Decoding protein signatures and protein interactions in oral potentially malignant disorders: a systematic review and network analysis Authors: Bounika Esvanth Rao, Anju M. Nair, Preethi Ramesh, Vijayalakshmi Ramshankar, S. Gopinath, P. A. Abhinand, Artnora Ndkoraj, Marta Mazur, Saman Warnakulasuriya, Divyambika Catakapatri Venugopal Published: 2026 ABSTRACT BackgroundProteomic profiling offers thorough insights into protein structure and function, as well as it acts as an essential approach for analyzing molecular changes at the tissue level. However, because of the proteome's diversity and dynamic nature, biomarker discovery remains challenging. By combining proteomics with bioinformatics, the level of understanding in relation to molecular interactions and disease processes can be improved. Through an integrative approach, few limitations can be addressed, thereby promoting proteomic profiling for the discovery of new therapeutic targets and novel biomarkers for a variety of disorders.AimTo identify differentially expressed protein markers and their key molecular pathways associated with Oral Potentially Malignant Disorders.MethodsSystematic Review was conducted following the PRISMA guidelines and the protocol registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the registration ID number CRD42024557545. A comprehensive literature review was performed using electronic databases, yielding 12,797, studies from which 15 eligible articles were selected. The Newcastle–Ottawa Scale was used to assess the risk of bias. Vote counting was performed to identify proteins reported in more than one study. A bipartite network was constructed using Cytoscape to identify shared and disease-specific protein markers. Lesion-wise protein–protein interaction networks were generated using STRING and analysed in Cytoscape to identify highly interconnected hub proteins, and pathway enrichment analysis for these hubs was performed using Reactome.ResultsA total of fifteen studies (Leukoplakia (LK) – n = 1, Proliferative Verrucous Leukoplakia (PVL) – n = 2, Oral Submucous Fibrosis (OSMF) – n = 7, and Oral Lichen Planus (OLP) – n = 5) were included. The Newcastle–Ottawa Scale was used to evaluate methodological quality and the quality of studies included in this systematic review was high for 4 articles and moderate in the remaining 11. The most commonly employed technique was mass spectrometry. A total of 318 candidate proteins (LK – 14, PVL – 82, OSMF – 172, and OLP – 50) were identified across the oral potentially malignant disorders. Key markers identified through vote counting included ERO1A, NUCB1, RHOA, and IL36A for PVL; LUM, KRT1, KRT9, ALB, and VIM for OSMF; and ALB, LYZ, HP, HBB, and AMY1A for OLP. The bipartite network showed that OSMF and OLP shared the highest number of proteins, indicating the strongest overlap among lesions. Network analysis further highlighted distinct hub proteins for each lesion: for LK— AMY1A, AMY1B and APOA1; for PVL— CFL1, RHOA and CDC42; for OSMF— HSP90AA1, ENO1 and SERPINA1; and for OLP— HP, B2M, and ORM1. Lesion-specific pathway enrichment revealed that LK was associated with epithelial differentiation, PVL with oncogenic signaling, OSMF with stress-driven fibrosis, and OLP with immune-mediated inflammation.ConclusionsProteomic expression offers insights into disease pathogenesis by identifying important molecular changes across OPMDs. However, the majority of biomarkers are still in the exploratory stage due to the considerable variation in lesion types, sample sources, proteomic techniques, and reporting systems. In order to create reliable and clinically applicable biomarkers, future studies should concentrate on combining multi-omics techniques with large-scale, standardized cohorts.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12903-026-08849-8. Introduction Oral squamous cell carcinoma (OSCC) is a malignant neoplasm arising from the stratified squamous epithelial cells lining the oral mucosa. Data from the Global Cancer Observatory (GLOBOCAN) 2020 database indicate approximately 90% of head and neck cancers are squamous cell carcinomas [1]. The GLOBOCAN 2022 India Fact Sheet reports that, lip and oral cavity cancer is the second most common cancer in India with 143,759 new cases (10.2% of all cancers), and 2nd in deaths with 79,979 deaths (8.7% of all cancer deaths) [2, 3]. A considerable proportion of OSCC arises from oral potentially malignant disorders (OPMDs), a group of precursor lesions with an estimated global prevalence of approximately 5% [4]. These include oral leukoplakia (LK), proliferative verrucous leukoplakia (PVL), oral submucous fibrosis (OSMF), and oral lichen planus (OLP), each exhibiting varying rates of malignant transformation. The reported malignant transformation rates range from 7.2–9.8% for LK [5], 43.87% for PVL [6], 7–13% for OSMF [7], and 0.44%–2.28% for OLP [8]. The late clinical detection of these lesions, along with the lack of reliable molecular markers for risk assessment, are major contributors to poor survival outcomes in oral cancer. Malignant development in OPMDs is triggered by complex genetic and molecular changes, such as chromosomal instability, gene mutations, and metabolic dysregulation [9]. These molecular mechanisms of cancer have been better explored by the use of 'omics' technologies such as transcriptomics, proteomics, metabolomics, and genomics. Proteomics is distinct among those because it can capture the whole picture of cellular protein expression, interactions, and post-translational modifications, which makes it an effective tool for exploring the factors involved in cancer. Despite advancements in analytical platforms, proteomic research on OPMDs is still in its nascent stage and remains far from fully integrated. Most of the research that is currently available focuses on specific lesions or small subsets of proteins; it does not offer comparative information across other lesion types or disease progression phases. Even with increasing evidence, the role of differential protein expression in malignant transformation remains incompletely understood. Also, the absence of thorough analysis combining pathway enrichment and protein–protein interaction (PPI) networks makes it difficult to identify the molecular networks promoting disease progression. Therefore, mapping important protein markers, clarifying dysregulated pathways linked to malignant transformation, and synthesizing data from proteomic research on OPMDs are the goals of this systematic review and network analysis. Methodology Study protocol The present systematic review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; Registration No: CRD42024557545). The review was conducted in accordance with established methodological standards for systematic reviews and followed the recommendations outlined in the Cochrane Handbook for Systematic Reviews of Interventions. Reporting of the study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The PRISMA checklist is provided in Fig. 1.Fig. 1PRISMA 2020 flow diagram [10]. * Records were excluded after title/abstract screening (n = 10,867), including in-vitro studies (n = 1223), in-vivo studies (n = 526), case reports (n = 197), review articles (n = 838), OSCC-focused studies (n = 2826), studies on other systemic diseases (n = 3483), randomized controlled trials (n = 421), use of non-relevant techniques (n = 875), and other omics-based studies (n = 478) PRISMA 2020 flow diagram [10]. * Records were excluded after title/abstract screening (n = 10,867), including in-vitro studies (n = 1223), in-vivo studies (n = 526), case reports (n = 197), review articles (n = 838), OSCC-focused studies (n = 2826), studies on other systemic diseases (n = 3483), randomized controlled trials (n = 421), use of non-relevant techniques (n = 875), and other omics-based studies (n = 478) Study question Population Individuals clinically and histopathologically diagnosed with OPMDs, including LK, PVL, OSMF, and OLP. Intervention Assessment of altered protein-based biomarkers in serum, saliva, or tissue samples of patients with OPMD. Comparison Protein biomarker levels in healthy individuals without OPMDs. Outcome Identification of significant alterations in proteomic markers in OPMD patients compared to healthy controls, with potential implications for early detection. Eligibility criteria Inclusion:Patients diagnosed with OPMDs, including LK, PVL, OSMF and OLPLesions located at any oral anatomical site, without restrictions on age or genderStudies comparing proteomic profiles between OPMD patients and healthy controlsStudies reporting qualitative (upregulated or downregulated expression) and/or quantitative proteomic data, including fold change and/or p-values.Observational study design: case–control, cross-sectional, and cohort studies published between Jan 2000 to Jun 2024Studies employing proteomic techniques such as mass spectrometry, liquid chromatography coupled with tandem mass spectrometry, electrospray ionization, matrix-assisted laser desorption/ionization, and Fourier transform ion cyclotron resonanceArticles published in the English language Patients diagnosed with OPMDs, including LK, PVL, OSMF and OLP Patients diagnosed with OPMDs, including LK, PVL, OSMF and OLP Lesions located at any oral anatomical site, without restrictions on age or gender Lesions located at any oral anatomical site, without restrictions on age or gender Studies comparing proteomic profiles between OPMD patients and healthy controls Studies comparing proteomic profiles between OPMD patients and healthy controls Studies reporting qualitative (upregulated or downregulated expression) and/or quantitative proteomic data, including fold change and/or p-values. Studies reporting qualitative (upregulated or downregulated expression) and/or quantitative proteomic data, including fold change and/or p-values. Observational study design: case–control, cross-sectional, and cohort studies published between Jan 2000 to Jun 2024 Observational study design: case–control, cross-sectional, and cohort studies published between Jan 2000 to Jun 2024 Studies employing proteomic techniques such as mass spectrometry, liquid chromatography coupled with tandem mass spectrometry, electrospray ionization, matrix-assisted laser desorption/ionization, and Fourier transform ion cyclotron resonance Studies employing proteomic techniques such as mass spectrometry, liquid chromatography coupled with tandem mass spectrometry, electrospray ionization, matrix-assisted laser desorption/ionization, and Fourier transform ion cyclotron resonance Articles published in the English language Articles published in the English language Exclusion:Studies involving neoplasms, muco-cutaneous diseases, infections, localized trauma, recurrent aphthous stomatitis (RAS), systemic conditions, or conditions requiring chronic medicationStudies focusing on non-protein biomarkersIn-vitro or in-vivo experimental studiesNon-observational study designs (e.g., reviews, case reports, case series, editorials, and conference abstracts) Studies involving neoplasms, muco-cutaneous diseases, infections, localized trauma, recurrent aphthous stomatitis (RAS), systemic conditions, or conditions requiring chronic medication Studies involving neoplasms, muco-cutaneous diseases, infections, localized trauma, recurrent aphthous stomatitis (RAS), systemic conditions, or conditions requiring chronic medication Studies focusing on non-protein biomarkers Studies focusing on non-protein biomarkers In-vitro or in-vivo experimental studies In-vitro or in-vivo experimental studies Non-observational study designs (e.g., reviews, case reports, case series, editorials, and conference abstracts) Non-observational study designs (e.g., reviews, case reports, case series, editorials, and conference abstracts) Information source The following electronic databases were systematically searched: MEDLINE (PubMed), Cochrane Library, Scopus, Google Scholar, and the Latin American and Caribbean Health Sciences Literature (LILACS), from January 2000 to June 2024. Search strategy (Proteomics OR ("PROTEIN EXPRESSION") OR ("protein based biomark*")) AND (serum OR blood OR plasma OR tissue OR saliva) AND(("oral precancer*"[Title/Abstract]) OR ("oral pre malignancy"[Title/Abstract]) OR ("oral potentially malignant disorder"[Title/Abstract]) OR ("opmd"[Title/Abstract]) OR ("oral submucous fibrosis"[Title/Abstract]) OR ("osmf"[Title/Abstract]) OR ("oral lichen planus"[Title/Abstract]) OR ("leukoplakia"[Title/Abstract])), restricted from January 2000 to June 2024. Screening and data extraction Two trained reviewers (BER and AMN) independently screened the titles, abstracts, and keywords to identify studies that met the eligibility criteria. Rayyan QCRI was used as the reference manager to perform the screening of the articles and removal of duplicates. Electronic mail or communication through Research Gate was sent to obtain the full texts of studies that required additional information or where abstracts were unclear or unavailable. Subsequently, the full texts of the articles that met the inclusion criteria during the initial screening were obtained. If the full text was unavailable, up to three attempts were made to contact the corresponding author and libraries; if unsuccessful, the study was excluded. Blinding of the articles was not performed with respect to the journals published, authors, or institutions. Disagreements among the reviewers were resolved through mutual consensus. When consensus could not be reached, a third reviewer (CVD) acted as an arbitrator to resolve the disagreement. All excluded studies at this stage were documented in an Excel spreadsheet (Excel 10, Microsoft Corp., Redwood City, Calif., USA), along with the reasons for exclusion. The reviewers evaluated the included studies to extract data regarding the first author, year of publication, place of study, study design, number of patients and controls, and the name and type of protein biomarker. Risk of bias (quality) assessment Quality assessment was performed independently by two reviewers (BER and PR) using the Newcastle–Ottawa Scale (NOS) for case–control studies. This tool evaluated studies across three domains, which included selection of study groups (maximum 4 stars), comparability of groups (maximum 2 stars), and outcome/exposure assessment (maximum 3 stars), with a maximum score of 9 stars. The Selection domain assures the adequacy of case definition, representativeness of cases, selection of controls, and definition of controls. Studies were awarded stars based on the extent to which they satisfied these criteria. The Comparability domain evaluates whether studies controlled for potential confounding factors. One star was awarded if the study controlled for the most important confounder, and an additional star was assigned if other relevant factors were also considered in the design or analysis. The Exposure domain assesses the method of determination for exposure, the use of the same ascertainment method for cases and controls, and the accuracy of non-response rates in describing them and comparing them between groups. Stars were allotted accordingly. Based on the total NOS score, studies were classified as low risk of bias (7–9 stars), moderate risk (4–6 stars), or high risk (< 4 stars). The certainty of evidence for each lesion-specific outcome was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. Selection of protein for network analysis Proteins reported to be differentially expressed in the articles included in the systematic review were selected for analysis. As there was a heterogeneity in the reporting criteria across the included studies, a uniform threshold for differential expression could not be applied. Restricting inclusion to proteins with standardized statistical measures (p-value, q-value or fold change) would substantially reduce the number of eligible proteins and potentially lead to selection bias. Therefore, an inclusive approach was incorporated, where all the differentially reported proteins were included. These proteins were subsequently categorized based on lesions for further analysis. Vote counting analysis of differentially expressed proteins The sample sizes were limited, and there was substantial heterogeneity across the included studies, mainly in terms of sample types, proteomic techniques, reporting metrics and analytical depiction of data, making this systematic review unfavourable for meta-analysis. To overcome this, a vote-counting approach was employed as a semi-quantitative method to identify consistently reported protein expression patterns specific to each lesion. Proteins reported in at least two independent studies were included in the vote-counting analysis. For each protein, a vote score was assigned based on the direction and frequency of expression reported across studies. Specifically, each instance of up-regulation was assigned a score of + 1, while each instance of down-regulation was assigned a score of − 1. The final vote score for each protein was calculated as the cumulative sum of these values across all included studies. Construction of bipartite disease–protein network To understand the association between OPMDs and the protein marker, a bipartite (two-mode) disease–protein network was constructed using Microsoft Excel and Cytoscape v3.8.0. An edge list was prepared in CSV format, in which lesions (LK, PVL, OSMF and OLP) were represented as source nodes and their associated proteins as target nodes, with each row representing a disease–protein interaction. A separate node attribute file was generated to classify nodes into two types: disease and protein. The edge list was imported into Cytoscape to generate the network, and the nodes were incorporated into an attribute table to assign node categories. The network was visualized using the organic layout to ensure optimal spatial distribution and clustering of related nodes. PPI network construction and hub protein identification The well-known housekeeping or high-abundance proteins were removed from the protein list obtained from the systematic review, as network analysis has a greater tendency to be inherently biased towards well-studied proteins such as ALB and ACTB, due to the greater availability of experimental evidence and higher annotation density. The lesion-wise (LK, PVL, OSMF and OLP) segregated protein datasets were used to construct protein–protein interaction (PPI) networks using the STRING database (https://string-db.org/). A high confidence score of (≥ 0.7) was included to reduce spurious connections driven by text-mining or co-citation bias. The resulting four interaction networks were imported into Cytoscape v3.8.0 [11] one by one to identify the hub proteins. These hubs were represented as highly connected nodes within a protein–protein interaction network that play central roles in maintaining network integrity and regulating key biological processes. Using the CytoHubba plugin in Cytoscape, the top hub proteins per lesion network were identified employing Maximal Clique Centrality (MCC), EPC (Edge Clustering Coefficient) degree, betweenness and closeness algorithms. These algorithms were selected because of their effectiveness in identifying highly connected and biologically significant nodes within protein–protein interaction networks. Hence, five sets for hub proteins were obtained for each lesion. Pathway enrichment analysis A lesion-wise pathway enrichment analysis was performed considering hub proteins that were obtained via MCC, degree, EPC, betweenness and closeness. An online gene set enrichment analysis tool, Enrichr [12], was used to analyse the pathways. The Reactome 2024 pathway database was selected to find biological pathways. To account for multiple testing, both raw and Benjamini–Hochberg adjusted p-values were presented. Pathways with an adjusted p-value (false discovery rate, FDR) of less than 0.05 were considered statistically significant. The output included enriched pathway names, associated genes, p-values, adjusted p-values, odds ratios, and combined scores. Results Study selection Through comprehensive searches across five major databases, including Cochrane (n = 2516), EMBASE (n = 1025), LILACS (n = 15), Scopus (n = 1240), and PubMed (n = 8001), a total of 12,797 records were identified. After removing duplicate records (n = 1887), 10,910 records were screened. Of these, 10,867 were excluded based on their “titles and/or abstracts” (in-vitro studies (n = 1223), in-vivo studies (n = 526), case reports (n = 197), review articles (n = 838), OSCC-focused studies (n = 2826), studies on other systemic diseases (n = 3483), randomized controlled trials (n = 421), use of non-relevant techniques (n = 875), and other omics-based studies (n = 478). Forty-three reports were sought for retrieval. During eligibility assessment, further exclusions were made due to the following reasons: articles not in English (n = 1), transcriptomic/metabolomic studies (n = 5), conference abstracts (n = 2), studies focusing primarily on OSCC (n = 13), unspecified protein names (n = 6), and replication of markers across articles (n = 1). Ultimately, 15 studies were included for qualitative synthesis in the systematic review (Fig. 1). Study characteristics and data collection Among the 15 included studies, LK (n = 1), PVL (n = 2), OSMF (n = 7), and OLP (n = 5) were investigated. Based on the sample type of the reported articles, saliva was the most commonly analyzed biological fluid, used in seven studies. Tissue samples were employed in six studies; one study utilized serum, and one used serum and saliva. Through this systematic review, a total of 318 proteins were identified. Among these, 146 proteins were upregulated, 158 were downregulated, and 14 proteins were reported without a specified direction of expression. For LK, a total of 14 markers were identified, all of which were salivary biomarkers. Among these, 3 proteins were upregulated, 2 were downregulated, and the remaining 9 were unspecified in terms of expression. In PVL, 82 markers were reported, all derived from salivary samples. Of these, 8 proteins were upregulated, while 74 were downregulated. The highest number of markers was observed in OSMF, with a total of 172 proteins identified across saliva, serum, and tissue samples. Among these, 91 proteins were upregulated in tissue samples. Downregulated proteins included 5 in saliva, 3 in serum, and 68 in tissue, while 5 proteins were reported as unspecified. For OLP, 50 markers were identified. Among these, upregulated proteins included 29 in saliva, 1 in serum, and 14 in tissue samples. Downregulated proteins included 5 in saliva and 1 in tissue (Tables 1, 2).Table 1Characteristics and key findings of included proteomic studies in Oral Potentially Malignant Disorders (OPMDs)Sl. noStudy IDAuthorSampleCountrySample sizeTechniqueProteomic markerValidationLKMarkerp -valueFCUp-regulated1.Ig alpha-2 chain C region (IGHA2)< 0.05↑4.22.Apolipoprotein A1 (APOA1)< 0.05↑3.43.Mature metal chelatase catalytic antibody with hapten, chain A (MMCA)< 0.01↑10.1Down-regulated4.Cystatin-SN precursor (CST1)< 0.01↓5.55.Serum albumin (ALB)< 0.01↓11.41.309,495,928Camisasca et al., 2017 [13]LK (Saliva)BrazilNormal (n = 10); LK (n = 10)2DE, MALDI-TOF/TOF MSUnspecifiedIHC:Tissue: KRT106.Alpha-amylase (AMY1A)NANA7.Keratin-10 (KRT10)NANA8.IgM kappa IIIb (IgM κ IIIb)NANA9.Lysozyme precursor (LYZ)NANA10.Polymeric immunoglobulin receptor (PIGR)NANA11.Alpha-amylase 1B (AMY1B)NANA12.Ig gamma-1 chain C region (IGHG1)NANA13.IgG kappa chain (IGHG)NANA14.Immunoglobulin kappa light chain (IGK)NANAPVLMarkerq -valueFCUp-regulated1.Nucleobindin-2 (NUCB2).010.202.Chitotriosidase-1 (CHIT1).032.18Down-regulated3.Adenosyl homocysteinase (AHCY).002- 1.844.Beta-2 microglobulin (B2M).002- 1.415.BPI fold–containing family B member 2 (BPIFB2).01- 1.716.Carboxypeptidase E (CPE).007- 1.767.Cell division control protein 42 homolog (CDC42).001- 2.118.Chloride intracellular channel protein 1 (CLIC1).001- 2.199.Carboxylesterase 2 (CES2)<.001- 1.7210.Desmoglein-1 (DSG1).002- 1.7511.ERO1-like protein alpha (ERO1A)<.001- 1.8512.Heme-binding protein 2 (HEBP2)<.001- 1.8013.IgG Fc-binding protein (FCGBP).03- 1.7714.Interleukin-1 receptor antagonist protein (IL1RN).001- 1.4015.Interleukin-36 alpha (IL36A)<.001- 2.5416.Kallikrein-11 (KLK11).04- 1.1317.Keratin type I cytoskeletal 13 (KRT13).01- 1.1018.Keratin type I cytoskeletal 14 (KRT14).02- 1.7719.Carboxylesterase 1 (CES1).01- 2.0320.Ly6/PLAUR domain-containing protein 3 (LYPD3).008- 2.4721.Mesothelin (MSLN).01- 3.1722.Mucin-5B (MUC5B).04- 2.3323.Nucleobindin-1 (NUCB1).01- 1.6724.Nucleotide exchange factor SIL1 (SIL1).01- 1.2725.Prostaglandin reductase 1 (PTGR1).001- 1.642.81,332,839Arroyo et al., 2023 [14]PVL (Saliva)Brazil, Spain, and FinlandPVL (n = 7); LK (n = 8); Normal (n = 5)LC–MS/MS26.Proteasome activator complex subunit 2 (PSME2)<.001- 2.27Validation has been performed for tissue markers for which the control group was inflammatory fibrous hyperplasia hence it was not considered27.Cytochrome P450 2D7 (CYP2D7).04- 2.9928.Ras-related protein Rap-1A (RAP1A).003- 1.5929.Transforming protein RhoA (RHOA).04- 2.2330.Transgelin-2 (TAGLN2).001- 1.4831.Cystatin-S (CST4).04- 1.7432.Cysteine-rich secretory protein 3 (CRISP3).02- 0.6933.Protein FAM3D (FAM3D).009- 0.3634.Trypsin-1 (PRSS1).02- 1.0635.6-Phosphogluconate dehydrogenase (PGD)<.001- 1.436.Aldehyde dehydrogenase 3 (ALDH3)<.001- 1.2937.Secretory leukocyte protease inhibitor (SLPI)<.001- 2.1438.BPI fold–containing family A member 1 (BPIFA1).02- 2.5339.BPI fold–containing family B member 1 (BPIFB1).01- 2.6740.Cathepsin B (CTSB)<.001- 1.2541.Cellular retinoic acid–binding protein 2 (CRABP2).003- 1.3842.Cornulin (CRNN).004- 1.5643.Coronin-1A (CORO1A).04- 1.2444.Desmocollin-2 (DSC2).001- 1.2245.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH).01- 1.3646.GTP-binding nuclear protein Ran (RAN).03- 1.3447.Immunoglobulin heavy constant gamma 2 (IGHG2).001- 1.3848.L-lactate dehydrogenase B chain (LDHB).014- 1.2849.Lipocalin-1 (LCN1).01- 2.8350.Serpin B5 (SERPINB5).02- 1.2051.Small proline-rich protein 3 (SPRR3).04- 1.49Markerp -valueFCUp-regulated52.Histone H2B type 1-D (HIST1H2BD)0.04139.0853.Keratin, type II cytoskeletal 1 (KRT1)0.04182.2454.Ig lambda chain V-III region LOI (IGLV3)0.01542.2455.Lipocalin-1 (LCN1)0.0051.9156.Cystatin-B (CSTB)0.03381.9157.Fatty acid-binding protein epidermal (FABP5)0.0091.79Down-regulated58.Actin, alpha cardiac muscle 1 (ACTC1)0.0446−1.4259.Submaxillary gland androgen-regulated protein 3B (SMR3B)0.0148−1.6660.Glucose-6-phosphate isomerase (GPI)0.0309−2.0061.Leukocyte elastase inhibitor (SERPINB1)0.002- 2.1262.Peroxiredoxin-1 (PRDX1)0.0106−2.4363.Thymosin beta-4 (TMSB4X)0.0323−2.7764.ERO1-protein alpha (ERO1A)0.0126−2.943.309,506,765Flores et al., 2016 [15]PVL (Saliva)BrazilPVL (n = 15); Normal (n = 15)LC–MS/MS65.Cofilin-1 (CFL1)0.0423−3.12NA66.Nucleobindin-1 (NUCB1)0.007−3.1267.Rho GDP-dissociation inhibitor 2 (ARHGDIB)0.0317−3.3368.Adenylyl cyclase-associated protein 1 (CAP1)0.005−3.8469.Moesin (MSN)0.0107−4.1670.Transforming protein RhoA (RHOA)0.0453−4.1671.Calreticulin (CALR)0.0172−4.7672.Isoform 2 of Glycogen phosphorylase, liver form (PYGL)0.0239−5.0073.Actin-related protein 3 (ACTR3)0.0268−7.6974.Nicotinamide phosphoribosyl transferase (NAMPT)0.0139−7.6975.Chitinase-3- protein 2 (CHI3L2)0.0143−9.0976.Peroxiredoxin-6 (PRDX6)0.0255−12.577.Angiotensinogen (AGT)0.0006−12.578.Complement C4-B (C4B)0.004−14.2879.Dipeptidyl peptidase 1 (CTSC)0.015NC80.Isoform 3 of Nucleoside diphosphate kinase B (NME2)0.000NC81.Ubiquitin-modifier-activating enzyme 1 (UBA1)0.0189NC82.Interleukin-36 alpha (IL36A)0.0197NCOSMFMarkerp -valueFCDown-regulation4.53,131,473Mathew et al., 2018 [16]OSMF (Serum)IndiaOSMF (n = 5); normal (n = 5)LC–MS/MS1.Apolipoprotein A-II preproprotein (APOA2)NANANA2.Maleyl aceto-acetate isomerase isoform 2 (GSTZ1)NANA3.Apolipoprotein C-I precursor (APOC1)NANAMarkerp -valueFCUp-regulated4.Myosin light chain 1 (MYL1)NANA5.Heat shock 70 kDa protein (HSP70)NANA6.Beta-Enolase (ENO3)NANA7.Alpha-Enolase (ENO1)NANA8.Lumican (LUM)NANA9.Carbonic Anhydrase 2 (CA2)NANA10.Carbonic Anhydrase 1 (CA1)NANA11.Glutathione S-Transferase Pi 1 (GSTP1)NANA12.Hemoglobin Subunit Beta (HBB)NANA13.Keratin, type I cytoskeletal 9 (KRT9)NANA14.Keratin, type II cytoskeletal 3 (KRT3)NANA5.81,332,981Venugopal et al., 2022 [17]OSMF (Tissue)IndiaOSMF (n = 20); OSCC arising from OSMF(n = 5); Normal (n = 5)2D-GE and MALDI-TOF and LC–MS/MS15.Keratin, type II cytoskeletal 3 (KRT6C)NANAIHC- 14–3-3ε, CA1, HSP7016.Keratin 77 (KRT77)NANA17.Hemoglobin Subunit Alpha 1 (HBA1)NANA18.Albumin (ALB)NANA19.Annexin A2 (ANXA2)NANA20.Vimentin (VIM)NANA21.14–3-3 Protein Epsilon (YWHAE)NANA22.SCCA2/SCCA1 fusion protein isoform 1 (SERPINB4/SERPINB3)NANA23.Myosin Light Chain 6 (MYL6)NANA24.Profilin 1 (PFN1)NANA25.Heat Shock Protein 90 Alpha Family Class A Member 1 (HSP90AA1)NANA26.Dermcidin (DCD)NANAMarkerp -valueFCUp-regulated27.Hypothetical protein bcere 0001_7850NANA28.Heat shock protein 70 1B (HSP701B)NANA29.Lumican precursor (LUM)NANA30.Calreticulin (CRT)NANA31.Chain A apo form of human mitochondrial aldehyde dehydrogenase (ALDH2)NANA32.Actin α 1 (ACTA1)NANA33.Keratin 10 (KRT10)NANA34.Keratin 6B (KRT6B)NANA35.Galectin-7 (LGALS7)NANA36.Galectin-8 (LGALS8)NANA6.81,333,253Das et al., 2018 [18]OSMF (Tissue)IndiaOSMF (n = 63); Normal (n = 50)2 D gel Electrophoresis and MALDI-TOF37.Cyclophilin A (CypA)NANART-qPCR: Hsp-70 1B, CRT, LUM, ENO138.Squamous Cell Carcinoma Antigen Recognized by T-cells 3 (SART3)NANA39.Allo Albumin Venezia (mutALB)NANA40.Keratin 1 (KRT1)NANA41.ADP ribosylation factor 6 (ARF6)NANADown-regulated42.Mitochondrial ATP synthase subunit beta (ATP5B)NANA43.Enolase 1 (ENO1)NANA44.Acetyltransferase (GNAT)NANA45.Myosin subunit light chain (MYL)NANA46.Fatty acid binding protein 5 (FABP5)NANA47.Chain B high resolution X-ray study of deoxy recombinant human hemoglobin β-G (HBB)NANAMarkerp -valueFCUp-regulated48.Serum albumin (ALB)NANA49.40S ribosomal protein SA (RPSA)NANA50.lumican (LUM)NANA51.Keratin type I cytoskeletal 14 (KRT14)NANA52.Keratin type II cytoskeletal 1 (KRT1)NANA53.Keratin type I cytoskeletal 9 (KRT9)NANA54.Keratin type II cytoskeletal 4 (KRT4)NANA55.Keratin type II cytoskeletal 5 (KRT5)NANA56.Keratin type II cytoskeletal 8 (KRT8)NANA57.Myosin light chains 1 (MYL1)NANA58.Myosin light chains 3 (MYL3)NANA59.Myosin light chain 6B (MYL6B)NANA60.Myosin light polypeptide 6 (MYL6)NANA61.Myosin light chain (MYL2)NANA62.Myosin-7 (MYH7)NANA63.Hemoglobin subunits beta (HBB)NANA64.Hemoglobin subunit alpha (HBA)NANA65.(HBB2_TRENE)NANA66.Aspartate aminotransferase (AAT2)NANA67.Peptidylprolyl isomerase A (PPIA)NANA68.F-ATPase delta subunit (ATP5D)NANA69.Carbonic anhydrase isoforms 1 (CA1)NANA70.Carbonic anhydrase isoforms 3 (CA3)NANA71.Muscle creatine kinase ATP synthase subunit delta (ATPD)NANA72.Vimentin (VIM)NANA73.Beta-enolase (ENO3)NANA74.Glutathione S-transferase P (GSTP1)NANA75.Puromycin-sensitive aminopeptidase (PSA)NANA76.Microfibril-associated glycoprotein 4 (MFAP4)NANA77.Shank1 (SHANK1)NANA78.NADH dehydrogenase 1 alpha subcomplex 5 ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1 (UQCRFS1)NANA79.Mimecan (OGN)NANA80.profilin-1 (PFN1)NANA81.Cyclophilin A (CYPA)NANADown-regulated82.Glial fibrillary acidic protein (GFAP)NANA83.Cytokeratin (Krt9)NANA84.Keratin type II cytoskeletal 73 (KRT73)NANA85Keratin type II cytoskeletal 74 (KRT74)NANA86.Keratin type II cytoskeletal 75 (KRT75)NANA7.81,333,347Hou et al., 2017 [19]OSMF (Tissue)ChinaOSMF (n = 25); Normal (n = 25)2DE, MALDI-Q-TOF–MS87Calmodulin (CALM1)NANART-PCR and Western blot:CYPA88.Protein S100-A6 (S100A6)NANA89.Protein S100-A9 (S100A9)NANA90.Protein S100-A11 (S100A11)NANA91.Annexin A1 (ANXA1)NANA92.Annexin A2 (ANXA2)NANA93.Annexin A4 (ANXA4)NANA94.Annexin A5 (ANXA5)NANA95.Annexin A11 (ANXA11)NANA96.Protein disulfide-isomerase (P4HB)NANA97.TP-beta (HADHB)NANA98.Glycine amidino transferase (GATM)NANA99.Tubulin alpha-1A chain (TUBA1A)NANA100.Tubulin alpha-1B chain (TUBA1B)NANA101.Triose phosphate isomerase (TPI1)NANA102.Sero transferrin (TF)NANA103.Fatty acid-binding protein (FABP5)NANA104.Rab GDP dissociation inhibitor alpha (GDI1)NANA105.Actin cytoplasmic 1 (ACTB)NANA106.ANKRD26-family member 1 A (A26C1A)NANA107.Prolargin (PRELP)NANA108.alpha-1-acid glycoprotein 1 (ORM1)NANA109.Mitochondrial 28S ribosomal protein S31 (MRPS31)NANA110.Ferritin light chain (FTL)NANA111.Pyruvate kinase isozymes M2 (PKM2)NANA112.Replication protein A 32 kDa subunit (RPA2)NANA113.Stabilin-2 (STAB2)NANA114.Thioredoxin (TXN)NANA115.Carma 2 (CARD14)NANA116.Collagen alpha-3 VI chain (COL6A3)NANA117.Galectin-1 (LGALS1)NANA118.Proteasome activator complex subunit 2 (PSME2)NANA119.Alpha-1-antitrypsin (SERPINA1)NANA120.Superoxide dismutase (SOD1)NANA121.Superoxide dismutase (SOD2)NANA122.Glutathione S-transferase Mu 3 (GSTM3)NANA123.NADH-cytochrome b5 reductase 3 (CYB5R3)NANA124.Nitric oxide synthase endothelial (NOS3)NANA125.SH3 domain-binding glutamic acid-rich-protein 3 (SH3BGRL3)NANA126.Immunoglobulin gamma-4 chain C region (IGHG4)NANA127.Fibrinogen beta (FGB)NANA128.Fibrinogen gamma (FGG)NANA129.Zinc finger MYM-type protein 4 (ZMYM4)NANA130.Heat shock protein beta-8 (HSPB8)NANA131.Immunoglobulin kappa chain C region transthyretin (TTR)NANAMarkerp -valueFCUp-regulated132.α-Enolase (ENO1)NANA8.81,333,274Bag et al., 2018 [20]OSMF (Tissue)IndiaOSMF without dysplasia (n = 23); OSCC (n = 23); Normal (n = 13)1DE, nano LC MALDI tandem MSUnspecified regulation statusIHC: ENO1133.Keratin, type II cytoskeletal 1 (KRT1)NANA134.Keratin, type I cytoskeletal 9 (KRT9)NANA135.Pyruvate kinase (PKM)NANA136.Hornerin (HRNR)NANA137.Albumin (ALB)NANAMarkerp -valueFCDown-regulated138.Calmodulin-protein 3 (CALML3)NANA9.53,131,193Babu et al., 2021 [21]OSMF (Saliva)IndiaOSMF (n = 5); Normal (n = 5)Label-Free LC–MS/MS139.Small proline-rich protein 3 (SPRR3)NANANA140.Alpha-amylase 1 isoform X1 (AMY1A)NANA141.Kallikrein-6 isoform A preproprotein (KLK6)NANA142.Histone H4 (HIST1H4A)NANAMarkerp -valueFCUp-regulated143.Annexin A4 (ANXA4)NA2.051162243144.Microfibril-Associated Glycoprotein 4 (MFAP4)NA5.7543993145.Keratin 19 (KRT19)NA18.03017807146.Histone H4 Family Member I (HIST1H4I)NA2.582260132147.Vimentin (VIM)NA2.558585882148.Filamin-A (FLNA)NA2.128139019149.Keratin 7 (KRT7)NA4.285485268150.Collagen Alpha-2(I) Chain (COL1A2)NA2.167704105151.Collagen Alpha-1(I) Chain (COL1A1)NA2.884031534152.Glycerol-3-Phosphate Dehydrogenase 1 (GPD1)NA4.055085182153.Collagen Alpha-1(VI) Chain (COL6A1),NA2.83139205154.Myeloperoxidase (MPO)NA2.466039419155.Glutamate Oxaloacetate Transaminase 1, Cytoplasmic (GOT1)NA8.709635735156.Alcohol Dehydrogenase 4 (ADH4)NA5.64936971710.81,333,362Liu et al., 2016 [22]OSMF (Tissue)ChinaOSMF (n = 6); OSCC (n = 6); Normal (n = 6)iTRAQ-LC–MS/MS157.Glutathione S-Transferase Mu 1 (GSTM1)NA9.549925804Western blot and IHC: ANAX4, FLNA158.Calmodulin 1 (CALM1)NA13.55189419159.Cathepsin G (CTSG)NA5.105050087160.Serum Amyloid P Component (APCS)NA4.830587864Down-regulated161.Glycine Amidino-Transferase (GATM),NA0.343557954162.Carboxylesterase 1 (CES1),NA0.108642563163.Proteasome Activator Complex Subunit 1 (PSME1),NA0.296483129164.Leukotriene B4 12-Hydroxy-dehydrogenase (LTB4DH),NA0.405508548165.Heat Shock Protein 90 Beta Family Member 1 (HSP90B1),NA0.310455948166.Protein Disulfide-Isomerase A3 (PDIA3),NA0.237684026167.Transferrin (TF),NA0.322106868168.Fibrinogen Alpha Chain (FGA),NA0.432513833168.Protein Disulfide-Isomerase (P4HB),NA0.280543357170.Protein Disulfide-Isomerase A4 (PDIA4),NA0.187068209171.Epoxide Hydrolase 1 (EPHX1),NA0.157036275172.Heat Shock Protein Family A -Hsp70 Member 5 (HSPA5)NA0.23120648OLPMarkerp -valueFCUp-regulated1.Albumin (ALB)NANA2.Zinc-alpha-2-glycoprotein (AZGP1)NANA3.S100A9NANA4.S100A8NANA5.Haptoglobin (HP)NANA6.Haemoglobin subunit Beta (HBB)NANA11.81,333,249Souza et al., 2018 [23]OLP (Saliva)BrazilOLP (n = 10); Normal (n = 10)Mass Spectrometry (LC–MS/MS)7.Haemoglobin subunit Alpha (HBA2)NANACytokines has been validated whereas protein markers has not been validated8.Haemoglobin subunit Delta (HBD)NANA9.Catalase (CAT)NANA10.prolactin-inducible protein (PIP)NANA11.lysozyme (LYZ)NANA12.Beta-2-microglobulin (B2M)NANA13.Heat shock proteins (HSPA5)NANA14.Thymosin beta-4 (TMSB4)NANA15.Alpha-1-acidic glycoprotein 1 (ORM1)NANAMarkerp -valueFCUp-regulated16.Lysozyme (LYZ)NANA17.Tapasin (TAPBP)NANA18.Annexin A6 (ANXA6)NANA19.Intercellular Adhesion Molecule 1 (ICAM1)NANA20.Cathepsin Z (CTSZ)NANA21.Lamin B Receptor (LBR)NANA22.Tryptophanyl-tRNA Synthetase (WARS)NANA23.Protein Tyrosine Phosphatase Receptor Type C-Associated Protein (PTPRCAP)NANA12.81,332,878Xie et al., 2023 [24]OLP (Tissue)United StatesIndolent OLP (n = 6);Transforming OLP (n = 6);Normal (n = 5)LCM and nanoLC-MS/MS24.Guanylate-Binding Protein 1 (GBP1)NANANA25.Endoplasmic Reticulum Protein 29 (ERP29)NANA26.Protein Tyrosine Phosphatase Receptor Type C (PTPRC)NANA27.Superoxide Dismutase 2 (SOD2)NANA28.Microtubule-associated protein 7 Domain Containing 1 (MAP7D1)NANA29.Alpha subunit of the HLA-DR molecule (HLA-DRA)NANADown-regulated30.Alpha-2-Macroglobulin-1 (A2ML1)NANA(Requested for supplementary tables)Markerp> -valueFCUp-regulated31.Serum albumin (ALB)NA3.032.Alpha-amylase (AMY1A)NA2.133.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)NA2.834.Short-chain specific acyl-CoA dehydrogenase, mitochondrial (ACADS)NA2.635.Ig J-chain, partial (JCHAIN)NA2.036.Neutrophil gelatinase-associated lipocalin (NGAL/LCN2)NA2.437.NOL1/NOP2/Sun domain family, member 4 (NSUN4)NA3.213.81,333,268Talungchit et al., 2018 [25]OLP (Saliva)ThailandOLP (n = 5); Normal (n = 5)2DE38.14–3-3 protein zeta/delta (YWHAZ)NANAELISA and Immunoblotting analysis:C3c, Fibrinogen fragment D, Cystatin SA39.Haptoglobin (Hp)NANA40.Chain C Human Complement Component (C3c)NANA41.Chain B Crystal Structure of Fibrinogen Fragment D (-)NANA42.Cytokeratin 9 (KRT9)NANADown-regulated43.Carbonic anhydrase isozyme VI (CA6)NA−1.844.Cystatin precursor (CST3)NA−1.545.Cystatin S precursor (CST4)NA−2.146.Cystatin SA (CST2)NA−1.6 14.81,333,714Yang et al., 2006 [26]OLP (Saliva)ChinaOLP (n = 6); Normal (n = 6)2DE, MALDI-TOF MSMarkerpBoldItalic> -valueFCNAUp-regulated47.Prokallikrein (KLKB1)NANADown-regulated48.short palate, lung, and nasal epithelium carcinoma-associated protein (PLUNC)NANAMarkerp -valueFCUp-regulated49.Fibrinogen gamma chain isoform gamma-A precursor (FGG) [Serum],NANA15.309,495,827Guadalupe et al., 2018 [27]OLP (Serum, Saliva)MexicoOLP (n = 1); Normal (n = 4)1DE, 2DE, LC–MS/MS50.Chain A, human salivary amylase (AMY1A) [Saliva]NANANA(Tissue samples were included but tissue markers were not mentioned)NANATable 2Comprehensive overview of protein expression in OPMDsOPMDsNo. of StudiesUp regulatedDown regulatedUnspecifiedTotalSalivaSerumTissueSalivaSerumTissueSalivaSerumTissueSalivaSerumTissueLK1--3--2--9--14PVL2--8--74-----82OSMF115--915368--5172OLP3 + 1* + 1*1291145-1---5015401105863699-5318+ 1*Serum and saliva sample Table 1Characteristics and key findings of included proteomic studies in Oral Potentially Malignant Disorders (OPMDs)Sl. noStudy IDAuthorSampleCountrySample sizeTechniqueProteomic markerValidationLKMarkerp -valueFCUp-regulated1.Ig alpha-2 chain C region (IGHA2)< 0.05↑4.22.Apolipoprotein A1 (APOA1)< 0.05↑3.43.Mature metal chelatase catalytic antibody with hapten, chain A (MMCA)< 0.01↑10.1Down-regulated4.Cystatin-SN precursor (CST1)< 0.01↓5.55.Serum albumin (ALB)< 0.01↓11.41.309,495,928Camisasca et al., 2017 [13]LK (Saliva)BrazilNormal (n = 10); LK (n = 10)2DE, MALDI-TOF/TOF MSUnspecifiedIHC:Tissue: KRT106.Alpha-amylase (AMY1A)NANA7.Keratin-10 (KRT10)NANA8.IgM kappa IIIb (IgM κ IIIb)NANA9.Lysozyme precursor (LYZ)NANA10.Polymeric immunoglobulin receptor (PIGR)NANA11.Alpha-amylase 1B (AMY1B)NANA12.Ig gamma-1 chain C region (IGHG1)NANA13.IgG kappa chain (IGHG)NANA14.Immunoglobulin kappa light chain (IGK)NANAPVLMarkerq -valueFCUp-regulated1.Nucleobindin-2 (NUCB2).010.202.Chitotriosidase-1 (CHIT1).032.18Down-regulated3.Adenosyl homocysteinase (AHCY).002- 1.844.Beta-2 microglobulin (B2M).002- 1.415.BPI fold–containing family B member 2 (BPIFB2).01- 1.716.Carboxypeptidase E (CPE).007- 1.767.Cell division control protein 42 homolog (CDC42).001- 2.118.Chloride intracellular channel protein 1 (CLIC1).001- 2.199.Carboxylesterase 2 (CES2)<.001- 1.7210.Desmoglein-1 (DSG1).002- 1.7511.ERO1-like protein alpha (ERO1A)<.001- 1.8512.Heme-binding protein 2 (HEBP2)<.001- 1.8013.IgG Fc-binding protein (FCGBP).03- 1.7714.Interleukin-1 receptor antagonist protein (IL1RN).001- 1.4015.Interleukin-36 alpha (IL36A)<.001- 2.5416.Kallikrein-11 (KLK11).04- 1.1317.Keratin type I cytoskeletal 13 (KRT13).01- 1.1018.Keratin type I cytoskeletal 14 (KRT14).02- 1.7719.Carboxylesterase 1 (CES1).01- 2.0320.Ly6/PLAUR domain-containing protein 3 (LYPD3).008- 2.4721.Mesothelin (MSLN).01- 3.1722.Mucin-5B (MUC5B).04- 2.3323.Nucleobindin-1 (NUCB1).01- 1.6724.Nucleotide exchange factor SIL1 (SIL1).01- 1.2725.Prostaglandin reductase 1 (PTGR1).001- 1.642.81,332,839Arroyo et al., 2023 [14]PVL (Saliva)Brazil, Spain, and FinlandPVL (n = 7); LK (n = 8); Normal (n = 5)LC–MS/MS26.Proteasome activator complex subunit 2 (PSME2)<.001- 2.27Validation has been performed for tissue markers for which the control group was inflammatory fibrous hyperplasia hence it was not considered27.Cytochrome P450 2D7 (CYP2D7).04- 2.9928.Ras-related protein Rap-1A (RAP1A).003- 1.5929.Transforming protein RhoA (RHOA).04- 2.2330.Transgelin-2 (TAGLN2).001- 1.4831.Cystatin-S (CST4).04- 1.7432.Cysteine-rich secretory protein 3 (CRISP3).02- 0.6933.Protein FAM3D (FAM3D).009- 0.3634.Trypsin-1 (PRSS1).02- 1.0635.6-Phosphogluconate dehydrogenase (PGD)<.001- 1.436.Aldehyde dehydrogenase 3 (ALDH3)<.001- 1.2937.Secretory leukocyte protease inhibitor (SLPI)<.001- 2.1438.BPI fold–containing family A member 1 (BPIFA1).02- 2.5339.BPI fold–containing family B member 1 (BPIFB1).01- 2.6740.Cathepsin B (CTSB)<.001- 1.2541.Cellular retinoic acid–binding protein 2 (CRABP2).003- 1.3842.Cornulin (CRNN).004- 1.5643.Coronin-1A (CORO1A).04- 1.2444.Desmocollin-2 (DSC2).001- 1.2245.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH).01- 1.3646.GTP-binding nuclear protein Ran (RAN).03- 1.3447.Immunoglobulin heavy constant gamma 2 (IGHG2).001- 1.3848.L-lactate dehydrogenase B chain (LDHB).014- 1.2849.Lipocalin-1 (LCN1).01- 2.8350.Serpin B5 (SERPINB5).02- 1.2051.Small proline-rich protein 3 (SPRR3).04- 1.49Markerp -valueFCUp-regulated52.Histone H2B type 1-D (HIST1H2BD)0.04139.0853.Keratin, type II cytoskeletal 1 (KRT1)0.04182.2454.Ig lambda chain V-III region LOI (IGLV3)0.01542.2455.Lipocalin-1 (LCN1)0.0051.9156.Cystatin-B (CSTB)0.03381.9157.Fatty acid-binding protein epidermal (FABP5)0.0091.79Down-regulated58.Actin, alpha cardiac muscle 1 (ACTC1)0.0446−1.4259.Submaxillary gland androgen-regulated protein 3B (SMR3B)0.0148−1.6660.Glucose-6-phosphate isomerase (GPI)0.0309−2.0061.Leukocyte elastase inhibitor (SERPINB1)0.002- 2.1262.Peroxiredoxin-1 (PRDX1)0.0106−2.4363.Thymosin beta-4 (TMSB4X)0.0323−2.7764.ERO1-protein alpha (ERO1A)0.0126−2.943.309,506,765Flores et al., 2016 [15]PVL (Saliva)BrazilPVL (n = 15); Normal (n = 15)LC–MS/MS65.Cofilin-1 (CFL1)0.0423−3.12NA66.Nucleobindin-1 (NUCB1)0.007−3.1267.Rho GDP-dissociation inhibitor 2 (ARHGDIB)0.0317−3.3368.Adenylyl cyclase-associated protein 1 (CAP1)0.005−3.8469.Moesin (MSN)0.0107−4.1670.Transforming protein RhoA (RHOA)0.0453−4.1671.Calreticulin (CALR)0.0172−4.7672.Isoform 2 of Glycogen phosphorylase, liver form (PYGL)0.0239−5.0073.Actin-related protein 3 (ACTR3)0.0268−7.6974.Nicotinamide phosphoribosyl transferase (NAMPT)0.0139−7.6975.Chitinase-3- protein 2 (CHI3L2)0.0143−9.0976.Peroxiredoxin-6 (PRDX6)0.0255−12.577.Angiotensinogen (AGT)0.0006−12.578.Complement C4-B (C4B)0.004−14.2879.Dipeptidyl peptidase 1 (CTSC)0.015NC80.Isoform 3 of Nucleoside diphosphate kinase B (NME2)0.000NC81.Ubiquitin-modifier-activating enzyme 1 (UBA1)0.0189NC82.Interleukin-36 alpha (IL36A)0.0197NCOSMFMarkerp -valueFCDown-regulation4.53,131,473Mathew et al., 2018 [16]OSMF (Serum)IndiaOSMF (n = 5); normal (n = 5)LC–MS/MS1.Apolipoprotein A-II preproprotein (APOA2)NANANA2.Maleyl aceto-acetate isomerase isoform 2 (GSTZ1)NANA3.Apolipoprotein C-I precursor (APOC1)NANAMarkerp -valueFCUp-regulated4.Myosin light chain 1 (MYL1)NANA5.Heat shock 70 kDa protein (HSP70)NANA6.Beta-Enolase (ENO3)NANA7.Alpha-Enolase (ENO1)NANA8.Lumican (LUM)NANA9.Carbonic Anhydrase 2 (CA2)NANA10.Carbonic Anhydrase 1 (CA1)NANA11.Glutathione S-Transferase Pi 1 (GSTP1)NANA12.Hemoglobin Subunit Beta (HBB)NANA13.Keratin, type I cytoskeletal 9 (KRT9)NANA14.Keratin, type II cytoskeletal 3 (KRT3)NANA5.81,332,981Venugopal et al., 2022 [17]OSMF (Tissue)IndiaOSMF (n = 20); OSCC arising from OSMF(n = 5); Normal (n = 5)2D-GE and MALDI-TOF and LC–MS/MS15.Keratin, type II cytoskeletal 3 (KRT6C)NANAIHC- 14–3-3ε, CA1, HSP7016.Keratin 77 (KRT77)NANA17.Hemoglobin Subunit Alpha 1 (HBA1)NANA18.Albumin (ALB)NANA19.Annexin A2 (ANXA2)NANA20.Vimentin (VIM)NANA21.14–3-3 Protein Epsilon (YWHAE)NANA22.SCCA2/SCCA1 fusion protein isoform 1 (SERPINB4/SERPINB3)NANA23.Myosin Light Chain 6 (MYL6)NANA24.Profilin 1 (PFN1)NANA25.Heat Shock Protein 90 Alpha Family Class A Member 1 (HSP90AA1)NANA26.Dermcidin (DCD)NANAMarkerp -valueFCUp-regulated27.Hypothetical protein bcere 0001_7850NANA28.Heat shock protein 70 1B (HSP701B)NANA29.Lumican precursor (LUM)NANA30.Calreticulin (CRT)NANA31.Chain A apo form of human mitochondrial aldehyde dehydrogenase (ALDH2)NANA32.Actin α 1 (ACTA1)NANA33.Keratin 10 (KRT10)NANA34.Keratin 6B (KRT6B)NANA35.Galectin-7 (LGALS7)NANA36.Galectin-8 (LGALS8)NANA6.81,333,253Das et al., 2018 [18]OSMF (Tissue)IndiaOSMF (n = 63); Normal (n = 50)2 D gel Electrophoresis and MALDI-TOF37.Cyclophilin A (CypA)NANART-qPCR: Hsp-70 1B, CRT, LUM, ENO138.Squamous Cell Carcinoma Antigen Recognized by T-cells 3 (SART3)NANA39.Allo Albumin Venezia (mutALB)NANA40.Keratin 1 (KRT1)NANA41.ADP ribosylation factor 6 (ARF6)NANADown-regulated42.Mitochondrial ATP synthase subunit beta (ATP5B)NANA43.Enolase 1 (ENO1)NANA44.Acetyltransferase (GNAT)NANA45.Myosin subunit light chain (MYL)NANA46.Fatty acid binding protein 5 (FABP5)NANA47.Chain B high resolution X-ray study of deoxy recombinant human hemoglobin β-G (HBB)NANAMarkerp -valueFCUp-regulated48.Serum albumin (ALB)NANA49.40S ribosomal protein SA (RPSA)NANA50.lumican (LUM)NANA51.Keratin type I cytoskeletal 14 (KRT14)NANA52.Keratin type II cytoskeletal 1 (KRT1)NANA53.Keratin type I cytoskeletal 9 (KRT9)NANA54.Keratin type II cytoskeletal 4 (KRT4)NANA55.Keratin type II cytoskeletal 5 (KRT5)NANA56.Keratin type II cytoskeletal 8 (KRT8)NANA57.Myosin light chains 1 (MYL1)NANA58.Myosin light chains 3 (MYL3)NANA59.Myosin light chain 6B (MYL6B)NANA60.Myosin light polypeptide 6 (MYL6)NANA61.Myosin light chain (MYL2)NANA62.Myosin-7 (MYH7)NANA63.Hemoglobin subunits beta (HBB)NANA64.Hemoglobin subunit alpha (HBA)NANA65.(HBB2_TRENE)NANA66.Aspartate aminotransferase (AAT2)NANA67.Peptidylprolyl isomerase A (PPIA)NANA68.F-ATPase delta subunit (ATP5D)NANA69.Carbonic anhydrase isoforms 1 (CA1)NANA70.Carbonic anhydrase isoforms 3 (CA3)NANA71.Muscle creatine kinase ATP synthase subunit delta (ATPD)NANA72.Vimentin (VIM)NANA73.Beta-enolase (ENO3)NANA74.Glutathione S-transferase P (GSTP1)NANA75.Puromycin-sensitive aminopeptidase (PSA)NANA76.Microfibril-associated glycoprotein 4 (MFAP4)NANA77.Shank1 (SHANK1)NANA78.NADH dehydrogenase 1 alpha subcomplex 5 ubiquinol-cytochrome c reductase Rieske iron-sulfur polypeptide 1 (UQCRFS1)NANA79.Mimecan (OGN)NANA80.profilin-1 (PFN1)NANA81.Cyclophilin A (CYPA)NANADown-regulated82.Glial fibrillary acidic protein (GFAP)NANA83.Cytokeratin (Krt9)NANA84.Keratin type II cytoskeletal 73 (KRT73)NANA85Keratin type II cytoskeletal 74 (KRT74)NANA86.Keratin type II cytoskeletal 75 (KRT75)NANA7.81,333,347Hou et al., 2017 [19]OSMF (Tissue)ChinaOSMF (n = 25); Normal (n = 25)2DE, MALDI-Q-TOF–MS87Calmodulin (CALM1)NANART-PCR and Western blot:CYPA88.Protein S100-A6 (S100A6)NANA89.Protein S100-A9 (S100A9)NANA90.Protein S100-A11 (S100A11)NANA91.Annexin A1 (ANXA1)NANA92.Annexin A2 (ANXA2)NANA93.Annexin A4 (ANXA4)NANA94.Annexin A5 (ANXA5)NANA95.Annexin A11 (ANXA11)NANA96.Protein disulfide-isomerase (P4HB)NANA97.TP-beta (HADHB)NANA98.Glycine amidino transferase (GATM)NANA99.Tubulin alpha-1A chain (TUBA1A)NANA100.Tubulin alpha-1B chain (TUBA1B)NANA101.Triose phosphate isomerase (TPI1)NANA102.Sero transferrin (TF)NANA103.Fatty acid-binding protein (FABP5)NANA104.Rab GDP dissociation inhibitor alpha (GDI1)NANA105.Actin cytoplasmic 1 (ACTB)NANA106.ANKRD26-family member 1 A (A26C1A)NANA107.Prolargin (PRELP)NANA108.alpha-1-acid glycoprotein 1 (ORM1)NANA109.Mitochondrial 28S ribosomal protein S31 (MRPS31)NANA110.Ferritin light chain (FTL)NANA111.Pyruvate kinase isozymes M2 (PKM2)NANA112.Replication protein A 32 kDa subunit (RPA2)NANA113.Stabilin-2 (STAB2)NANA114.Thioredoxin (TXN)NANA115.Carma 2 (CARD14)NANA116.Collagen alpha-3 VI chain (COL6A3)NANA117.Galectin-1 (LGALS1)NANA118.Proteasome activator complex subunit 2 (PSME2)NANA119.Alpha-1-antitrypsin (SERPINA1)NANA120.Superoxide dismutase (SOD1)NANA121.Superoxide dismutase (SOD2)NANA122.Glutathione S-transferase Mu 3 (GSTM3)NANA123.NADH-cytochrome b5 reductase 3 (CYB5R3)NANA124.Nitric oxide synthase endothelial (NOS3)NANA125.SH3 domain-binding glutamic acid-rich-protein 3 (SH3BGRL3)NANA126.Immunoglobulin gamma-4 chain C region (IGHG4)NANA127.Fibrinogen beta (FGB)NANA128.Fibrinogen gamma (FGG)NANA129.Zinc finger MYM-type protein 4 (ZMYM4)NANA130.Heat shock protein beta-8 (HSPB8)NANA131.Immunoglobulin kappa chain C region transthyretin (TTR)NANAMarkerp -valueFCUp-regulated132.α-Enolase (ENO1)NANA8.81,333,274Bag et al., 2018 [20]OSMF (Tissue)IndiaOSMF without dysplasia (n = 23); OSCC (n = 23); Normal (n = 13)1DE, nano LC MALDI tandem MSUnspecified regulation statusIHC: ENO1133.Keratin, type II cytoskeletal 1 (KRT1)NANA134.Keratin, type I cytoskeletal 9 (KRT9)NANA135.Pyruvate kinase (PKM)NANA136.Hornerin (HRNR)NANA137.Albumin (ALB)NANAMarkerp -valueFCDown-regulated138.Calmodulin-protein 3 (CALML3)NANA9.53,131,193Babu et al., 2021 [21]OSMF (Saliva)IndiaOSMF (n = 5); Normal (n = 5)Label-Free LC–MS/MS139.Small proline-rich protein 3 (SPRR3)NANANA140.Alpha-amylase 1 isoform X1 (AMY1A)NANA141.Kallikrein-6 isoform A preproprotein (KLK6)NANA142.Histone H4 (HIST1H4A)NANAMarkerp -valueFCUp-regulated143.Annexin A4 (ANXA4)NA2.051162243144.Microfibril-Associated Glycoprotein 4 (MFAP4)NA5.7543993145.Keratin 19 (KRT19)NA18.03017807146.Histone H4 Family Member I (HIST1H4I)NA2.582260132147.Vimentin (VIM)NA2.558585882148.Filamin-A (FLNA)NA2.128139019149.Keratin 7 (KRT7)NA4.285485268150.Collagen Alpha-2(I) Chain (COL1A2)NA2.167704105151.Collagen Alpha-1(I) Chain (COL1A1)NA2.884031534152.Glycerol-3-Phosphate Dehydrogenase 1 (GPD1)NA4.055085182153.Collagen Alpha-1(VI) Chain (COL6A1),NA2.83139205154.Myeloperoxidase (MPO)NA2.466039419155.Glutamate Oxaloacetate Transaminase 1, Cytoplasmic (GOT1)NA8.709635735156.Alcohol Dehydrogenase 4 (ADH4)NA5.64936971710.81,333,362Liu et al., 2016 [22]OSMF (Tissue)ChinaOSMF (n = 6); OSCC (n = 6); Normal (n = 6)iTRAQ-LC–MS/MS157.Glutathione S-Transferase Mu 1 (GSTM1)NA9.549925804Western blot and IHC: ANAX4, FLNA158.Calmodulin 1 (CALM1)NA13.55189419159.Cathepsin G (CTSG)NA5.105050087160.Serum Amyloid P Component (APCS)NA4.830587864Down-regulated161.Glycine Amidino-Transferase (GATM),NA0.343557954162.Carboxylesterase 1 (CES1),NA0.108642563163.Proteasome Activator Complex Subunit 1 (PSME1),NA0.296483129164.Leukotriene B4 12-Hydroxy-dehydrogenase (LTB4DH),NA0.405508548165.Heat Shock Protein 90 Beta Family Member 1 (HSP90B1),NA0.310455948166.Protein Disulfide-Isomerase A3 (PDIA3),NA0.237684026167.Transferrin (TF),NA0.322106868168.Fibrinogen Alpha Chain (FGA),NA0.432513833168.Protein Disulfide-Isomerase (P4HB),NA0.280543357170.Protein Disulfide-Isomerase A4 (PDIA4),NA0.187068209171.Epoxide Hydrolase 1 (EPHX1),NA0.157036275172.Heat Shock Protein Family A -Hsp70 Member 5 (HSPA5)NA0.23120648OLPMarkerp -valueFCUp-regulated1.Albumin (ALB)NANA2.Zinc-alpha-2-glycoprotein (AZGP1)NANA3.S100A9NANA4.S100A8NANA5.Haptoglobin (HP)NANA6.Haemoglobin subunit Beta (HBB)NANA11.81,333,249Souza et al., 2018 [23]OLP (Saliva)BrazilOLP (n = 10); Normal (n = 10)Mass Spectrometry (LC–MS/MS)7.Haemoglobin subunit Alpha (HBA2)NANACytokines has been validated whereas protein markers has not been validated8.Haemoglobin subunit Delta (HBD)NANA9.Catalase (CAT)NANA10.prolactin-inducible protein (PIP)NANA11.lysozyme (LYZ)NANA12.Beta-2-microglobulin (B2M)NANA13.Heat shock proteins (HSPA5)NANA14.Thymosin beta-4 (TMSB4)NANA15.Alpha-1-acidic glycoprotein 1 (ORM1)NANAMarkerp -valueFCUp-regulated16.Lysozyme (LYZ)NANA17.Tapasin (TAPBP)NANA18.Annexin A6 (ANXA6)NANA19.Intercellular Adhesion Molecule 1 (ICAM1)NANA20.Cathepsin Z (CTSZ)NANA21.Lamin B Receptor (LBR)NANA22.Tryptophanyl-tRNA Synthetase (WARS)NANA23.Protein Tyrosine Phosphatase Receptor Type C-Associated Protein (PTPRCAP)NANA12.81,332,878Xie et al., 2023 [24]OLP (Tissue)United StatesIndolent OLP (n = 6);Transforming OLP (n = 6);Normal (n = 5)LCM and nanoLC-MS/MS24.Guanylate-Binding Protein 1 (GBP1)NANANA25.Endoplasmic Reticulum Protein 29 (ERP29)NANA26.Protein Tyrosine Phosphatase Receptor Type C (PTPRC)NANA27.Superoxide Dismutase 2 (SOD2)NANA28.Microtubule-associated protein 7 Domain Containing 1 (MAP7D1)NANA29.Alpha subunit of the HLA-DR molecule (HLA-DRA)NANADown-regulated30.Alpha-2-Macroglobulin-1 (A2ML1)NANA(Requested for supplementary tables)Markerp> -valueFCUp-regulated31.Serum albumin (ALB)NA3.032.Alpha-amylase (AMY1A)NA2.133.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)NA2.834.Short-chain specific acyl-CoA dehydrogenase, mitochondrial (ACADS)NA2.635.Ig J-chain, partial (JCHAIN)NA2.036.Neutrophil gelatinase-associated lipocalin (NGAL/LCN2)NA2.437.NOL1/NOP2/Sun domain family, member 4 (NSUN4)NA3.213.81,333,268Talungchit et al., 2018 [25]OLP (Saliva)ThailandOLP (n = 5); Normal (n = 5)2DE38.14–3-3 protein zeta/delta (YWHAZ)NANAELISA and Immunoblotting analysis:C3c, Fibrinogen fragment D, Cystatin SA39.Haptoglobin (Hp)NANA40.Chain C Human Complement Component (C3c)NANA41.Chain B Crystal Structure of Fibrinogen Fragment D (-)NANA42.Cytokeratin 9 (KRT9)NANADown-regulated43.Carbonic anhydrase isozyme VI (CA6)NA−1.844.Cystatin precursor (CST3)NA−1.545.Cystatin S precursor (CST4)NA−2.146.Cystatin SA (CST2)NA−1.6 14.81,333,714Yang et al., 2006 [26]OLP (Saliva)ChinaOLP (n = 6); Normal (n = 6)2DE, MALDI-TOF MSMarkerpBoldItalic> -valueFCNAUp-regulated47.Prokallikrein (KLKB1)NANADown-regulated48.short palate, lung, and nasal epithelium carcinoma-associated protein (PLUNC)NANAMarkerp -valueFCUp-regulated49.Fibrinogen gamma chain isoform gamma-A precursor (FGG) [Serum],NANA15.309,495,827Guadalupe et al., 2018 [27]OLP (Serum, Saliva)MexicoOLP (n = 1); Normal (n = 4)1DE, 2DE, LC–MS/MS50.Chain A, human salivary amylase (AMY1A) [Saliva]NANANA(Tissue samples were included but tissue markers were not mentioned)NANA Characteristics and key findings of included proteomic studies in Oral Potentially Malignant Disorders (OPMDs) Table 2Comprehensive overview of protein expression in OPMDsOPMDsNo. of StudiesUp regulatedDown regulatedUnspecifiedTotalSalivaSerumTissueSalivaSerumTissueSalivaSerumTissueSalivaSerumTissueLK1--3--2--9--14PVL2--8--74-----82OSMF115--915368--5172OLP3 + 1* + 1*1291145-1---5015401105863699-5318+ 1*Serum and saliva sample Comprehensive overview of protein expression in OPMDs + 1*Serum and saliva sample Across the included articles, seven had performed validation of proteomic markers using techniques such as ELISA, qRT-PCR, Western blotting, immunohistochemistry, and multiplex cytokine analysis. The systematic review identified 318 proteins, of which only 14 were validated markers. The validated markers for LK- KRT10; for OSMF – 14–3-3ε, CA1, HSP70, Hsp-70 1B, CRT, LUM, ENO1, CYPA, ANAX4 and FLNA; OLP- C3c, cystatin SA and Fibrinogen Fragment D. There was no validation study reported for PVL. ENO1 was validated in two independent OSMF studies; however, conflicting expression patterns were observed, with one study reporting upregulation [20] and another reporting downregulation [18]. Vote counting analysis of differentially expressed proteins A total of three lesions (PVL, OSMF, and OLP) were included in the vote-counting analysis, while LK was excluded due to the availability of only a single study, limiting the ability to assess consistency across studies. PVL revealed a predominant downregulation pattern with ERO1A, NUCB1, RHOA, and IL36A consistently decreased across two independent studies. The downregulation of ERO1A and NUCB1 may promote cellular proliferation and inhibit apoptosis, with ERO1A potentially acting through the PI3K/AKT pathway. RHOA exhibits a dual function, as it can both promote and suppress cancer progression depending on the cellular context. The downregulation of pro-inflammatory cytokine IL36A may contribute to tumor progression. The aggressive nature of the lesion can be supported by these alterations in combination. The expression pattern of LCN1 was contradictory (upregulated in one study and downregulated in another), which made it hard to reach a clear conclusion. In three studies, OSMF showed consistent upregulation of LUM, KRT1, KRT9, ALB, and VIM. Two studies showed a consistent upregulation of HSP70, CA1, GSTP1, and PFN1, while a group of proteins, including FABP5, PSME2, TF, and GATM, showed a consistent downregulation. Proteins like ENO1, HBB, and ANXA2 showed contradictory expression patterns. The frequently identified two proteins were LUM and VIM, and they appeared to be functionally interconnected. Modulation of the miR-200 family has been linked to the upregulation of LUM, resulting in a decrease in E-cadherin expression and an increase in VIM expression. The promotion of this shift contributes to the epithelial-mesenchymal transition (EMT), which is a crucial pathological pathway in OSMF. OLP demonstrated a uniform upregulation of ALB, LYZ, HP, HBB, and AMY1A consistently increased across studies. Among those, protein LYZ and HP seem to be increased in the case of inflammation,which substantiates the progression of OLP (Fig. 2).Fig. 2Assessment of differentially expressed protein markers across PVL, OSMF, and OLP using Vote Counting method. Green = PVL; Orange = OSMF; Yellow = OLP. Positive values indicate upregulation, negative values indicate downregulation. The height of each bar reflects how many studies support the differential expression. Study sources are indicated by author abbreviations listed above the graph. Abbreviations: A23 (Arroyo 2023), F16 (Flores 2016), V22 (Venugopal 2022), D18 (Das 2018), H17 (Hou 2017), B18 (Bag 2018), L16 (Liu 2016), S18 (Souza 2018), T18 (Talungchit 2018), X23 (Xie 2023), G18 (Guadalupe 2018) Assessment of differentially expressed protein markers across PVL, OSMF, and OLP using Vote Counting method. Green = PVL; Orange = OSMF; Yellow = OLP. Positive values indicate upregulation, negative values indicate downregulation. The height of each bar reflects how many studies support the differential expression. Study sources are indicated by author abbreviations listed above the graph. Abbreviations: A23 (Arroyo 2023), F16 (Flores 2016), V22 (Venugopal 2022), D18 (Das 2018), H17 (Hou 2017), B18 (Bag 2018), L16 (Liu 2016), S18 (Souza 2018), T18 (Talungchit 2018), X23 (Xie 2023), G18 (Guadalupe 2018) OSMF and OLP share common upregulated proteins, particularly ALB and HBB, likely reflecting inflammation-association. In contrast, PVL exhibits a distinct proteomic signature marked by widespread downregulation, underscoring a potentially different pathogenic mechanism that warrants further investigation. Quality assessment The NOS was applied to score and classify the quality of the included articles. According to NOS, the majority of the studies (n = 11) were classified as moderate quality, with 6 stars and the remaining studies (n = 4) were classified as high quality, with 7 stars. The main limitation across the included studies was observed in the comparability domain. Most of the included studies failed to match confounding factors such as age, gender, and habit-related exposures (e.g., tobacco or areca nut use) among groups. This leads to a potential risk of confounding bias. All the included studies exhibited a strong selection domain, indicating the presence of adequate case definition and control selection. In the exposure domain, minor limitations were noted, particularly due to inadequate reporting of non-response rates and lack of blinding in some studies (Table 3).Table 3Risk of Bias—NOSStudySelection Stars (max 4)Comparability Stars (max 2)Exposure Stars (max 3)Total Stars (max 9)QualityCamisasca et al. 2017 [13]★★★★-★★6ModerateArroyo et al. 2023 [14]★★★★-★★6ModerateFlores et al. 2016 [15]★★★★★★★7HighMathew et al. 2018 [16]★★★★★★★7HighVenugopal et al. 2022 [17]★★★★-★★6ModerateDas et al. 2018 [18]★★★★-★★6ModerateHou et al. 2017 [19]★★★★-★★6ModerateBag et al. 2018 [20]★★★★-★★6ModerateBabu et al. 2021 [21]★★★★-★★6ModerateLiu et al. 2016 [22]★★★★-★★6ModerateSouza et al. 2018 [23]★★★★-★★6ModerateXie et al. 2023 [24]★★★★-★★6ModerateTalungchit et al. 2018 [25]★★★★★★★7HighYang et al. 2006 [26]★★★★★★★7HighGuadalupe et al. 2018 [27]★★★★-★★6Moderate Table 3Risk of Bias—NOSStudySelection Stars (max 4)Comparability Stars (max 2)Exposure Stars (max 3)Total Stars (max 9)QualityCamisasca et al. 2017 [13]★★★★-★★6ModerateArroyo et al. 2023 [14]★★★★-★★6ModerateFlores et al. 2016 [15]★★★★★★★7HighMathew et al. 2018 [16]★★★★★★★7HighVenugopal et al. 2022 [17]★★★★-★★6ModerateDas et al. 2018 [18]★★★★-★★6ModerateHou et al. 2017 [19]★★★★-★★6ModerateBag et al. 2018 [20]★★★★-★★6ModerateBabu et al. 2021 [21]★★★★-★★6ModerateLiu et al. 2016 [22]★★★★-★★6ModerateSouza et al. 2018 [23]★★★★-★★6ModerateXie et al. 2023 [24]★★★★-★★6ModerateTalungchit et al. 2018 [25]★★★★★★★7HighYang et al. 2006 [26]★★★★★★★7HighGuadalupe et al. 2018 [27]★★★★-★★6Moderate Risk of Bias—NOS The GRADE framework revealed the certainty of evidence to be low for LK, very low for PVL, OSMF and LP. The detailed GRADE assessments are presented in Supplementary Table S1. Bipartite disease–protein network analysis The bipartite disease–protein network revealed that each lesion was represented by a unique cluster, and between the clusters, shared markers were present. LK is represented as a small cluster with shared proteins between OSMF and OLP; the lack of unique markers indicated the deficit of studies about LK. The PVL cluster formed as a moderately dense cluster with a good number of unique markers and a few shared ones. OSMF had the biggest and densest cluster among the lesions, which indicates the complicated pathological profile of the disease. OLP formed a moderately dense cluster, with adequate unique markers and a chunk of shared markers between the other three lesions. The bipartite network revealed that FGG, ORM1, SOD2, HSPA5, KRT9, S100A9, and HBB were shared between OSMF and OLP. CES1, FABP5, KRT1, SERPINA3, KRT14, and PSME2 were found to be shared between PVL and OSMF. CST4 and B2M were found between PVL and OLP. LYZ was common between LK and OLP. KRT10 was shared between LK and OSMF, and AMY1A and ALB were shared among LK, OLP, and OSMF. The proteins that were shared between the OPMDs indicate that they share fundamental molecular pathways, particularly involving inflammation, immune response, and oxidative stress, which may be responsible for disease progression (Fig. 3).Fig. 3A bipartite network illustrating the association between PVL, LK, OLP, and OSMF—and their corresponding proteins constructed using Cytoscape. Circular nodes represent diseases and square nodes represent proteins. Edges denote reported associations between leisons and proteins derived from the included studies. A color gradient ranging from light orange to dark red was applied to protein nodes to enhance visual differentiation. Yellow denotes the lesion, orange represents the lesion specific proteins, red indicates the protein shared between 2 lesions and dark red represent the protein shared across three lesions A bipartite network illustrating the association between PVL, LK, OLP, and OSMF—and their corresponding proteins constructed using Cytoscape. Circular nodes represent diseases and square nodes represent proteins. Edges denote reported associations between leisons and proteins derived from the included studies. A color gradient ranging from light orange to dark red was applied to protein nodes to enhance visual differentiation. Yellow denotes the lesion, orange represents the lesion specific proteins, red indicates the protein shared between 2 lesions and dark red represent the protein shared across three lesions On the other hand, the disease-specific protein marker indicated unique molecular pathogenesis that revealed the distinct biological processes of each lesion. However, the lesion clusters were characterized by a combination of shared molecular mechanisms and lesion-specific alterations, which supports a continuum model of disease progression rather than completely independent pathological entities. Protein–protein interaction network and hub protein analysis For each of the four lesion groups, PPI networks were constructed, which revealed distinct structural and functional characteristics. In the LK network, 7 proteins (nodes) interacted with 2 edges, with a clustering coefficient of 0.571 and an average node degree of 0.571, which suggested a moderately connected, functionally coherent group. The network showed strong enrichment (p = 0.0002), indicating a meaningful biological network. At the core of this network AMY1A, AMY1B, APOA1, and LYZ form a tightly linked cluster related to salivary and immune functions. Notably, CytoHubba analysis consistently identified AMY1A and AMY1B as the top hub proteins in LK across all five ranking methods. The PVL network was considerably larger and more intricate than LK, with 72 nodes, 38 edges, an average node degree of 1.06, and a clustering coefficient of 0.391. The significant enrichment (p = 7.77 × 10⁻1⁶), suggests that these proteins were deeply interconnected. A dense central cluster within the network further pointed to a group of key regulatory proteins that orchestrated multiple biological processes at once. Among all the proteins in the network, CFL1 was the top-ranked hub protein across MCC, Degree, EPC, closeness and betweenness algorithms. Other consistently highly ranked hub proteins included CDC42 and RHOA. These proteins demonstrated high connectivity within the network, indicating their central role in PVL-associated molecular interactions. The OSMF network was the most densely connected of all, bringing together 130 nodes and 207 edges, which is a striking contrast to the other lesion groups. With an average node degree of 3.18 and a clustering coefficient of 0.502, the network reflected a highly active and tight network. The observed interactions nearly tripling the predicted 280 connections, and the enrichment was highly significant (p < 1.0e − 16). CytoHubba analysis of the OSMF network revealed distinct hub protein patterns across different algorithms. The MCC method identified SERPINA1 as the top-ranking hub. Whereas, Degree, EPC, closeness and betweenness analyses identified HSP90AA1 and ENO1 as the most connected proteins, which were stress-related proteins. The OLP network comprised 43 nodes and 33 edges, with an average node degree of 1.53 and a clustering coefficient of 0.404. The highly significant enrichment (p < 1.0 × 10⁻1⁶) reflected strong functional connectivity among the constituent proteins, suggesting that they participate in closely related and biologically meaningful processes. A moderately dense central cluster was apparent within the network, indicative of coordinated regulatory activity and the likely presence of influential hub proteins. CytoHubba analysis identified HP, B2M and ORM1 as the top-ranked hub proteins across MCC, Degree, EPC, betweenness and closeness algorithms. The high connectivity of these proteins within the network underscores the central involvement of immune and inflammatory pathways in the pathogenesis of OLP (Figs. 4 and 5).Fig. 4Protein–Protein Interaction (PPI) Network Maps of OPMDs created using STRING (https://string-db.org/): a Leukoplakia (LK), b Proliferative verrucous leukoplakia (PVL), c Oral submucous fibrosis (OSMF), and d Oral lichen planus (OLP)Fig. 5Top-Ranked Hub Proteins Identified by MCC, Degree, EPC, Closeness and Betweenness Algorithms in OPMDs (a-d) using CytoHubba plugin from Cytoscape (https://cytoscape.org/), a Leukoplakia (LK), b Proliferative verrucous leukoplakia (PVL), c Oral submucous fibrosis (OSMF), and d Oral lichen planus (OLP), Color gradient (red to yellow) represents ranking, where red indicates the highest-ranked hub proteins and yellow indicates lower-ranked proteins; Reactome Pathway Enrichment Analysis of Hub Proteins in OPMDs (e–h), e LK, f PVL, g OSMF, and h OLP. * Maximal Clique Centrality (MCC), EPC (Edge Clustering Coefficient) Protein–Protein Interaction (PPI) Network Maps of OPMDs created using STRING (https://string-db.org/): a Leukoplakia (LK), b Proliferative verrucous leukoplakia (PVL), c Oral submucous fibrosis (OSMF), and d Oral lichen planus (OLP) Top-Ranked Hub Proteins Identified by MCC, Degree, EPC, Closeness and Betweenness Algorithms in OPMDs (a-d) using CytoHubba plugin from Cytoscape (https://cytoscape.org/), a Leukoplakia (LK), b Proliferative verrucous leukoplakia (PVL), c Oral submucous fibrosis (OSMF), and d Oral lichen planus (OLP), Color gradient (red to yellow) represents ranking, where red indicates the highest-ranked hub proteins and yellow indicates lower-ranked proteins; Reactome Pathway Enrichment Analysis of Hub Proteins in OPMDs (e–h), e LK, f PVL, g OSMF, and h OLP. * Maximal Clique Centrality (MCC), EPC (Edge Clustering Coefficient) Pathway analysis LK and OSMF showed strong enrichment in keratinization, cornified envelope formation, and developmental biology, indicating roles in epithelial differentiation and tissue remodeling. PVL was strongly associated with Rho GTPase signaling, axon guidance, and neural signaling processes, suggesting enhanced cytoskeletal regulation and cell migration. OLP was characterized by up-regulation of immune and platelet activation pathways, reflecting prominent inflammation and immune modulation. Discussion OPMDs has higher likelihood of progressing to OSCC. Regardless of differences in pathogenesis among OPMDs, they all share molecular pathways such as inflammation, oxidative stress, hypoxia, and extracellular matrix (ECM) remodelling during malignant transformation. Nevertheless, the underlying pathogenesis of OPMDs remains complex and not fully elucidated, warranting further investigation (Fig. 6) [28].Fig. 6Represents a conceptual disease progression model illustrating the molecular and pathogenic continuum from oral potentially malignant disorders to oral squamous cell carcinoma Represents a conceptual disease progression model illustrating the molecular and pathogenic continuum from oral potentially malignant disorders to oral squamous cell carcinoma According to epidemiological evidence, OPMDs may account for up to 80% of oral malignancies in Southeast Asia [29], with a malignant transition rate of 2–44% [30–32]. As proteomics reflects post-translational modifications, pathway dysregulation, and real-time cellular activity, it has emerged as a promising tool for precision medicine and disease research [33]. Biofluids-based proteomics analysis can enable the differentiation of normal healthy individuals from OPMD patients and those with oral cancer [34]. This systematic review identified a total of 318 proteins associated with OPMDs, including both lesion-specific and shared biomarkers. However, proteomic research has certain methodological limitations. Low reproducibility and sensitivity limit the use of 2DE approaches, which were extensively employed in previous investigations. Precise protein identification is also difficult with one-dimensional SDS-PAGE due to its low resolution. Integrating it with cutting-edge methods like nano-LC–MS/MS improves protein identification sensitivity and accuracy [25]. This variability in analytical techniques may contribute to inconsistencies in protein identification across studies. The use of pooled samples for proteome analysis is another drawback. Although pooling lowers variability, it may mask minute inter-individual variances, making it more difficult to identify variations that were physiologically significant. Additionally, it was noted that using tissue samples for OSMF proteomic analysis had certain drawbacks, such as the tissues' brittleness and hardness from excessive collagen deposition, making protein extraction difficult [18]. The majority of research emphasized the necessity of independent validation cohorts and greater sample size [13, 15, 16, 21, 23]. Research on PVL was mostly from Brasilia; studies on OLP and LK were more geographically diverse, with contributions from Brazil, Mexico, Thailand, and the United States. The OSMF studies showed a clear geographical trend, with the majority being carried out in Asian countries, especially in China and India, which probably reflects the high prevalence of areca nut use in these regions. This geographical variation may also influence proteomic profiles due to differences in environmental and lifestyle risk factors. To the best of our knowledge, this is one of the first studies to comprehensively integrate the four major OPMD lesions (OSMF, OLP, LK and PVL) and their potential diagnostic sample types (serum, saliva and tissue). This enables lesion- and sample-specific mapping of proteomic evidence and identification of research gaps. A key strength of this study is the application of network analysis. Several identified hub proteins across OPMDs, some of which overlapped with previously validated biomarkers (HSP in OSMF [18, 32], and C3 in OLP [25]). These findings suggest that network-based approaches can effectively prioritize biologically relevant biomarkers for future validation. The included studies exhibited heterogeneity in terms of sample types, proteomic techniques, analytical workflows, and reporting standards. This variability may influence the composition of pooled protein datasets used for protein–protein interaction network construction and hub protein identification. Although lesion-specific analyses and vote-counting approaches were employed to mitigate these effects, residual heterogeneity may still impact the stability and reproducibility of the identified networks. Limitations One of the key limitations of this study is that, despite a wide search period from 2000 to 2024, the included studies were comparatively limited in number (n = 15). This illustrates that only a few OPMD-related proteome studies had employed specialised proteomic techniques. Protein-specific analysis is still a challenge for low- and middle-income nations, especially those in Southeast Asia, where most of the studies included in this systematic review were conducted.Meta-analysis was not feasible due to the absence of common biomarkers among the lesions, different proteomics techniques employed, and the lack of essential statistical parameters such as mean and standard deviation across the included studies.The clinical feasibility of the biomarkers that have been identified from this systematic review is an important aspect to be concentrated on for their translation into routine practice. The majority of the markers were derived from tissue (179), followed by saliva (135) and serum (4). Since tissue-based markers can be directly obtained from the lesion site, they can be comparatively accurate and allow precise evaluation of molecular alteration, but being invasive is their drawback. Salivary markers can be obtained via a non-invasive approach and may contain the exfoliated cells, proteins and nucleic acids from the lesion site. While serum-based markers were the least reported among the articles included, this might be due to patient compliance. On observing the feasibility of the techniques employed, mass spectrometry-based proteomic techniques were the most common analytical technique. Despite their high sensitivity and ability to detect low-abundance biomarkers, these techniques were not widely used in clinical applications due to their high cost and specialized infrastructure.As this systematic review focused on the early detection of OPMDs, the study required healthy controls as a comparison group. Hence, several potentially pertinent papers were excluded from the systematic review because the majority of the articles employed OSCC as the control group, and some studies did not provide lesion-specific data; instead, the article grouped OPMDs as a single entity. One of the key limitations of this study is that, despite a wide search period from 2000 to 2024, the included studies were comparatively limited in number (n = 15). This illustrates that only a few OPMD-related proteome studies had employed specialised proteomic techniques. Protein-specific analysis is still a challenge for low- and middle-income nations, especially those in Southeast Asia, where most of the studies included in this systematic review were conducted. One of the key limitations of this study is that, despite a wide search period from 2000 to 2024, the included studies were comparatively limited in number (n = 15). This illustrates that only a few OPMD-related proteome studies had employed specialised proteomic techniques. Protein-specific analysis is still a challenge for low- and middle-income nations, especially those in Southeast Asia, where most of the studies included in this systematic review were conducted. Meta-analysis was not feasible due to the absence of common biomarkers among the lesions, different proteomics techniques employed, and the lack of essential statistical parameters such as mean and standard deviation across the included studies. Meta-analysis was not feasible due to the absence of common biomarkers among the lesions, different proteomics techniques employed, and the lack of essential statistical parameters such as mean and standard deviation across the included studies. The clinical feasibility of the biomarkers that have been identified from this systematic review is an important aspect to be concentrated on for their translation into routine practice. The majority of the markers were derived from tissue (179), followed by saliva (135) and serum (4). Since tissue-based markers can be directly obtained from the lesion site, they can be comparatively accurate and allow precise evaluation of molecular alteration, but being invasive is their drawback. Salivary markers can be obtained via a non-invasive approach and may contain the exfoliated cells, proteins and nucleic acids from the lesion site. While serum-based markers were the least reported among the articles included, this might be due to patient compliance. On observing the feasibility of the techniques employed, mass spectrometry-based proteomic techniques were the most common analytical technique. Despite their high sensitivity and ability to detect low-abundance biomarkers, these techniques were not widely used in clinical applications due to their high cost and specialized infrastructure. The clinical feasibility of the biomarkers that have been identified from this systematic review is an important aspect to be concentrated on for their translation into routine practice. The majority of the markers were derived from tissue (179), followed by saliva (135) and serum (4). Since tissue-based markers can be directly obtained from the lesion site, they can be comparatively accurate and allow precise evaluation of molecular alteration, but being invasive is their drawback. Salivary markers can be obtained via a non-invasive approach and may contain the exfoliated cells, proteins and nucleic acids from the lesion site. While serum-based markers were the least reported among the articles included, this might be due to patient compliance. On observing the feasibility of the techniques employed, mass spectrometry-based proteomic techniques were the most common analytical technique. Despite their high sensitivity and ability to detect low-abundance biomarkers, these techniques were not widely used in clinical applications due to their high cost and specialized infrastructure. As this systematic review focused on the early detection of OPMDs, the study required healthy controls as a comparison group. Hence, several potentially pertinent papers were excluded from the systematic review because the majority of the articles employed OSCC as the control group, and some studies did not provide lesion-specific data; instead, the article grouped OPMDs as a single entity. As this systematic review focused on the early detection of OPMDs, the study required healthy controls as a comparison group. Hence, several potentially pertinent papers were excluded from the systematic review because the majority of the articles employed OSCC as the control group, and some studies did not provide lesion-specific data; instead, the article grouped OPMDs as a single entity. Future directions It is essential to conduct such kind of exploratory studies in a varied population because it would help in the development of candidate markers with great reliability and generalizability. Conducting this type of diagnostic research would facilitate the standardization of protocol for sample collection and processing, which would reduce the variability and increase the reproducibility.To obtain a bigger picture of the underlying disease mechanisms, in the future, single-omics studies could be replaced with multi-omics techniques. By doing this, the process of exploring the pathogenesis and acquiring reliable biomarker panels becomes channelized. Further validation of these biomarker panels in a larger sample size would enable the analysis of their clinical significance.In this systematic review, only normal healthy volunteers were considered as the comparison group, for future analysis if OSCC should be considered as an additional comparator. This would enable the identification of biomarkers specifically associated with malignant transformation, thereby improving understanding of malignant transformation of OPMDs and facilitating early risk stratification. It is essential to conduct such kind of exploratory studies in a varied population because it would help in the development of candidate markers with great reliability and generalizability. Conducting this type of diagnostic research would facilitate the standardization of protocol for sample collection and processing, which would reduce the variability and increase the reproducibility. It is essential to conduct such kind of exploratory studies in a varied population because it would help in the development of candidate markers with great reliability and generalizability. Conducting this type of diagnostic research would facilitate the standardization of protocol for sample collection and processing, which would reduce the variability and increase the reproducibility. To obtain a bigger picture of the underlying disease mechanisms, in the future, single-omics studies could be replaced with multi-omics techniques. By doing this, the process of exploring the pathogenesis and acquiring reliable biomarker panels becomes channelized. Further validation of these biomarker panels in a larger sample size would enable the analysis of their clinical significance. To obtain a bigger picture of the underlying disease mechanisms, in the future, single-omics studies could be replaced with multi-omics techniques. By doing this, the process of exploring the pathogenesis and acquiring reliable biomarker panels becomes channelized. Further validation of these biomarker panels in a larger sample size would enable the analysis of their clinical significance. In this systematic review, only normal healthy volunteers were considered as the comparison group, for future analysis if OSCC should be considered as an additional comparator. This would enable the identification of biomarkers specifically associated with malignant transformation, thereby improving understanding of malignant transformation of OPMDs and facilitating early risk stratification. In this systematic review, only normal healthy volunteers were considered as the comparison group, for future analysis if OSCC should be considered as an additional comparator. This would enable the identification of biomarkers specifically associated with malignant transformation, thereby improving understanding of malignant transformation of OPMDs and facilitating early risk stratification. Conclusion To conclude, proteomic research makes it feasible to identify important proteins, enriched pathways, and molecular mechanisms that contribute to the pathophysiology of OPMD. Immunoglobulins in LK, cytoskeletal proteins in PVL, heat shock proteins in OSMF, and HP in OLP, which coincides with the markers that were reported to be validated in the included articles, were identified via network analysis. Although these findings demonstrate the potential of proteomic biomarkers for improving molecular understanding and early detection techniques, the majority of the available data comes from exploratory research with insufficient validation. Therefore, rather than considering these markers as a reliable diagnostic marker, they should be considered exploratory markers. In order to consider these markers as clinical diagnostic markers, future research should concentrate on bigger cohort studies, consistent techniques, and multi-centre validation of protein panels across varied populations. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 1. Supplementary Material 2. Supplementary Material 3.