The potential value of oral microbial signatures for prediction of oral squamous cell carcinoma based on machine learning algorithms.

OBJECTIVE: This study aimed to explore the potential predictive value of oral microbial signatures for oral squamous cell carcinoma (OSCC) risk based on machine learning algorithms. METHODS: The oral microbiome signatures were assessed in the unstimulated saliva samples of 80 OSCC patients and 179 healthy individuals using 16S rRNA gene sequencing. Four different machine learning classifiers were used to develop prediction models. RESULTS: Compared with control participants, OSCC patients had a higher microbial dysbiosis index (MDI, p < 0.001). Among four machine learning classifiers, random forest (RF) provided the best predictive performance, followed by the support vector machines, artificial neural networks and naive Bayes. After controlling the potential confounders using propensity score matching, the optimal RF model was further developed incorporating a minimal set of 20 bacteria genera, exhibiting better predictive performance than the MDI (AUC: 0.992 vs. 0.775, p < 0.001). CONCLUSIONS: The novel MDI and RF model developed in this study based on oral microbiome signatures may serve as noninvasive tools for predicting OSCC risk.

A composite oral hygiene score and the risk of oral cancer and its subtypes: a large-scale propensity score-based study.

OBJECTIVE: This study aimed to investigate the potential relationship between oral hygiene and the risk of oral cancer and its subtypes after controlling the effects of several confounding factors. MATERIALS AND METHODS: A large-scale case-control study was conducted from January 2010 to August 2019, recruiting a total of 1,288 oral cancer cases with newly diagnosed and 4,234 healthy controls. Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) were utilized to minimize confounding effects. Conditional logistic regression was used to evaluate the effects of oral hygiene indicators on oral cancer. RESULTS: A composite oral hygiene score was developed based on five indicators selected based on PSM and IPTW analysis (including tooth loss, dentures wearing, the frequency of tooth brushing, regular dental visits, and recurrent dental ulcer). Participants with a higher score, compared with their lower counterparts, showed a 49% increased risk (the odds ratio (OR) was 1.49 (95% confidence interval (CI): 1.26-1.75). A similar association pattern was found following IPTW analyses (OR = 1.32; 95% CI: 1.22-1.42). Of note, the adverse effects of poor oral hygiene were more evident among the sites of gingival and buccal (PSM analysis: 2.03-fold and 2.68-fold increased risk; IPTW analysis: 1.57-fold and 2.07-fold increased risk, respectively). Additionally, a greater positive association was observed between poor oral hygiene and oral squamous cell carcinoma, compared with other pathological types. CONCLUSION: This study establishes a composite oral hygiene score and provides supportive evidence of poor oral hygiene associated with a higher risk of oral cancer, particularly in the gingival and buccal mucosa sites and in the squamous cell carcinoma. CLINICAL RELEVANCE: The data highlights the importance of improving poor oral hygiene habits, which has public health implications for the prevention of oral cancer.

A novel prognostic score based on systemic inflammatory biomarkers for patients with oral squamous cell carcinoma.

OBJECTIVE: To explore the prognostic value of systemic inflammatory biomarkers (albumin/globulin ratio [AGR], neutrophil/lymphocyte ratio [NLR], and platelet/lymphocyte ratio [PLR]) in patients with oral squamous cell carcinoma (OSCC), and further develop a novel prognostic score (AGR-NLR). METHODS: A large-scale prospective study enrolling 792 eligible patients from December 2002 to June 2018 was carried out at the First Affiliated Hospital of Fujian Medical University. Three multivariate Cox regression models were performed to assess the association of overall survival (OS) with systemic inflammatory biomarkers, quantified by Akaike information criterion (AIC). Then, a novel AGR-NLR score was established and incorporated into a prognostic nomogram. RESULTS: In the univariate analysis, the increased AGR was associated with a reduced risk of death. Conversely, the higher NLR and PLR, the worse the OS. In the multivariate Cox regression models, AGR and NLR were stably independent prognostic indicators in all models, with Model 2 showing a lowest AIC (AGR: HR = 0.56, 95%CI: 0.41-0.78; NLR: HR = 1.80, 95%CI: 1.07-3.04). Then, a novel AGR-NLR score was established, which showed a more excellent performance than either AGR or NLR alone (area under curve [AUC]: 0.589, 0.559, and 0.556, respectively). The C-index of the nomogram based on AGR-NLR was superior to that of traditional TNM staging system (C-index: 0.658 versus. 0.596, p < .001). Similar results were also showed by decision curve analysis, indicating the nomogram had more positive net benefit compared to TNM staging system. CONCLUSION: The novel AGR-NLR score is strongly associated with outcome in patients with OSCC and could be serve as a useful tool to accurately predict the OS of OSCC patients.

Association Between Rare Earth Element Cerium and the Risk of Oral Cancer: A Case-Control Study in Southeast China.

Cerium (Ce), the most abundant of rare earth elements in the earth's crust, has received much health concerns due to its wide application in industry, agriculture, and medicine. The current study aims to evaluate whether there is an association between Ce exposures and the risk of developing oral cancer. Serum Ce level of 324 oral cancer patients and 650 matched healthy controls were measured by inductively coupled plasma mass spectrometry. Association between Ce level and the risk of oral cancer was estimated with an unconditional logistic regression model. Serum Ce concentrations in the oral cancer patients and controls were 0.57 (0.21-3.02) µg/L and 2.27 (0.72-4.26) µg/L, respectively. High level of Ce was associated with a decreased risk of oral cancer (OR: 0.60, 95% CI: 0.43-0.84). Stronger inverse associations between high level of Ce and oral cancer risk were observed among those with smoking (OR: 0.46, 95% CI: 0.27-0.79), drinking (OR: 0.50, 95% CI: 0.26-0.96), limited intake of leafy vegetables (OR: 0.40, 95% CI: 0.22-0.71) and fish (OR: 0.52, 95% CI: 0.33-0.83). There were significant multiplicative interactions between Ce level and alcohol drinking or intake of leafy vegetables and fish (all Pinteraction <0.05). This preliminary case-control study suggests an inverse association between high serum Ce level and the risk of oral cancer. Further prospective studies with a larger sample size are needed to confirm the findings.

A dynamic prognostic nomogram to predict the benefit from surgical treatment modality for patients with laryngeal squamous cell carcinoma.

BACKGROUND: Although nonsurgical treatment strategy is increasingly adopted in patients with locoregionally advanced laryngeal squamous cell carcinoma (LSCC), survival disparities were reported between surgical treatment modality and organ preservation protocols, highlighting the great importance for accurate patients' selection. METHOD: This secondary analysis used data from the Surveillance, Epidemiology, and End Results database between 2010 and 2015 with follow-up data up to 2018. We developed and validated a dynamic prognostic nomogram for overall survival (OS) in 4237 patients with LSCC and subgroup of 2087 patients with locoregionally advanced laryngeal squamous cell carcinoma (LALSCC). Based on the total risk score derived from the dynamic nomogram, two well-matched risk groups (i.e., low- and high-risk) were created via X-tile software and 1-to-1 propensity score matching (PSM); surgical treatment modality was compared with nonsurgical one in each risk group. RESULTS: A more accurate and convenient dynamic prognostic nomogram based on age, marital status, T category, N category, M category, tumor size, and tumor differentiation was developed and validated, of which the predictive performance was superior to that of TNM staging system. For high-risk LALSCC selected by the dynamic nomogram, after 1-to-1 PSM, significantly improved OS was observed in patients with receiving surgical treatment compared to those receipt of nonsurgical management (restricted mean survival time at 36-month: 26.6 vs 22.7, p < 0.001; restricted mean survival time at 60-month: 36.7 vs 31.0, p = 0.003). CONCLUSION: We establish and validate a more accurate and convenient dynamic prognostic nomogram for patients with LSCC, which may predict the benefit from surgical treatment modality for patients with high-risk LALSCC.

Prognostic value of preoperative lymphocyte-to-monocyte ratio in oral cancer patients and establishment of a dynamic nomogram.

OBJECTIVE: To assess the association of preoperative lymphocyte-to-monocyte ratio (LMR) and overall survival (OS) in patients with oral cancer and develop a dynamic nomogram for individualized survival prediction. METHOD: The prognostic value of LMR was evaluated in a large-scale cohort with 651 postoperative patients with oral cancer between January 2010 and December 2017. Propensity score-matched (PSM) analysis and inverse probability of treatment weighting (IPTW) analysis were performed to further verify the prognostic value of LMR. A dynamic nomogram was then developed based on the LMR and clinicopathological features, and its predictive performance and clinical utility were evaluated. RESULTS: A high LMR was significantly associated with better OS of patients with oral cancer (HR = 0.65; 95% CI = 0.44-0.98). The similar association was also observed in the PSM and IPTW analyses. Moreover, compared with TNM staging system, the dynamic nomogram based on the LMR exhibited more excellent predictive performance (0.72 versus 0.64, p < .001), with calibration curves (1,000 bootstrap resamples) suggesting good match between the actual and predicted probabilities. Decision curve analyses (DCAs) showed a more significant positive net benefit in the practical ranges of threshold probabilities using the dynamic nomogram. CONCLUSION: Preoperative LMR may serve as an easily accessible and non-invasive prognostic biomarker for predicting the prognosis of patients with oral cancer. A dynamic nomogram based on the LMR may show more convenience in survival prediction for patients with oral cancer. Further future studies are warranted to confirm our findings.

Prognostic value of preoperative systemic inflammation response index in patients with oral squamous cell carcinoma: Propensity score-based analysis.

BACKGROUND: The aim of the study was to elucidate the relationship between systemic inflammation response index (SIRI) and the prognosis of postoperative oral squamous cell carcinoma (OSCC) patients. METHODS: The prognostic value of SIRI was evaluated in a prospective cohort consisting of 535 OSCC patients with surgical resection. Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) analyses were used to further verify the prognostic value of SIRI. RESULTS: Patients with a higher SIRI had a significantly increased risk of mortality compared with those with a low SIRI (HR [hazard ratio]: 1.60, 95% CI [confidence interval]: 1.04-2.47). The similar association pattern was observed following PSM (HR: 1.97, 95% CI: 1.14-3.40) and IPTW (HR: 1.70, 95% CI: 1.29-2.24) analyses. Of note, receiving postoperative chemotherapy resulted in a 72% of decreased risk of death among patients with a higher SIRI (HR: 0.28, 95% CI: 0.08-0.95). Additionally, a novel prognostic nomogram, based on TNM stage, tumor differentiation, and SIRI, demonstrated superior accuracy for the prediction of overall survival than that of the seventh edition of the AJCC staging system. CONCLUSION: Preoperative SIRI may be a valuable tool for prediction of survival of OSCC patients.

Selenoprotein genetic variants may modify the association between serum selenium and oral cancer risk.

OBJECTIVE: The relationship between selenium (Se) and oral cancer is still controversial, and the selenoprotein genes play crucial roles in selenium metabolism. We aim to investigate the potential effect of selenoprotein genes (including GPx and TXNRD) in the association of serum Se with oral cancer risk. METHODS: A case-control study including 235 oral cancer cases and 406 controls from September 2011 to December 2018 was conducted in Fujian, China. The peripheral blood samples were obtained from each participant. Genotyping was performed by MassARRAY system, and serum Se levels were measured by inductively coupled plasma mass spectrometry (ICP-MS). RESULTS: Compared with the lowest tertile of Se concentration, those with Se levels in the third tertile were associated with the lower risk of oral cancer (OR = 0.228; 95% CI: 0.135, 0.384). After additional adjustment for genetic risk score (GRS, derived from selenoprotein genetic variants), the model demonstrated the superior goodness of fit. When stratified by GRS, the negative correlation of serum Se was more pronounced among those with low risk (i.e., lower GRS). Moreover, there is a multiplicative interaction between serum Se and GRS for the risk of oral cancer (p = .001). CONCLUSIONS: The present study suggests that serum Se levels may be significantly associated with oral cancer risk, yet the association may be modified by the effects of selenoprotein genetic variants.