Sufficient dimension reduction for censored predictors

Estimating the AUC with a Graphical Lasso Method for High-dimensional Biomarkers with LOD

This manuscript estimates the area under the receiver operating characteristic curve (AUC) of combined biomarkers in a high-dimensional setting. We propose a penalization approach to the inference of precision matrices in the presence of the limit of detection. A new version of expectation-maximization algorithm is then proposed for the penalized likelihood, with the use of numerical integration and the graphical lasso method. The estimated precision matrix is then applied to the inference of AUCs. The proposed method outperforms the existing methods in numerical studies. We apply the proposed method to a data set of brain tumor study. The results show a higher accuracy on the estimation of AUC compared with the existing methods.

Inflammatory profiles in Chilean Mapuche and non-Mapuche women with gallstones at risk of developing gallbladder cancer

Chile has high incidence rates of gallbladder cancer globally, particularly among Amerindian women, who also have a high prevalence of gallstones. We examined differences in inflammatory biomarkers between Mapuche and non-Mapuche women from the Chile Biliary Longitudinal Study, a cohort of women with ultrasound-detected gallstones. We randomly selected 200 Mapuche women frequency matched to non-Mapuche women on age and statin use Inflammatory biomarkers were analyzed using a multiplex assay and linear regression to assess associations of a priori markers (CCL20, CXCL10, IL-6, and IL-8) with ethnicity. Novel biomarkers were analyzed using exploratory factor analysis (EFA) and sufficient dimension reduction (SDR) to identify correlated marker groups, followed by linear regression to examine their association with ethnicity. The mean values of IL-8 were higher in Mapuche than non-Mapuche women (P = 0.04), while CCL20, CXCL10, and IL-6 did not differ significantly by ethnicity. EFA revealed two marker groups associated with ethnicity (P = 0.03 and P < 0.001). SDR analysis confirmed correlation between the biomarkers and ethnicity. We found higher IL-8 levels among Mapuche than non-Mapuche women. Novel inflammatory biomarkers were correlated with ethnicity and should be studied further for their role in gallbladder disease. These findings may elucidate underlying ethnic disparities in gallstones and carcinogenesis among Amerindians.