154 Chapter 6 Biosystems, Foster City, CA, USA), using Epitect Multiplex PCR Mastermix (Qiagen, Venlo, Netherlands) and 2.5–5.0 µM of each primer and 5.0–10.0 µM of each probe in a total volume of 12.5 µl. As a positive control, double-stranded gBlocks™ Gene Fragments (Integrated DNA Technologies) containing the target regions were taken along. H2O was taken along as a negative control during each run. Samples with a ACTB Cycle threshold (Ct) value exceeding 32 were excluded from methylation analysis to ensure sample quality and sufficient input. Methylation marker abundance was calculated relative to ACTB levels (Ct-ratio), using the following formula: 2 ^ (CtMARKER – CtACTB) * 100. Data analysis For comparison of categorical data between groups, the χ2 test was used. All calculations of methylation levels were performed using square root transformed Ctratios. Differences in DNA methylation levels between cases and controls, smokers and non-smokers, stages, histological subtypes, and tumors with and without nodal involvement were compared using the Mann Whitney U test. P-values <.05 (two-sided) were considered statistically significant. The performance of individual methylation markers was assessed by univariate logistic regression analysis. To determine whether a combination of markers improved discrimination between cases and controls, multivariate logistic regression using backward selection was applied. The predicted probabilities obtained from the logistic regressions, representing the probability for the presence of NSCLC, were visualized using receiver operating characteristic (ROC) curves, including the area under the curve (AUC) value for each sample and individually per sample. Model performance was evaluated by AUC values with confidence intervals, and sensitivity and specificity at the Youden’s Index (J) threshold (30). This threshold was used to define marker cut-offs based on the predicted probabilities that maximizes the sum of sensitivity and specificity. The predictive performance of the individual markers and marker combination were assessed outside the set by leave-one-out cross-validation (LOOCV). Samples were considered positive if any of the individual markers was classified as positive (‘believe-the-positive’) (31). Statistical testing was performed using SPSS (SPSS 22.0, IBM Corp., NY, USA). Logistic regression analyses and LOOCV were executed using R version 4.0.3 (Vienna, Austria. UR). RESULTS Study population A total of 46 patients who underwent pulmonary surgery with curative intent for NSCLC and 50 controls were included. NSCLC patients and controls showed no statistically significant
RkJQdWJsaXNoZXIy MjY0ODMw