HPV infection in relation to OSCC histological grading and TNM stage. Evalutation by traditional statistics and fuzzy logic model

We aimed to evaluate if in oral squamous cell carcinoma (OSCC) there is a relationshipbetween histological grading (HG), TNM clinical stage and HPV infection; and to study theperformance of fuzzy logic compared to traditional statistics, in the analysis of HPV status andcorrelates of OSCC. In cross-sectional analysis, the study group comprised 63 patients (meanage 68.89 years (SD ± 11.78), range (32–93); males 28 (44.4%), females 35 (55.6%)) with OSCChistologically diagnosed. HPV-DNA was studied in exfoliated oral epithelial cells by nested PCR(MY09/MY11 and GP5+/GP6+ primers). Data were analysed in parallel by traditional statisticswith multivariate analysis and a fuzzy logic (FL) technique (membership functions as input,the ANFIS methodology, and the Sugeno’s model of first order). HPV infection was detectedin 24/63 (38.1%) of OSCC, as being HPV+ve 14/36 (38.9%) in G1, 7/18 (38.9%) in G2, and 3/9(33.3%) in G3; HPV+ve 8/33 (24.2%) in Stage I, 9/12 (75.0%) in Stage II, 6/11(54.5%) in StageIII, and 1/7 (14.3%) in Stage IV. In both methods of analysis, no significantly increased risk ofHPV infection was found for any HG score; whereas, TNM stage II was significantly associatedto HPV infection (p = 0.004; OR = 9.375 (95% CI = 2.030:43.30); OR0 = 11.148 (95%CI = 1.951:43.30)), and, in particular, to primary tumour size T2 (p = 0.0036; OR = 7.812 (95% CI = 1.914:31.890); OR0 = 9.414 (95% CI = 1.846:48.013)); FL (% of prevision: 79.8; Root MeanSquareError (RMSE): 0.29). No association was found between HPV infection and anydemographical variable. Our findings show an association between HPV infection with TNM(stage II – T2), but not with histological grading of OSCC. Also, FL seems to be an additionaleffective tool in analysing the relationship of HPV infection with correlates of OSCC.c 2005 Elsevier Ltd. All rights reserved.