Koos Boeve

156 Chapter 8 To evaluate the possible clinical relevance for the detection of tumor cells in saliva independent on methylation status in the original OSCC tissue, we determined the optimal cut-off to discriminate between OSCC and non-cancer control DNA in saliva cells for each marker. ROC analysis among all 20 patients (10 OSCC patients versus 10 controls), revealed a high area under the curve (AUC) with a 100% sensitivity and 100% NPV of EDNRB (AUC 0.82) and KCNA5 (AUC 0.99) for detecting patients with OSCC using saliva cells (Table 3A). The other five markers showed a lower sensitivity and NPV when using the most optimal cut-off. The analysis on the age-matched patients (10 OSCC versus five aged-matched controls) resulting in other optimal cut-offs also showed a 100% sensitivity and 100% NPV for EDNRB (AUC 0.68) (Table 3B). For KCNA5 (AUC 0.98) both the sensitivity (90%) and NPV (83%) decreased slightly, but interestingly with the highest specificity (100%) and positive predictive value (PPV 100%) (Table 3B). As none of the markers had a 100% diagnostic potential, we combined one or more methylation markers in age matched samples. This analysis revealed that KCNA5 combined with TIMP3 had the highest diagnostic potential (100% for this limited dataset) in detecting saliva cells in patients with OSCC (data not shown). DISCUSSION DNA methylation of OSCC specific tumor markers might be useful as biomarkers for early detection of new primaries or local recurrences in OSCC patients, preferably prior to clinical manifestation. In this study we used the methylome of tissue biopsies of 12 patients with OSCC generated using genome-wide methylation screening by MethCapSeq analysis [23] to identify DNA methylation biomarkers with a high diagnostic potential for the detection of OSCC. Seven new OSCC-specific biomarkers representing six genes were identified by selection of equally methylated markers between all 12 OSCC and not methylated in two pools with leukocytes from four different individuals. Moreover, the acquired highest ranking methylated candidate markers were compared to a vast methylome database of over 80 different samples considered as negative control samples. For the validation of these markers using QMSP, we could design optimal primers/probes assays for three markers ( C11orf85, KCNA5 and SIPA1 ). To evaluate biomarkers with the highest performance, DNA was isolated from saliva cells acquired from 10 OSCC patients and their corresponding tumor tissues. Saliva cells fromfive younger controls and five age-matched controls planned to undergo benign surgery served as healthy (non-cancer) controls. KCNA5 was the best marker (independent of age) as it was significantly hypermethylated in OSCC saliva cells in comparison to control saliva. The possible clinical relevance of KCNA5 is further illustrated by the very high sensitivity (90%), NPV (83%), specificity (100%) and PPV (100%), the highest of all markers tested in this study (Table 3B). Moreover, a panel of KCNA5 and TIMP3 could