Potential noninvasive biomarkers for the malignant transformation of oral leukoplakia: A systematic review and meta‐analysis

Abstract Background The rising cancer incidence in patients with oral leukoplakia (OL) highlights the importance of identifying potential biomarkers for high‐risk individuals and lesions because these biomarkers are useful in developing personalized management strategies for OL patients. This study systematically searched and analyzed the literature on potential saliva and serum biomarkers for OL malignant transformation. Methods PubMed and Scopus were searched for studies published up to April 2022. The primary outcome of this study was the difference in biomarker concentrations in saliva or serum samples from healthy control (HC), OL and oral cancer (OC) populations. Cohen's d with 95% credible interval was calculated and pooled using the inverse variance heterogeneity method. Results A total of seven saliva biomarkers were analyzed in this paper, including interleukin‐1alpha, interleukin‐6 (IL‐6), interleukin‐6‐8, tumor necrosis factor alpha (TNF‐α), copper, zinc, and lactate dehydrogenase. IL‐6 and TNF‐α exhibited statistically significant deviations in comparisons between HC versus OL and OL versus OC. A total of 13 serum biomarkers were analyzed, including IL‐6, TNF‐α, C‐reactive protein, total cholesterol, triglycerides, high‐density lipoproteins, low‐density lipoproteins, albumin, protein, β2‐microglobulin, fucose, lipid‐bound sialic acid (LSA), and total sialic acid (TSA). LSA and TSA exhibited statistically significant deviations in comparisons between HC versus OL and OL versus OC. Conclusion IL‐6 and TNF‐α in saliva have strong predictive values for OL deterioration, and LSA and TSA concentration levels in serum also have the potential to serve as biomarkers for OL deterioration.


| INTRODUCTION
A white patch of oral mucosa is a condition defined as oral leukoplakia (OL) by the World Health Organization that cannot be erased or classified otherwise, either clinically or histopathologically. 1 OL does not refer to the presence or absence of epithelial dysplasia at any stage; it is merely a clinical definition and is not related to a specific histology. 2,3 The prevalence of leukoplakia is estimated at 4.1% worldwide. 4 OL can be clinically classified into homogeneous and nonheterogeneous lesions. Homogeneous leukoplakias are flat and uniform in appearance with small superficial cracks, and they are mostly benign and more common. Nonhomogeneous leukoplakias have a variety of features; they can be exophytic, nodular, white or red in color, flat and speckled, or papillary/verrucous. 2,5 Patients with OL are prone to developing oral cancer (OC). 6 OC is closely associated with smoking, alcohol consumption, fungal, bacterial, and viral infections, and hormonal disorders. OL is one of the most common oral potentially malignant disorders (OPTMs, defined as any oral mucosal abnormality that is associated with a statistically increased risk of developing OC. 7 ) and is responsible for approximately 11% of squamous carcinomas. 4 Studies have shown a wide range of OL malignant transformation (MT), from 0.13% to 34.0%. 8 The risk of MT is currently assessed by biopsy analysis of histopathological markers. The prognosis of OL may be affected by some clinical and pathological features, especially signs of epithelial dysplasia, such as asymmetric epithelial stratification, myoepithelial basocellular hyperplasia, dyskeratosis, hyperchromatic nuclei, and pleomorphism. 9 However, studies have shown that grading epithelial dysplasia is difficult in the clinical setting, and treatment planning of OL patients is usually made on the basis of subjective and poorly reproducible judgment. Patients diagnosed with OPMDs can develop cancer anywhere in the mouth during their lifetime. The cancer does not necessarily occur at the site of the visibly altered mucosa. A key challenge for clinicians managing patients with OPMD is identifying the small proportion of patients who are most likely to develop future malignancy. 7 Therefore, there is growing interest in minimally invasive biomarkers for diagnostic procedures. Oral carcinogenesis usually stems from a genetic mutation due to long-term exposure to carcinogens that change tissue homeostasis and increase the possibility of MT. These changes occur in multiple pathways, such as cell proliferation signaling, evasion of immune destruction, invasion and angiogenesis, and reprogramming of cellular energetics. 10 The proteins and genes involved in these pathways are frequently deregulated during oral carcinogenesis. [11][12][13] Saliva sample analysis can be noninvasive, rapidly collected, and easily processed. These advantages make saliva analysis more attractive than biopsy-based histopathological analysis or the use of another matrix and is a promising option for the diagnosis and prognosis of OC.
This study systematically searched and analyzed the literature on potential salivary and serum biomarkers for OL MT. We hope that our study will provide strong clues for finding biomarkers, especially noninvasive biomarkers, for OL MT.

| MATERIALS AND METHODS
The study was designed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) requirements. 14 This systematic review and meta-analysis were registered in the INPLASY database under the registration number INPLASY202250166 (https://inpla sy.com/inpla sy-2022-5-0166/).

| Literature search strategy
PubMed and Scopus were searched for studies published up to April 2022. The literature search strategy was ("oral" AND "leukoplakia") AND ("saliva" OR "spit" OR "serum"). English was set as a language restriction. We also checked the reference list of all identified articles for additional relevant studies, including hand-searching reviews and previous meta-analyses.

| Eligibility criteria and selection of studies
The inclusion criteria for selecting the articles were as follows: The case-control studies reported research on biomarkers in human saliva or serum samples. The samples were obtained from individuals with OL and proliferative verrucous leukoplakia. The articles reported concentrations of biomarkers in OL patients as well as samples from OC patients and healthy controls (HCs) and their association with possible malignant degeneration. Two authors independently examined the titles and abstracts of citations. Full texts of potentially eligible studies were obtained, and disagreements were resolved by discussion.

| Data extraction
Two reviewers extracted the data independently. The first author, publication year, study design, number of patients in each group, assay method, country, biomarker, and concentration of biomarkers in each arm were recorded. We tabulated the concentrations of the biomarkers of each study in different groups and sorted out the data of the planned comparison group for the synthesis. If the concentration units of the biomarker were inconsistent, they were converted to the same unit.

| Quality assessment
The Newcastle-Ottawa Scale (NOS) was used to evaluate the quality of the included studies. It evaluates casecontrol studies through three categories and eight items, including study population selection, comparability, exposure evaluation and outcome evaluation. The NOS adopts the semiquantitative principle of the star system in the evaluation of study quality, with a full score of nine stars.

| Statistical analysis
This study analyzed the standard mean difference of the saliva or serum concentration of the biomarkers in HC versus OL and OL versus OC populations. The synthesis results of all biomarkers extracted from the included studies were tabulated and the results of each study and synthesis were visualized using forest plots. Only biomarkers with a significant difference in both comparisons could be considered potential biomarkers for predicting OL MT. Cohen's d with a 95% credible interval (CI) was calculated and pooled using the inverse heterogeneity (IVhet) of variances method. IVhet is an improved alternative to the random effects (RE) model for meta-analysis of heterogeneous studies. It is shown that the known issues of underestimation of the statistical error and spuriously overconfident estimates with the RE model can be resolved by the use of an estimator under the fixed effect model assumption with a quasi-likelihood based variance structure. Extensive simulations confirm that this estimator retains a correct coverage probability and a lower observed variance than the RE model estimator, regardless of heterogeneity. 15 The Chi 2 test and Higgins I 2 statistics were used to assess heterogeneity among the included studies. 16 In addition, sensitivity analyses were performed by leave-one-out analysis. Publication bias was not assessed because fewer than 10 studies were included in each comparison. The meta-analysis was conducted in MetaXL 5.3 (EpiGear International).

| Literature search and characteristics of the included studies
A total of 538 studies were obtained through preliminary retrieval, of which 113 were subjected to full-text evaluation. Finally, 32 articles involving 3223 subjects (including 1050 OL patients, 957 OC patients, and 1216 HCs) were included in this meta-analysis ( Figure 1). The characteristics of the included studies are shown in Table 1. All included studies were case-control studies, of which 10 studies investigated salivary biomarkers, 17 focused on serum biomarkers, and six analyzed both. The studies covered six groups of biomarkers, including cytokines , glycoconjugates (fucose, lipid-bound sialic acid (LSA) and total sialic acid (TSA)), trace elements ((copper and zinc), and cytoplasmic enzymes lactate dehydrogenase (LDH)). The NOS scale was used to evaluate the studies' quality, and the scores are shown in Table 1. Due to the small sample sizes, the studies were considered to have average quality, and most studies lacked a detailed description of whether important additional factors were controlled for.

| Salivary Biomarkers for the MT of OL
A total of seven salivary biomarkers were analyzed in this paper, including IL-1α, IL-6, IL-8, TNF-α, copper, zinc, and LDH. IL-6 and TNF-α exhibited statistically significant deviations in comparisons between HC versus OL and OL versus OC.  Figures S15-S18). These results showed that the concentration levels of LSA and TSA in the serum of HC, OL, and OC indicated an increasing trend, while the other 11 serum biomarkers (IL-6, TNF-α, CRP, TC, TG, HDL, LDL, albumin, protein, β2-M, and fucose) displayed no differences ( Figures S19-S39).

| Sensitivity analyses
Only IL-6 remained stable as we tested the sensitivity of the potential salivary biomarkers IL-6 and TNF-α, as well as the serum biomarkers LSA and TSA. The statistical significance of the other three potential biomarkers varied with the removal of factors from the analysis one by one. See Table 3 for details.

| DISCUSSION
A total of 32 studies involving 3223 participants were included in this study, and six kinds of potential biomarkers in saliva and serum were analyzed, including cytokines, lipid profiles, proteins, glycoconjugates, cytoplasmic enzymes, and trace elements. The preponderance of evidence indicates that certain molecular alterations can be considered biomarkers for the risk of MT. This study aims to find statistically consistent evidence that these molecules may serve as potential biomarkers for OL deterioration.
The present study revealed that in OL and OC patients, salivary cytokines, such as IL-6 and TNF-α, showed statistically consistent deviations, suggesting that OL patients F I G U R E 1 Flowchart of study selection. had higher IL-6 and TNF-α concentration levels in saliva than healthy people but had lower IL-6 and TNF-α concentrations than OC patients. This finding suggests that altered cytokine responsiveness is not only closely associated with OC development but also correlated with premalignant lesions such as OL; this result was consistent with that in previous studies. 18,49 In normal cells, growth inhibition is accompanied by stimulation with proinflammatory cytokines, and in OC cells, upregulation of positive cell cycle regulators such as nuclear factor kappa B (NF-κB) occurs with the stimulation of proinflammatory cytokines. Various studies have shown that abnormal activation of NF-κB is associated with the development and progression of human cancers, including head and neck squamous cell carcinoma. This leads to the upregulation of antiapoptotic factors, proangiogenic factors and proinflammatory cytokines responsible for tumor growth. 50 When a heterodimeric receptor, which is composed of the common cytokine receptor signal-transducing subunit gp130 51 and ligand-binding IL-6α chain, binds to IL-6 as a multifunctional cytokine, the JAK family of tyrosine kinases is activated, and multiple pathways are stimulated, involving STATs, PI3Ks, MAPKs, and other signaling proteins. 52 Various subcomponents of these pathways are involved in the oncogenic cascade, affecting various cancers, including oral squamous cell carcinoma (OSCC) and premalignant oral lesions. [53][54][55] TNF-α is also a pleiotropic cytokine. The TNF-TNF receptor system is known as an important player in the process of MT that undergoes programmed cell death, angiogenesis, proliferation, and inflammation. 56 TNF-α may directly induce DNA damage in cells, which is related to cell MT, by inducing reactive oxygen species. 57 Moreover, TNF family members are also involved in immunosuppression. 58 According to G et al, 26 the specificity and sensitivity of salivary TNF-α were 96.7% and 100%, respectively, in OSCC and 93.3% and 90% in oral epithelial dysplasia. The significantly reliable and valid values suggest that salivary TNF-α can be used as a biomarker for OSCC and oral epithelial dysplasia. Saliva can reflect the healthy state and systemic disease state of individuals. Thus, salivary biomarkers are ideal for making pathological diagnoses. Because saliva directly comes from the oral cavity, saliva testing is an ideal choice for assessing potential malignancy and malignant oral lesions and identifying individuals at risk. 59,60 Therefore, we believe that IL-6 and TNF-α are potential biomarkers for OL deterioration and can be used as useful screening tools.
The levels of IL-6 and TNF-α in saliva vary with the progression of OL; therefore, they can be used for the diagnosis and prognosis of OL MT, providing clinicians with a valuable noninvasive procedure as a clinical finding supplement.
According to our study, the molecular concentration levels of LSA and TSA in serum glycoconjugates also show statistically consistent deviations in OL and OC. The surface of the cell membrane is mainly composed of glycolipids and glycoproteins. Sialic acid, a family of acetylated derivatives of neuraminic acid, is widely found in mammals and usually occurs as a terminal component at the nonreducing end of carbohydrate chains of glycolipids and glycoproteins. 61 Sialic acid, mainly LSA, is an important factor in determining cell surface properties and is associated with immunogenicity, adhesion, and cell invasiveness. 62 Any alteration in the intracellular microenvironment may result in a change in the composition of the surface membrane. The altered carbohydrate composition of glycolipids and glycoproteins on the surface of malignant cells is responsible for antigenicity, cell adhesion, aberrant intercellular recognition, and invasiveness. These glycolipids and glycoproteins can be released into the serum by shedding, secretion, and increased turnover. 63 Therefore, LSA and TSA concentration levels in serum have potential diagnostic and prognostic values for OL deterioration.
Most biomarkers showed no statistically consistent deviations in our study. LDH concentration levels in saliva showed statistically significant deviations in OL versus OC but not in HC versus OL. The enzyme LDH is found in cells of almost all body tissues. Increased LDH levels are due to an increased mitotic index and lactic acid production by tumor cells due to the breakdown of glycoproteins. Our results, however, cannot deny the potential role of salivary LDH.
Several studies have found significant hypocholesterolemia and varied serum lipoprotein (TC, TG, HDL, LDL) levels in OC and oral precancerous lesions. [64][65][66] The concentration levels of copper and zinc tend to decrease in the blood of patients with head and neck cancer. 67,68 However, no difference was detected in our meta-analysis in either the OL versus HC comparison or the OL versus OC comparison.
The oxidation of proteins plays an important role in the pathogenesis of OC. 35,69 Hyperproteinemia is commonly observed in oral malignancies and is expressed as cachexia. 70 However, no deviation in serum albumin, protein, or β2-M was detected in our pooled values for OL versus OC.
All studies included in this meta-analysis were casecontrol studies of average quality and small sample sizes from hospitals rather than communities. The assay method and other methods and means used varied from study to study, resulting in considerable heterogeneity in the biomarker concentration levels. Our findings need to be interpreted with caution, although we have reduced the impact of the assay method by applying Cohen's d rather than the weighted mean difference to evaluate the difference in concentrations. Furthermore, due to the large number of biomarkers and the scattered focus of research, the studies included for each biomarker are insufficient. The sensitivity analysis confirmed the potential of salivary IL-6 as a biomarker, but also suggested the unstable results of salivary TNF-α and serum TSA. Since only two studies were involved in the synthesis of serum LSA, the sensitivity analysis was not performed. Its results should also be interpreted with caution. Due to the large number of salivary biomarkers, we mainly considered three aspects when selecting the target biomarkers in this study. First, we reviewed previous reviews and metaanalyses for the hot OL biomarkers to make a preliminary selection. [71][72][73][74][75] Second, since the same disease was studied with different instruments and experimental designs, it is not surprising that different biomarkers were proposed for the same pathology, ELISA was chosen for our study. 7 Third, understudied biomarkers were abandoned by full-text reading when the search strategy was chosen for inclusion in the study. Due to the subjective factors in the above process, OL salivary and serum biomarkers involved in this study may still be lacking. We found only one biomarker with stable statistical difference, providing clues for future large-sample, multi-instrument verification experiments. It is expected that follow-up research will focus on a certain number of potential biomarkers, such as salivary IL-6 synthesized in this study, and find more accurate evidence. Abbreviations: HC, health control; HCI, high confidence interval; IL-6, interleukin-6; LCI, low confidence interval; OC, oral cancer; OL, Oral leukoplakia; TNF-α, tumor necrosis factor alpha; TSA, total sialic acid.

| CONCLUSION
Our study shows that IL-6 and TNF-α in saliva have strong predictive values for OL deterioration, and LSA and TSA concentration levels in serum also have the potential to serve as biomarkers for OL deterioration. This study is obviously insufficient in quantity and quality despite the statistically consistent evidence. Some larger and multicenter studies are needed to ensure the reliability of the consistent evidence before these biomarkers are applied in clinical practice.