Infection with MCPyV, KIV, WUV, and HPV as potential risk factors for lung cancer

Abstract

<bold>Background</bold>: Liquid bead microarray antibody (LBMA) assays are used to assess pathogen-cancer associations, yet analytic methods differ between studies, limiting comparability. <bold>Methods</bold>: To assess methods for analyzing LBMA data, we generated 10,000 Monte Carlo-type simulations of log-normal antibody distributions (exposure) with 200 cases and 200 controls (outcome). We estimated type I error rates, statistical power, and bias associated with three types of analytic techniques: (a) t-tests; (b) logistic regression with a linear predictor; and (c) logistic regression with predictors dichotomized according to four methods of defining cutpoints: 200 or 400 MFI determined a priori; the mean MFI among controls plus two standard deviations; and the optimal value based upon receiver operating characteristic (ROC) curve analysis. We also applied these models, and data visualizations (kernel density plots, ROC curves, predicted probability plots, Q-Q plots), to empirical data evaluating the association between HPV16 L1 antibody response and colorectal polyps to assess the consistency of the exposure-outcome relationship. <bold>Results</bold>: All strategies had acceptable type I error rates (0.030≤P≤0.048), except for the dichotomization according to optimal sensitivity and specificity (type I error rate = 0.27). Among the remaining methods, logistic regression with a linear predictor and t-tests had the highest power (Power=1.00 for both) to detect a mean difference of 1.0 MFI (median fluorescence intensity) on the log scale and were unbiased. Dichotomization methods upwardly biased the risk estimates. <bold>Conclusion</bold>: Logistic regression with linear predictors and unpaired t-tests were superior to logistic regression with dichotomized predictors for assessing disease associations with LBMA data.