Proteomic analysis of saliva reveals changes in proteomic profiles during colorectal cancer in an Iranian cohort Authors: Sama Rezasoltani, Ali Biabani, Bente Siebels, Hamid Asadzadeh Aghdaei, Georg Conrads, Hartmut Schlüter, Mohammad Mehdi Feizabadi Published: 2026 ABSTRACT IntroductionColorectal cancer (CRC) is common and associated with poor survival, and early detection is pivotal to improving patient outcomes. Therefore, identifying reliable biomarkers reflecting early-stage tumor development is of major clinical importance. This study aimed to identify salivary proteins with potential as non-invasive biomarkers associated with early-stage CRC.MethodsA total of 104 saliva samples were collected from individuals undergoing CRC screening, including patients diagnosed with CRC stage I and non-cancer controls (NCC), at Taleghani Hospital, Tehran, Iran, in this case–control study. A quantitative label-free proteomics approach was applied to determine global proteome differences between CRC patients and NCCs and to identify proteins with altered abundance. Proteins were extracted from saliva samples, digested with trypsin, and peptides were analyzed using LC–MS/MS on an Orbitrap Fusion mass spectrometer.ResultsA total of 2,456 salivary proteins were identified, of which 181 showed significantly different abundance between CRC patients and NCCs. Unsupervised clustering demonstrated partial separation of groups with influence from demographic variables. Members of the small proline-rich protein (SPRR) family emerged as consistent core markers distinguishing CRC from NCCs, while additional immune- and metabolism-associated proteins contributed to group differentiation. Demographic subgroup analyses, including age-, sex-, and smoking-stratified comparisons, revealed subgroup-associated differences in protein abundance patterns; however, several CRC-associated alterations remained detectable across subgroups, including middle-aged individuals. Gene set enrichment analysis indicated suppression of cytoskeletal and epithelial structural pathways alongside enrichment of protease regulatory pathways. Ingenuity Pathway Analysis identified an interconnected immune–metabolic network characterized by inflammatory signaling and oxidative stress–related processes. These alterations indicate immune-related and metabolic signaling changes associated with CRC.DiscussionThis study supports salivary proteomics as a non-invasive approach for early CRC detection, and the observed proteomic signatures may reflect systemic tumor–immune interactions associated with CRC. Data are available via ProteomeXchange with identifier PXD078610. Background Colorectal cancer (CRC) persists as a significant healthcare challenge, comprising approximately 6.1% of all cancers globally (1). The International Agency for Research on Cancer provides estimates of CRC incidence and mortality across world regions. In 2020 alone, over 1.9 million new cases of CRC and more than 930,000 CRC-related deaths were estimated worldwide (2). When CRC is detected in its early stages (localized, stages 0 to II), the five-year survival rate exceeds 80%. However, this rate drops dramatically in advanced stages, particularly stage IV disease (3). Compounding the issue, many instances of CRC stem from initially benign colonic adenomas (4). Given that the survival rate steadily declines with advancing cancer stages at diagnosis, the imperative of early detection and removal of these precancerous adenomas cannot be overstated (5). Consequently, widespread population screening and prevention initiatives are strongly advocated globally (6). Contemporary population-based CRC screening programs predominantly rely on fecal immunochemical testing (FIT) to identify high-risk individuals, a method credited with lowering CRC mortality rates. However, FIT’s sensitivity is limited, yielding inconsistent outcomes across diverse populations (7). Furthermore, this approach often leads to unnecessary colonoscopies, an invasive and costly procedure (8). Consequently, there is a pressing demand for more precise and non-invasive screening methodologies to identify individuals in the early stages of CRC. Today, the potential value of saliva in the early diagnosis of oral diseases, as well as cancers, diabetes, and other systemic disorders is increasingly recognized (9–11). With the rapid advancement in salivaomics, saliva is a rich source of biomarkers (11). Proteomic analyses have identified over 3,700 human proteins in saliva and more than 2,000 detectable microbial proteins; however, the total proteomic diversity likely exceeds these numbers due to methodological and detection constraints (12). Recent studies have uncovered shifts in both microbial patterns and protein profiles in saliva, particularly in relation to various diseases, notably cancers (13, 14). Meanwhile, CRC is strongly influenced by epigenetic alterations, with accumulating evidence highlighting the role of gut microbiota in disease development (15, 16). Our recent studies have revealed distinct microbial patterns in the saliva of individuals with CRC compared to non-cancer controls (NCCs) within the Iranian population (17, 18). Our findings suggest that a range of microbial biomarkers may serve as candidates for the early detection of CRC. Building upon our previous investigations, this study aims to characterize the salivary protein profiles of CRC patients compared to NCCs in an Iranian cohort. Our objective is to determine whether the protein patterns in the saliva of CRC patients differ significantly from those of NCCs and to explore the feasibility of utilizing salivary protein biomarkers as candidates for CRC screening. We also examine demographic characteristics of patients and NCCs. This emphasis is crucial, as factors such as sex, age, lifestyle, diet, overall health, and geographic context might influence salivary protein profiles. Methods Study population The present study was a case-control analysis, with all saliva samples procured from patients referred for standard screening colonoscopy between February 2020 and January 2022 at Taleghani Hospital in Tehran, Iran. Enrolled participants, aged ≥18 years, were recruited from patients referred to Taleghani Hospital from different regions of Iran. A total of 104 saliva samples were collected, comprising 51 NCCs and 53 CRC cases stage I, prior to colonoscopy procedures. Participant status (CRC vs. NCCs) was determined by colonoscopy and subsequently confirmed by histopathological examination. Patients exhibiting symptoms such as rectal bleeding, changes in bowel habits, anemia, and abdominal pain were included in the study, while exclusion criteria comprised individuals: 1) who had taken antibiotics within the past three months (with the exception of one participant who had used antibiotics for less than one week), 2) who had regularly used probiotics within the past three months (with the exception of one NCC who had used probiotics for one week), 3) who had undergone invasive medical interventions within the same period; 4) with a previous history of CRC, inflammatory or infectious intestinal diseases; 5) with gastrointestinal disorders like inflammatory bowel disease, irritable bowel syndrome, liver disorders, or non-alcoholic fatty liver disease; and 6) categorized as high-risk for CRC, such as those with familial adenomatous polyposis or other hereditary cancer syndromes. Demographic and clinical data, including age, sex, profession, physical activity, disease or surgical history, family history, smoking status, alcohol consumption, dietary patterns, viral infection status, probiotic use, and antibiotic usage, were collected through standardized questionnaires. NCC participants were chosen based on normal colonoscopy results and a negative personal and family history of gastrointestinal diseases. Patients were approached during their initial hospital visit to explain the research protocol and obtain informed consent. Approval for this case-control study was granted by the Clinical Research Ethics Committee of Shahid Beheshti University of Medical Sciences and the Ethics Committee of Taleghani Hospital, Tehran, Iran (IR.SBMU.RIGLD.REC.1398.039). Human saliva sampling Saliva specimens were prospectively collected between 08:00 - 12:00, with participants instructed to abstain from eating or drinking for at least 2 h prior to sampling. Each saliva sample was collected into a 20 mL Falcon tube kept on ice. Approximately 1–2 mL of unstimulated saliva was collected over a period of 5–10 min. Following collection, the specimens were promptly transferred to new 2 mL microcentrifuge tubes and immediately frozen at −80 °C to preserve their integrity for subsequent evaluation (19). Protein extraction and tryptic digestion 500 µL of saliva were mixed with an equal volume of 6 M urea buffer, and 2 µL of Sera-Mag™ carboxylate-modified hydrophobic and hydrophilic magnetic beads were added. Proteins were reduced using 0.5 µL of 1 M dithiothreitol (DTT) for 30 min at 60 °C, followed by alkylation with 2 µL of 0.5 M iodoacetamide for 30 min at 37 °C in the dark. Proteolytic digestion was initiated by adding 1 µL of a trypsin solution (0.5 µg/µL) to each sample, prepared by dissolving 20 µg of trypsin in 40 µL of 50 mM ammonium bicarbonate. Digestion proceeded overnight at 37 °C using a shaker. Following enzymatic digestion, the samples were washed two times with 100% acetonitrile (ACN). Based on the SP3 protocol by Hughes et al. (2019), following proteolytic digestion the magnetic beads with bound peptides were washed twice, with 70% ethanol to remove residual contaminants, ensuring that peptides remained immobilized on the beads during the cleanup step (20). Peptides were eluted with 50 µL of a solution containing 2% dimethyl sulfoxide (DMSO) and 1% formic acid (FA). The eluates were dried using vacuum centrifugation, reconstituted in 0.1% FA, and stored at -20 °C until LC-MS/MS analysis (20–22). LC-MS/MS analysis using data-dependent acquisition Chromatographic separation of peptides was achieved with a two-buffer system (buffer A: 0.1% FA in H2O, buffer B: 0.1% FA in ACN) on a nano-UHPLC (Dionex Ultimate 3000 UHPLC system, Thermo Fisher). Attached to the UHPLC was a peptide trap (100 µm x 20 mm, 100 Å pore size, 5 µm particle size, C18, nanoViper, Thermo Fisher) for online desalting and purification, followed by a 25 cm C18 reversed-phase column (75 µm x 250 mm, 130 Å pore size, 1.7 µm particle size, peptide BEH C18, nanoEase, Waters). Peptides were separated using an 80-minute method with a linear increase in ACN concentration from 2% to 30% ACN over 60 minutes, followed by high organic wash and re-equilibration. MS/MS measurements were performed on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher). Eluting peptides were ionized using a nano-electrospray ionization (nano-ESI) source with a spray voltage of 1.8kV and analyzed in data-dependent acquisition (DDA) mode. For each MS1 scan, ions were accumulated for a maximum of 120 ms or until an AGC target of 2 × 105 ions was reached. Fourier-transform-based mass analysis in the Orbitrap covered m/z 400–1,300 at a resolution of 120,000 at m/z = 200. Peptides with charge states between 2+ to 5+ above an intensity threshold of 1,000 were isolated within a 1.6 m/z isolation window in Top Speed mode for 3 s per cycle and fragmented with a normalized collision energy of 30% using higher-energy collisional dissociation (HCD). MS2 scans were acquired in the ion trap at a rapid scan rate, covering a mass range starting at m/z 120 and accumulated for 60 ms or to an AGC target of 1 × 105. Dynamic exclusion was set to 30 s (23, 24). Data analysis Raw MS data were processed using Proteome Discoverer (PD) version 3.1.0.638 (Thermo Fisher Scientific), with MS/MS spectra searched against the Homo sapiens reference proteome (UniProtKB/Swiss-Prot canonical, TaxID 9606, release 2023-11-08). A static modification of cysteine (carbamidomethylation) was specified, and variable modifications included methionine oxidation and pyroglutamate formation. Trypsin was specified as the proteolytic enzyme, allowing a maximum of two missed cleavages. A strict false discovery rate (FDR) threshold of 1% was applied at the peptide-spectrum match (PSM) and protein levels to ensure high-confidence identifications. Label-free quantification (LFQ) was performed using the Minora Feature detector. The resulting protein abundances were log2-transformed and normalized using column-wise median normalization in Perseus (version 2.0.11) (23, 24). The large number of samples made it necessary to split the samples into different preparation and LC-MS/MS cohorts. Resulting batch effects were removed with the Batch-Effect Reduction Trees (BERT) algorithm, using the reference based method, which operates on data matrices with missing values and thus avoids listwise deletion (25). For statistical analysis, a two-sided Student’s t-test was applied. P-values were adjusted for multiple testing using the Benjamini-Hochberg false discovery rate (FDR) procedure. Proteins with an adjusted p-value (q-value) < 0.05 were considered statistically significant. Visualizations were performed in R Studio (Version 2025.05.0 + 496) using ggplot and pHeatmap. Gene set enrichment analysis (GSEA) was performed to identify biological pathways and functional categories associated with differential protein abundance between groups and was conducted in R using the clusterProfiler package (26). Gene sets were obtained from the Gene Ontology (GO) database, including the Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) ontologies. Gene sets containing fewer than 15 or more than 500 genes were excluded from the analysis. Statistical significance was assessed using Benjamini-Hochberg multiple testing correction, and enriched terms were evaluated based on the false discovery rate (FDR) q-value. Ingenuity pathway analysis (IPA, version 153384343, QIAGEN) was performed using proteins with nominal p-values < 0.05 and an absolute log2 fold change ≥ 1.0. Results Comparative analysis of demographic characteristics in CRC patients Table 1 summarizes the demographic and clinical characteristics of the study participants. As expected, CRC patients were significantly older than NCCs (P < 0.001). Significant differences were also observed in employment status, family history of CRC, disease or surgical history, and physical activity (all P < 0.05). In contrast, no statistically significant differences were detected between groups with respect to sex, viral infection status, smoking, alcohol consumption, dietary pattern, or regular probiotic and antibiotic use. Overall, the two groups were comparable in major lifestyle-related variables, while age and selected clinical characteristics differed between CRC patients and NCCs. Table 1Baseline demographic and clinical characteristics of colorectal cancer (CRC) patients and non-cancer controls (NCCs).VariableCategoryCRC stage I (n=53)NCC (n=51)P-valueAge59.19 ± 13.9536.20 ± 11.99<0.001SexFemale29 (54.72)31 (60.78)0.557Male24 (45.28)20 (39.22)ProfessionNo43 (81.13)12 (23.53)<0.001Yes10 (18.87)39 (76.47)Family historyNo28 (52.83)40 (78.43)0.008Yes25 (47.17)11 (21.57)Disease or surgical historyNo26 (49.06)36 (70.59)0.029Yes27 (50.94)15 (29.41)Viral infectionNo52 (98.11)51 (100.00)1.000Yes1 (1.89)0 (0.00)SmokingNo43 (81.13)46 (90.20)0.266Yes10 (18.87)5 (9.80)Alcohol consumptionNo50 (94.34)46 (90.20)0.484Yes3 (5.66)5 (9.80)Physical activityNo46 (86.79)20 (39.22)<0.001Yes7 (13.21)31 (60.78)Dietary patternHigh consumption of meat14 (26.42)5 (9.80)0.081High consumption of vegetables and fruits5 (9.43)8 (15.69)All food (vegetable and meat)34 (64.15)38 (74.51)Regular probioticNo53 (100.00)50 (98.04)0.490Yes0 (0.00)1 (1.96)Antibiotic usageNo52 (98.11)51 (100.00)1.000Yes1 (1.89)0 (0.00)Age is expressed as mean ± SD; categorical variables are presented as n (%). Baseline demographic and clinical characteristics of colorectal cancer (CRC) patients and non-cancer controls (NCCs). Age is expressed as mean ± SD; categorical variables are presented as n (%). Differential proteomic analysis to reveal biomarker candidates In total, 2,456 proteins were identified and quantified across saliva samples from CRC patients and NCCs. Differential abundance analysis revealed 181 proteins that were significantly altered between the two groups. To evaluate whether these proteins could distinguish CRC patients from NCCs, their expression profiles were visualized across all samples. The resulting heatmap (Figure 1) demonstrated partial separation between CRC and NCC samples, although considerable heterogeneity was observed within both groups. Clustered heatmap of differentially abundant proteins in saliva samples from colorectal cancer (CRC) patients and non-cancer controls (NCC). Hierarchical clustering was performed on both proteins (rows) and samples (columns) using Pearson-correlation distance with average linkage. Rows represent proteins; columns represent individual samples. Values are mean-centered per protein (row-wise; each protein’s across-sample mean was subtracted); red indicates higher abundance and blue indicates lower abundance. The top annotation bars denote sample annotations including demographic and clinical variables (e.g., age, sex, smoking). A partial, but not complete, separation between CRC and NCC was observed, consistent with considerable inter-individual heterogeneity in salivary proteomes. In this figure, the label “Healthy” refers to non-cancer controls (NCCs). Using a volcano plot, we visualized the log2 fold-change (log2FC) in protein abundance between CRC and NCC against the corresponding -log10 p-values (Figure 2). Proteins with an absolute log2 fold change ≥ 1 (≥ 2-fold change) and an q-value < 0.05 were considered differentially abundant (27). Several proteins exhibited markedly higher or lower abundance in CRC patients compared with NCCs. Notably, multiple members of the small proline-rich protein (SPRR) family, including SPRR1A, SPRR3, and SPRR2D showed significantly higher abundance in CRC. Increased abundance was also observed for thioredoxin domain-containing protein 17 (TXNDC17), diazepam binding inhibitor (DBI), peptidase inhibitor 3 (PI3/Elafin), kallikrein-related peptidase 6 (KLK6), WAP four-disulfide core domain protein 2 (WFDC2), and fatty acid binding protein 5 (FABP5). For clarity, only selected representative proteins are labeled in the plot. Potential demographic influences, including age, sex, and smoking status, were considered during the differential-abundance analysis and further explored through subgroup analyses. Volcano plot shows differentially abundant salivary proteins between colorectal cancer (CRC) patients and non-cancer controls (NCCs). The x-axis represents the log2 fold change (CRC vs. NCCs), where positive values indicate higher abundance in CRC samples, and the y-axis represents −log10-transformed nominal p-values. Differential abundance analysis was adjusted for age, sex, and smoking status, and Benjamini–Hochberg false discovery rate (FDR)-adjusted q-values were used to correct for multiple testing. Red dots indicate proteins with q < 0.05 and |log2FC| ≥ 1 (≥2-fold) showing higher abundance in CRC, whereas blue dots indicate proteins with q < 0.05 and |log2FC| ≥ 1 showing lower abundance in CRC. Pink and pale blue dots represent proteins with p < 0.05 and |log2FC| ≥ 1 that did not remain significant after FDR correction. Black dots indicate proteins with p < 0.05 but |log2FC| < 1, while gray dots represent non-significant proteins (p ≥ 0.05). Vertical dashed lines indicate |log2FC| = 1, and the horizontal dashed line indicates p = 0.05 (−log10(0.05) ≈ 1.3). Prior to expansion of the study cohort, an initial exploratory analysis was performed using a smaller subset of samples (20 CRC and 20 NCCs) to assess the reproducibility of the proteomic workflow and the detectability of CRC-associated salivary protein alterations. Notably, several candidate proteins identified in the final cohort, including SPRR3, FABP5, DBI, PI3, WFDC2, and SOD1, were also consistently detected in this preliminary analysis. These exploratory findings are summarized in Supplementary Table 1. To evaluate the potential influence of demographic variables on salivary protein patterns, subgroup analyses stratified by smoking status, age, and sex were performed (Figure 3). In the middle-aged subgroup (40–59 years; Figure 3B), several CRC-associated salivary protein alterations remained detectable, supporting that the observed proteomic differences were not solely attributable to age-related changes. Similar CRC-associated alterations were also observed across the smoking- and sex-stratified subgroups. In the non-smoker subgroup (Figure 3A), multiple proteins remained differentially abundant between CRC patients and NCCs. In the male subgroup (Figure 3C), proteins including CEACAM5, CSTB, SOD1, FABP5, DBI, and KLK13 demonstrated higher abundance in CRC samples. In the female subgroup (Figure 3D), several proteins exhibited higher abundance in CRC compared with NCCs, including members of the small proline-rich protein (SPRR) family (SPRR1A, SPRR3, and SPRR2A), as well as WFDC2 and TXNDC17. Overall, these findings suggest that the core CRC-associated salivary protein signature is detectable across demographic subgroups, including the middle-aged subgroup despite the overall age disparity between CRC patients and NCCs, while certain proteins may exhibit subgroup-specific abundance differences. Volcano plots of differentially abundant salivary proteins between colorectal cancer (CRC) patients and non-cancer controls (NCCs), stratified by demographic variables. (A) Non-smokers, (B) middle-aged individuals (40–59 years), (C) males, and (D) females. Differential abundance analysis was performed using Student’s t-test followed by Benjamini–Hochberg false discovery rate (FDR) correction. Vertical dashed lines indicate |log2FC| = 1, and the horizontal dashed line indicates p = 0.05. Gray dots represent non-significant proteins; black dots indicate proteins with p < 0.05 but |log2FC| < 1; pale blue and pink dots represent proteins with >2-fold lower or higher abundance, respectively, based on unadjusted p-values; blue and red dots indicate proteins with >2-fold lower or higher abundance, respectively, that remained significant after FDR correction (q < 0.05). In this study, we identified candidates in the saliva of CRC patients that demonstrated significant alterations compared to NCCs. These proteins were analyzed and compared with those reported in previous CRC studies (Table 2). Our findings suggest that several of these differentially abundant proteins have been previously described in CRC, primarily in tissue or serum samples. Their detection in saliva in our cohort further supports their potential relevance as non-invasive biomarkers. Table 2Candidate salivary protein biomarkers identified in colorectal cancer (CRC) versus non-cancer controls (NCC), compared with biomarkers reported in previous CRC studies.Protein nameGene nameUniProt accession numberMechanism in CRCSample sourceStage/phase of CRCReferencesSmall Proline-Rich Proteins (SPRR1A, SPRR3)SPRR1A, SPRR3P35321, P26447Enhances cell migration, invasion, and metastasis in CRC.TissueMetastatic CRC(27, 63)Diazepam Binding Inhibitor (DBI)DBIP07108Involved in lipid metabolism; implicated in tumor progression.TissueCRC progression(64)Kallikrein-Related Peptidase 6 (KLK6)KLK6Q92876Associated with protease activity, promoting metastasis and poor prognosis in CRC.Serum, tissueMetastatic CRC(45, 65)S100A8/A9 (Calprotectin)S100A8, S100A9P05109, P06702Linked to neutrophil deregulation, CRC progression, and inflammation.Serum, tissueCRC progression(66, 67)Superoxide Dismutase 1 (SOD1)SOD1P00441Implicated in oxidative stress response and tumor suppression in CRC.SerumEarly and late-stage CRC(68, 69)Peptidase Inhibitor 3 (Elafin, PI3)PI3P19957Plays a role in inflammation and immune response, associated with metastasis in CRC.SerumMetastatic CRC(70)Aspartate Beta-Hydroxylase (ASPH)ASPHQ12797Promotes cell migration and invasion, associated with metastatic CRC.TissueMetastatic CRC(71) Candidate salivary protein biomarkers identified in colorectal cancer (CRC) versus non-cancer controls (NCC), compared with biomarkers reported in previous CRC studies. Notably, some of the protein biomarker candidates detected in CRC were also reported in other oro-gastrointestinal cancers, including oral squamous cell carcinoma (OSCC), esophageal cancer, gastric cancer, liver cancer, and pancreatic cancer. These shared biomarkers, summarized in Table 3, indicate potential overlap in molecular processes across GI malignancies. While these proteins may have broader relevance in cancer biology, our findings emphasize the importance of further evaluating CRC-associated markers to improve diagnostic specificity. Table 3Comparison of salivary protein biomarker candidates identified in this study for colorectal cancer (CRC) versus non-cancer controls (NCC), with reports in other gastrointestinal (GI) cancers.Protein nameGene nameUniProtaccession numberMechanism in cancerSample sourceDiseaseReferencesAlpha-1 Antitrypsin (A1AT)SERPINA1P01009, P02763Protease inhibitor, tissue protection,modulating inflammation and immune responses.SerumOral squamous cell carcinoma (OSCC)(72)Interleukin-1 Receptor Antagonist Protein (IL1RN)IL1RNP18510Inhibits IL-1 signaling, suppressing inflammation and early carcinogenic events.Tissue, SerumOSCC(73, 74)Cystatin-B (CSTB)CSTBP04080Inhibitor of cysteine proteases, increased cell proliferation and migration, reducing apoptosis.Gastric cancer SGC-7901 cell lineGastric cancer (GC)(75)Triosephosphate Isomerase (TPI)TPI1P60174Involved in glycolysis; facilitating metabolic reprogramming in cancer cells, contributing to rapid cell proliferation.Gastric cancer MGC-803 cell lineGC(76)Aflatoxin B1 Aldehyde Reductase (AKR7A2)AKR7A2Q43488Detoxification of carcinogenic aldehydes, involved in resistance to chemotherapy, protection against oxidative stress.Pancreatic ductal adenocarcinoma tissue, normal adjacent pancreatic tissue, pancreatic benign cystadenoma tissuePancreatic ductal adenocarcinoma(77)Annexin A7 (ANXA7)ANXA7P20073Regulating cell proliferation, apoptosis, migration, and invasion facilitating tumor metastasis.Cancer tissues and adjacent normal tissuesVarious cancers including hepatocellular carcinoma(78)Sialic Acid Binding Ig-like Lectin 5 (SIGLEC5)SIGLEC5Q9Y336Immune evasion, modulation of tumor microenvironment, regulation of immune response.Tumor tissues, cell lines, patient samplesVarious cancers including colorectal cancer (CRC)(79)Y Box Binding Protein 1 (YBX1)YBX1P67809Involved in the regulation of oncogenes and tumor suppressors, promotion of metastasis, and resistance to therapy.Tissue and cell linesVarious human cancers including pancreatic cancer and CRC(80) Comparison of salivary protein biomarker candidates identified in this study for colorectal cancer (CRC) versus non-cancer controls (NCC), with reports in other gastrointestinal (GI) cancers. Functional analysis of salivary protein alterations in CRC To further elucidate the functional implications of salivary protein alterations between CRC patients and NCCs, gene set enrichment analysis (GSEA) was conducted across three major Gene Ontology (GO) categories, Biological Process (BP), Cellular Component (CC), and Molecular Function (MF), as shown in Figures 4A–C. Distinct patterns of enrichment were observed, with a predominant involvement of cytoskeleton-related pathways and proteolytic regulation (28, 29). Gene Set Enrichment Analysis (GSEA) of differentially abundant salivary proteins between CRC patients and non-cancer controls (NCCs), categorized by Gene Ontology (GO) domains: (A) Biological Process (BP), (B) Cellular Component (CC), (C) Molecular Function (MF). Dot plots depict the top enriched terms, separated into activated (higher representation in CRC) and suppressed (lower representation in CRC) categories. Bubble size represents the number of proteins in each term (set size); the color gradient indicates adjusted p-values (q-values). Enrichment was performed using the GO database and visualized with dot plots. Within the BP category (Figure 4a), several terms were relatively decreased in CRC compared to NCCs, particularly those related to cytoskeletal and intermediate filament organization. These included intermediate filament organization, intermediate filament cytoskeleton organization, cytoskeleton organization, and supramolecular fiber organization, suggesting altered representation of structural protein programs in CRC saliva (28–30). Epithelial-associated processes, such as keratin-related pathways, were observed within this broader pattern of reduced structural organization (29, 30). Conversely, fewer but notable activated processes were detected, with proteolysis emerging as a consistently enriched biological process (Figure 4A), suggesting altered protease activity in CRC saliva (31). Analysis of the CC category revealed that the most significantly affected compartments were structural elements associated with the cytoskeleton and filament-based assemblies (Figure 4B). Several cytoskeleton-related components, including intermediate filament, keratin filament, and supramolecular fiber, showed relatively lower representation in CRC compared to NCCs (28, 29), reflecting alterations in structural protein composition. Within the MF category, activated terms were dominated by protease inhibitory and regulatory activities, including enzyme inhibitor activity, serine-type endopeptidase inhibitor activity, and peptidase regulatory activity (Figure 4C). These findings point to enhanced regulatory mechanisms controlling proteolytic processes in CRC saliva (31, 32). In contrast, functions associated with cytoskeletal binding and structural molecule activity, such as actin binding, cytoskeletal protein binding, and structural constituent of cytoskeleton (Figure 4c), were relatively decreased in CRC compared to NCCs (28, 29). A highly interconnected immune–metabolic signaling architecture characterizes CRC saliva To further elucidate the functional interplay between dysregulated biological processes in CRC patients, we applied Ingenuity Pathway Analysis (IPA) “Overlapping Canonical Pathways” network analysis (Figure 5). This analysis revealed a highly interconnected signaling architecture among significantly enriched pathways in CRC saliva. Central nodes within the network included Antimicrobial Peptides, Clathrin-Mediated Endocytosis Signaling, Neutrophil Degranulation, and MSP-RON Signaling, all of which are implicated in inflammation, immune modulation, and cancer biology (33, 34). Metabolic pathways such as Reactive Oxygen Species (ROS) Detoxification, Fatty-acid β-Oxidation, and Melatonin Degradation III were also integrated into the network, suggesting a link between metabolic stress and tumorigenesis (35). Overlapping Canonical Pathways Network generated using Ingenuity Pathway Analysis (IPA). Red nodes represent significantly enriched canonical pathways (p < 0.05) identified in the salivary proteome of CRC patients versus non-cancer controls (NCCs). Color intensity corresponds to the significance of pathway enrichment (−log10 p-value). Blue edges denote shared molecular components (e.g., genes or proteins) between pathways, representing functional overlap or regulatory crosstalk. The network highlights interrelated immune, metabolic, and signaling pathways implicated in CRC biology. Pathways such as MSP-RON signaling, neutrophil degranulation, antimicrobial peptides, and reactive oxygen species (ROS) detoxification formed core hubs within the network. Discussion This study presents a comprehensive analysis of the salivary proteome in Iranian CRC patients compared with NCCs, revealing a distinct set of differentially abundant proteins with potential diagnostic relevance. By integrating demographic, lifestyle, and clinical characteristics with proteomic profiling and bioinformatics analyses, we identified coordinated alterations in immune-related, metabolic, and structural protein pathways. These findings not only reflect the systemic biological impact of CRC but also support the potential utility of saliva as a non-invasive source of disease-associated molecular signatures. Demographic and lifestyle factors Between-group comparisons revealed significant differences in age, profession, family history, disease or surgical history, and physical activity levels (Table 1). CRC patients were significantly older than NCCs, consistent with the known age distribution of CRC incidence. Lower physical activity levels were more frequently reported in the CRC group, in line with previous observations describing reduced activity levels among CRC patients (36). A higher proportion of CRC patients reported a family history and prior disease or surgical history within this cohort. According to the predefined study criteria, individuals with known hereditary CRC syndromes were excluded; thus, “family history” refers to self-reported familial occurrence without documented hereditary cancer syndromes. No statistically significant differences were observed in dietary pattern, smoking status, alcohol consumption, viral infection status, regular probiotic use, or antibiotic usage. Although dietary factors have been widely discussed in relation to CRC risk (37), dietary patterns did not differ significantly between CRC patients and NCCs in this study. Given the exclusion of participants with recent extensive antibiotic or probiotic exposure (with the exception of one short-term antibiotic case and one short-term probiotic case), variation in these variables was minimal within the cohort. Overall, major lifestyle-related factors were broadly comparable between groups. Salivary proteome alterations and key biomarker candidates Unsupervised hierarchical clustering of salivary proteomic profiles (Figure 1) demonstrated partial separation between CRC stage I patients and NCCs. Although complete segregation was not observed, several clusters showed visible alignment with clinical and demographic annotations, indicating inter-individual heterogeneity within both groups. This pattern suggests that salivary proteomic alterations in CRC coexist with background variability related to host characteristics. Global differential abundance analysis using a covariate-adjusted volcano plot (Figure 2) identified a subset of proteins significantly altered between CRC and NCC after adjustment for age, sex, and smoking status. Several proteins exhibited increased abundance in CRC saliva, including members of the SPRR family (SPRR1A, SPRR3, and SPRR2D), as well as TXNDC17, DBI, PI3 (Elafin), KLK6, WFDC2, and FABP5. Among these, the consistent elevation of SPRR family members is notable. SPRR proteins have been linked to epithelial differentiation, epithelial-to-mesenchymal transition (EMT), and tumor progression in CRC and other epithelial malignancies (38–41). Their increased abundance in saliva may reflect tumor-associated epithelial remodeling and inflammatory signaling processes (42–44). Similarly, KLK6 has been associated with protease activity and metastatic behavior in CRC (45), while DBI and FABP5 are involved in lipid metabolism and metabolic regulation (46). PI3 (Elafin) and WFDC2 have been connected to inflammatory and extracellular matrix–related mechanisms in cancer biology (47, 48). Collectively, these findings indicate coordinated alterations in epithelial integrity, protease regulation, metabolism, and inflammatory signaling in CRC saliva. In addition to elevated proteins, several proteins showed reduced abundance in CRC saliva, including myeloperoxidase (MPO), lysozyme (LYZ), histone cluster 1 H2B family member D (HIST1H2BD), annexin A3 (ANXA3), and acyl-CoA dehydrogenase very long chain (ACADVL). MPO and LYZ are central components of innate immune defense, while histone-associated proteins are linked to chromatin organization and cellular turnover processes (49–53). Their decreased representation in CRC saliva may reflect alterations in immune-related pathways or epithelial homeostasis within the disease context. Subgroup analyses (Figure 3) further demonstrated variability in the prominence of selected proteins across demographic strata. Carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5), cystatin B (CSTB), superoxide dismutase 1 (SOD1), DBI, and FABP5 showed stronger differential signals in males, whereas SPRR1A, SPRR3, small proline-rich protein 2A (SPRR2A), WFDC2, and TXNDC17 were more prominent in females. In middle-aged individuals, several SPRR family members and additional CRC-associated proteins remained differentially abundant between CRC patients and NCCs. These observations suggest that although demographic variables may influence specific salivary protein abundance patterns, multiple CRC-associated proteomic alterations remained detectable across demographic subgroups (54, 55). Overall, the identified salivary alterations support the presence of a multi-protein CRC-associated signature encompassing epithelial structural proteins, metabolic enzymes, immune-related mediators, and protease regulators. This integrated pattern reinforces the potential advantage of multiplex biomarker panels over single-marker approaches in non-invasive CRC detection (56, 57). Comparative analysis with other studies To contextualize our findings, we compared the salivary biomarker candidates identified in this study with previously reported markers in CRC and other oro-GI malignancies, including OSCC, as summarized in Tables 2 and 3. Several proteins detected in our cohort have also been described in CRC tissue or serum studies, as well as in other GI cancers, supporting their broader involvement in oncogenic and inflammatory processes. At the same time, the specific combination and relative abundance patterns observed in our cohort suggest the presence of a CRC-associated salivary proteomic profile, rather than reliance on a single disease-specific marker (58). These findings highlight the importance of further validation in independent cohorts to determine the reproducibility and diagnostic specificity of this candidate panel. Recent molecular classification efforts in CRC, such as the genome- and transcriptome-based subtyping approach described by Guinney et al. (2015), further emphasize the biological heterogeneity of CRC and the value of integrating multi-layer molecular data (59). In this context, salivary proteomic signatures may represent a complementary, non-invasive dimension for refining CRC characterization. Functional enrichment: GSEA and IPA insights Gene Set Enrichment Analysis (GSEA) (Figure 4) highlights a coordinated shift in structural and regulatory functions in the salivary proteome of CRC patients. The consistent suppression of cytoskeleton- and intermediate filament–related biological processes and cellular components across all three GO categories indicates a loss of structural organization, reflecting cytoskeletal remodeling commonly associated with malignant transformation (28–30). In parallel, the enrichment of protease inhibitory and regulatory molecular functions suggests altered control of proteolytic activity, potentially reflecting compensatory or tumor-associated mechanisms rather than generalized proteolysis (31, 32). Together, these patterns point to a saliva-based proteomic signature characterized by disrupted cellular architecture and enhanced protease regulation in CRC, underscoring the relevance of cytoskeletal integrity and proteolytic balance as hallmarks of disease-associated molecular alterations captured through non-invasive saliva sampling. IPA-based canonical pathway analysis (Figure 5) revealed a network of interconnected immune and metabolic pathways in CRC saliva. Immune-related hubs included Neutrophil Degranulation, Antimicrobial Peptides, and MSP-RON Signaling, highlighting coordinated inflammatory and immune-modulatory activity (33, 34). Concurrent enrichment of Reactive Oxygen Species (ROS) Detoxification and Fatty-acid β-Oxidation pathways indicates altered oxidative balance and metabolic adaptation, processes commonly associated with CRC biology (33–35). The presence of Melatonin Degradation within the enriched network further supports the involvement of metabolic regulatory mechanisms (60). Overall, this interconnected map reflects a multifaceted CRC-associated signaling landscape in saliva, where immune-related pathways, oxidative stress mechanisms, and metabolic regulatory processes appear concurrently represented within a shared molecular network (35, 61). Limitations and future directions Despite the robustness of our findings, several limitations should be acknowledged. The relatively modest sample size and lack of longitudinal data may limit the generalizability of the results. In addition, this study should be regarded as a discovery-phase pilot biomarker investigation intended to identify candidate salivary proteins associated with CRC. While encouraging differences were observed, future studies will require orthogonal validation and independent external cohorts to confirm reproducibility and clinical utility. Future investigations should include larger, multi-ethnic cohorts and consider integrating salivary proteomics with genomic, transcriptomic, and microbiome data to more comprehensively capture CRC complexity. Furthermore, the development of targeted proteomics platforms, such as selected reaction monitoring (SRM), may facilitate biomarker validation and support future clinical translation (62). Conclusions This study underscores the diagnostic potential of salivary proteomics in CRC. The observed dual signature, characterized by suppression of cytoskeletal and epithelial structural programs alongside activation of immune and metabolic pathways, reflects the systemic nature of CRC and the informative value of saliva. With further validation, the identified biomarker candidates may contribute to the development of non-invasive diagnostic platforms and support more personalized approaches in CRC management.