926	0.013
DDX58 IFIH1 mediated induction of interferon alpha beta	1.707	0.049	1.815	0.029	1.916	0.012

Abbreviations: FDR, false discovery rate; NES, normalized enrichment score; OldCon, old controls; OldPat, old patients; YouCon, young controls; YouPat, young patients.

TABLE 4.

List of the proteins contributing to gene set enrichment.

Gene set	Proteins for core enrichment
Cellular response to chemical stress	PRDX2, PSME2, CCS, CAT, SOD1, NUDT2, FABP1, BLVRB, PSMD9, TALDO1, UFD1, PSME1, GSTA1, TRIM21, BACH1, PRDX5, GSTA3, TXN, PRDX1, CCL2, PGD, NFKB1, NCF2, PRDX6, ATOX1, AKT2, CDKN1A
Detoxification of reactive oxygen species	PRDX2, CCS, CAT, SOD1, NUDT2, PRDX1, TXN, PRDX5, NCF2, PRDX6, ATOX1
Cellular response to stimuli	PRDX2, FKBP4, PSME2, DNAJB1, CCS, IMPACT, DNAJA4, CAT, SOD1, SEC31A, ST13, TPP1, NUDT2, FABP1, BLVRB, PSMD9, TALDO1, UFD1, PSME1, AKT1S1, DCTN1, GSTA1, ADD1, TRIM21, EGLN1, BACH1, PRDX5, GSTA3, HSBP1, BAG4, HSPA1A, CDC26, TXN, PRDX1, CCL2, PGD, HDGF, NFKB1, NCF2, HSPA2, PRDX6, CXCL8, ATOX1, AKT2, CDKN1A, STIP1, IL6, CDC27, LAMTOR5, MAPK9, TXNRD1, ZBTB17
Molecules associated with elastic fibers	BMP10, ITGAV, ITGB1, ITGB6, LTBP2, FBLN2, ITGB5, EFEMP1, MFAP4

4. Discussion

An increasing incidence of oral cancer among young adults is seen in certain regions of the world. Notably, the SCCOT subtype is more prevalent among young than old patients [14]. Given the shorter exposure time to conventional risk factors in the lower age group, age‐dependent differences in pathophysiology have been suspected. However, despite intensive investigations, no convincing findings for differences in etiological triggers, clinical phenotypes, histopathological features, mutational patterns, or survival outcomes have yet been identified to characterize young patients [15, 16, 17, 18, 19, 20]. On the other hand, there is evidence for distinct biological behaviors, for example, the differences in patient outcome seen for copy number changes, which provide poor prognosis specifically in young SCCOT patients [17]. In the present comprehensive analysis of circulating proteins, we attempted to identify differences in plasma proteins between young and old patients with OSCC, which could reflect differences in the underlying pathophysiological processes occurring in these two age groups. The definition of “young adult” varies between studies, with the cutoff age ranging from 20 to 45 years [21]. In this study, we used the 40‐year cutoff to maximize the sample size of “young adult” and to ensure a clear‐cut distinction from the majority of patients aged 50 years and over.

Patterns in blood protein levels change during various physical conditions, including aging [7, 9], and we found that levels of plasma proteins related to “Regulation of multicellular organismal process,” “Signaling,” and “Developmental process” increased with age. The protein most significantly upregulated with age is Elastin (ELN). Elastic fibers are major components of the extracellular matrix in large arterial wall, and their remodeling occurs progressively with age [22]. Consistent with previous findings, elevated plasma levels of ELN, the core protein of elastic fibers, in old individuals may result from increased endothelial permeability, which facilitates the release of elastin into the bloodstream [22, 23].

Comparing young patients to other groups, we identified elevated CCL26 levels specifically in young patients. CCL26, also known as eotaxin‐3, is a chemokine primarily involved in immune cell trafficking and is best known for its role in allergic inflammation and eosinophilic disorders, such as asthma and atopic dermatitis [24]. CCL26 has also been implicated in tumor biology, particularly through its effects on the tumor immune microenvironment and cancer cell migration [25, 26, 27]. In OSCC, an association between CCL26 expression and an altered invasive phenotype, as well as a tumor immunosuppressive microenvironment, has been suggested [28]. Strikingly, it was recently reported that when comparing tumor tissues from young OSCC patients to those from old patients, increased expression of genes related to extracellular matrix remodeling, epithelial‐mesenchymal transition [29] and tumor microenvironment immunosuppression [29, 30] was observed in the young patients. Our finding that plasma level of CCL26 was selectively elevated in young OSCC patients may reflect a distinct immune response or the activation of specific oncogenic pathways that are more prevalent in this age group. This is especially noteworthy given that oral cancers in young individuals often occur in the absence of traditional risk factors such as tobacco or alcohol exposure, pointing toward a potentially unique disease biology in this subgroup.

Interestingly, KLK4, a member of the kallikrein family of serine proteases involved in processes such as proteolysis, hormone processing, extracellular matrix degradation, and immune cell infiltration [31], also shows a young‐patient‐specific alteration. Its downregulation in young patients suggests a specific disruption in protease signaling and extracellular matrix interactions. Consistent with findings by Estephan et al. [29], this may indicate a distinct molecular subtype of OSCC in younger individuals, characterized by impaired tissue remodeling and altered tumor–host interactions.

The top differentially expressed protein in the plasma of young patients is PRDX2, a well‐known antioxidant enzyme [32]. Excluding patients with SCCOO, GSEA showed that “Detoxification of reactive oxygen species” was one of the top three pathways demonstrating young patient‐specific alterations. Reactive oxygen species (ROS) are unstable molecules originating from oxygen during cellular metabolism [33]. They are essential signaling agents controlling diverse aspects of cellular behavior and are mediators of inflammation combating pathogens. However, ROS also damage DNA, proteins and lipids [33]. Levels of ROS are therefore carefully controlled by antioxidant metabolites, antioxidant enzymes and metal ion sequestration. Increased generation of oxidants that exceeds the ability to detoxify their harmful effect leads to oxidative stress, which participates in a wide range of diseases including cancer [34]. The increased “Detoxification of reactive oxygen species” pathway seen in young patients with SCCOT indicates that the level of oxidative stress is particularly high in this group of patients.

Notably, we found that “Detoxification of reactive oxygen species” and “Molecules associated with elastic fibers” were negatively associated. Elastic fibers are essential constituents of the extracellular matrix of higher vertebrates, providing elasticity and resilience, and are responsible for growth factor signaling and the maintenance of tissue homeostasis [35]. A well‐known source of damage to elastic fibers is oxidative stress, which impairs their structure and functions [36]. Thus, we speculate that increased oxidative stress, along with aberrant elastic fiber‐related signaling and tissue remodeling, may be associated with tumor development in young patients with SCCOT. Similarly, based on spatial transcriptomics analysis of tumor tissues, Marina et al. suggested that oxidative stress maintained by immunosuppression, may be a driver of SCCOT in young individuals [30].

In our study, we observed differential plasma protein abundance of PRDX2, CCL26, and KLK4 between young and old patients. However, analysis of gene expression data from the TCGA OSCC cohort did not reveal significant differences in the mRNA levels of these genes between the two age groups. This discrepancy may be due to differences between cohorts and/or the complex regulation of protein abundance. Protein levels are influenced not only by mRNA expressions, but also by post‐transcriptional mechanisms, protein stability, secretion dynamics, and systemic physiological factors. Nevertheless, from a systemic perspective, both our plasma proteomics analysis and previous tissue transcriptomics studies [29, 30] point toward the same overall conclusion.

One limitation of this study is the small sample size, which may reduce statistical power and limit the generalizability of the findings. Furthermore, detailed lifestyle information or environmental factors other than smoking and alcohol consumption history may confound interpretation of age‐associated molecular differences. Future studies with larger, well‐characterized cohorts are warranted to validate these findings and control for potential confounders. Such efforts could lead to a deeper understanding of the origin of young SCCOT and inform strategies for its prevention and treatment.

5. Conclusions

Based on a comprehensive analysis of circulating proteins, our study identified a unique plasma proteomic signature in young patients with SCCOT, suggesting distinct underlying pathophysiological processes specific to this patient group. Our plasma‐based proteomics analysis supports recent tissue‐based findings identified in this specific group of patients that have an increasing prevalence of disease for reasons that are as yet unknown.

Author Contributions

X.G. analyzed data, wrote, and edited the manuscript. P.J.C. interpreted data, wrote, and edited the manuscript. L.W., S.V.G., N.S. interpreted data, and reviewed the manuscript. N.A., B.E., and M.M. reviewed the manuscript. K.N. supervised the project, wrote, and edited the manuscript. All authors approved the final version of the manuscript.

Ethics Statement

This study was approved by the Regional Ethics Review Board, Umeå, Sweden (Dnr 08‐003M and 2012‐131‐33M) and performed in accordance with the Declaration of Helsinki.

Consent

Written informed consent to participate was obtained from all study subjects.

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/jop.70020.

Supporting information

Data S1: Supplementary Figure.

Data S2: Supplementary Tables.

Gu X., Coates P. J., Wang L., et al., “A Unique Plasma Protein Signature Characterizes Squamous Cell Carcinoma of the Oral Tongue in Young Adults,” Journal of Oral Pathology & Medicine 54, no. 8 (2025): 706–714, 10.1111/jop.70020.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.